the epidemiology of childhood type 1 diabetes in …...the epidemiology of childhood type 1 diabetes...
TRANSCRIPT
The epidemiology of childhood Type 1 diabetes in
Western Australia: trends and determinants of
incidence
Dr Aveni Haynes
BA(Hons), MBBChir
This thesis is presented for the degree of Doctor of Philosophy of the University of
Western Australia
School of Population Health, University of Western Australia
2012
i
Statement of Contribution
All the work presented in this thesis comprises of my own original work except where
otherwise stated. I have been involved as the main researcher in all stages of the studies
whose findings form the basis of this thesis. This has included the formation of study
concepts, determination of appropriate study designs, as well as applying for grants and
obtaining all necessary ethics approvals. In addition, I have been solely responsible for
all aspects of data cleaning, validation and management, and for performing all the
statistical analyses with advice provided by my supervisors. I have also been
responsible for writing all draft manuscripts reporting the study findings. After
obtaining feedback and advice from my supervisors and co-authors, I have been
responsible for making relevant changes to the manuscripts and for submitting the final
versions for publication in peer-reviewed journals.
Due acknowledgement has been made in the text to the contributions made by co-
authors and all other material used. Co-authors have given their permission for the
work to be included in this thesis and details of their contributions are provided in
signed statements, which are presented in the relevant Appendix. The work contained
in this thesis has not been previously submitted for any other degree and has been
completed during my period of candidature for this degree at The University of Western
Australia.
Dr Aveni Haynes 18th June 2012
ii
I verify that the declaration made above by Dr Aveni Haynes is an accurate and true
representation of her contribution to the co-authored publications and work presented in
this thesis.
Co-supervisor: Professor Carol Bower
Signature:
Date: 18/06/2012
Co-supervisor: Associate Professor Elizabeth Davis
Signature:
Date: 19/06/2012
Co-ordinating Supervisor: Professor David Preen
Signature: ___________________________________________________________
Date: 20/06/2012
iii
Abstract
Type 1 diabetes is one of the most common chronic diseases in childhood and in many
countries the incidence has been increasing. The aims of this research were to use the
population-based data in Western Australia to document trends in childhood Type 1
diabetes and to identify risk factors for the disease.
All children diagnosed with Type 1 diabetes in Western Australia under the age of 15
years were identified using the Western Australian Children's Diabetes Database.
Record linkage to the Western Australian Data Linkage System was undertaken to
enable validation of case ascertainment and to obtain perinatal data for cases and all
remaining births in the State. Population data were obtained from the Australian
Bureau of Statistics. Incidence rates and incidence rate trends of childhood Type 1
diabetes in Western Australia were determined from 1985 to 2010 and analysed by
calendar year, gender and age group. The incidence was also examined for seasonal
variation by month of diagnosis and month of birth, and for variation by place of
residence and socioeconomic status at the time of diagnosis and the time of birth.
Associations between the incidence of childhood Type 1 diabetes and potential perinatal
risk factors were also examined.
Between 1985 and 2010, the mean annual age-standardised incidence for childhood
Type 1 diabetes in Western Australia was 18.1 per 100,000 person years. The incidence
increased by an average of 2.2% a year, increasing in both genders and all age groups,
with the lowest rate of increase observed in 0-4 year olds. A new finding was the
observation of a sinusoidal 5-year cyclical variation in the incidence rate trend of
childhood Type 1 diabetes. Significant seasonal variation was found with a higher rate
iv
of cases diagnosed in the autumn and winter months, but no variation found by month
of birth.
An increased incidence of childhood Type 1 diabetes was associated with higher
socioeconomic status and in children living in urban compared to non-urban areas of
Western Australia. Another new finding was that over the past decade, the incidence
increased at a greater rate in non-urban compared to urban areas of Western Australia,
resulting in a narrowing of the gap in incidence between urban and non-urban areas of
the State.
Analysis of perinatal factors showed that a higher incidence of childhood Type 1
diabetes was associated with older maternal age, higher birth weight, lower gestational
age and in first born compared to later born children.
The research findings presented in this thesis provide strong evidence for the role of
environmental risk factors in the aetiology of childhood Type 1 diabetes. Key
contributions to the global understanding of childhood Type 1 diabetes include the
observation of a cyclical variation in the incidence rate trend, a disproportionate rate of
increase in the incidence in non-urban compared to urban areas of the State, and a lower
rate of increase in 0-4 year olds compared to older children.
v
Publications arising from thesis
1. A Haynes, C Bower, MK Bulsara, TW Jones, EA Davis. Continued increase in the
incidence of childhood Type 1 diabetes in a population-based Australian sample
(1985-2002). Diabetologia 2004; 47(5):866-70. (Chapter 5, Appendix D)
2. A Haynes, EA Davis. Why is Type 1 diabetes becoming more common in children?
Diabetes Management Journal 2006; 17: 7. (Chapter 5, Appendix D)
3. A Haynes, C Bower, MK Bulsara, JP Codde, TW Jones, EA Davis. Independent
effects of socioeconomic status and place of residence on the incidence of Type 1
diabetes in Western Australia. Pediatric Diabetes 2006; 7:94-100. (Chapter 6,
Appendix E)
4. A Haynes, C Bower, MK Bulsara, J Finn, TW Jones, EA Davis. Perinatal risk
factors for childhood Type 1 diabetes in Western Australia - a population-based
study (1980-2002). Diabetic Medicine 2007; 24(5):564-70. (Chapter 7, Appendix
F)
5. Accepted for publication in Diabetes Care (May 2012, In Press) - A Haynes, MK
Bulsara, C Bower, TW Jones, EA Davis. Cyclical variation in the incidence of
childhood Type 1 diabetes in Western Australia (1985 – 2010). (Chapter 5,
Appendix D)
vi
vii
Table of Contents
Chapter 1: Introduction ............................................................................. 1
1.1 Introduction.............................................................................................................2
1.2 Specific research aims.............................................................................................6
1.3 Thesis Approach......................................................................................................6
Chapter 2: Literature Review.................................................................. 11
2.1 Introduction...........................................................................................................12
2.2 What is diabetes?...................................................................................................12
2.2.1 Type 1 diabetes ..............................................................................................13
2.2.2 Type 2 diabetes ..............................................................................................13
2.2.3 Less common types of childhood diabetes.....................................................14
2.3 Significance of Type 1 diabetes ............................................................................14
2.4 Epidemiology of childhood Type 1 diabetes ........................................................17
2.4.1 Incidence ........................................................................................................17
2.4.2 Gender ............................................................................................................19
2.4.3 Age .................................................................................................................19
2.4.4 Temporal trends .............................................................................................20
2.4.5 Childhood Type 1 diabetes in Western Australia so far ................................22
2.5 Aetiology of childhood Type 1 diabetes ...............................................................23
2.6 Genetic factors ......................................................................................................28
2.7 Environmental factors ...........................................................................................34
2.7.1 Geographical variation in incidence of childhood Type 1 diabetes...............35
2.7.2 Socioeconomic status.....................................................................................36
2.7.3 Hygiene hypothesis ........................................................................................36
viii
2.7.4 Seasonal variation ..........................................................................................37
2.7.5 Viral infections...............................................................................................38
2.7.6 Gut microbiome .............................................................................................39
2.7.7 Vitamin D deficiency .....................................................................................41
2.7.8 Dietary factors................................................................................................43
2.7.9 Rapid growth, overweight and obesity...........................................................45
2.7.10 Perinatal factors............................................................................................47
2.8 Chapter summary ..................................................................................................49
Chapter 3: Background............................................................................ 51
3.1 Introduction...........................................................................................................52
3.2 Western Australia..................................................................................................52
3.2.1 Size.................................................................................................................52
3.2.2 Climate ...........................................................................................................52
3.2.3 Population ......................................................................................................53
3.2.4 Ethnicity .........................................................................................................54
3.2.5 Population distribution...................................................................................54
3.3 Data sources ..........................................................................................................54
3.3.1 Western Australian Children’s Diabetes Database ........................................54
3.3.2 Western Australian Data Linkage System .....................................................56
3.3.3 Hospital Morbidity Data System....................................................................58
3.3.4 Midwives’ Notification System .....................................................................59
3.3.5 Deaths registry ...............................................................................................59
3.4 Chapter Summary..................................................................................................60
Chapter 4: Methods .................................................................................. 61
4.1 Introduction...........................................................................................................62
ix
4.2 Ethical Issues.........................................................................................................62
4.2.1 Ethics Approval..............................................................................................62
4.2.2 Consent...........................................................................................................62
4.2.3 Confidentiality ...............................................................................................63
4.3 Outcome of Interest - Type 1 diabetes ..................................................................63
4.4 Study Cohorts........................................................................................................65
4.4.1 Incidence and trends of childhood Type 1 diabetes in Western Australia.....65
Case identification...............................................................................................65
The capture-recapture method.............................................................................66
Population data and calculation of incidence rates .............................................68
4.4.2 Perinatal factors and childhood Type 1 diabetes in Western Australia .........69
Case definition and identification .......................................................................69
Population data....................................................................................................69
Calculation of incidence rates .............................................................................70
Record linkage and perinatal data extraction ......................................................70
Validation of record linkage process...............................................................71
Perinatal data validation......................................................................................73
Missing values.................................................................................................73
Improbable values ...........................................................................................74
4.5 Measuring Socioeconomic Status .........................................................................74
4.5.1 Socioeconomic Indexes for Area (SEIFA) ....................................................74
4.6 Statistical analysis .................................................................................................76
4.7 Chapter Summary..................................................................................................76
Chapter 5: Incidence and incidence rate trends of childhood Type 1
diabetes in Western Australia.................................................................. 79
x
5.1 Preface...................................................................................................................80
5.1.1 Hypotheses .....................................................................................................80
5.1.2 Initial study period: 1985 - 2002....................................................................81
5.1.3 Follow-up study period: 1985 - 2010.............................................................81
5.2 Initial study: Continued increase in the incidence of childhood Type 1 diabetes in
a population-based Australian sample (1985-2002) ...................................................82
5.2.1 Abstract ..........................................................................................................82
Aims/hypothesis..................................................................................................82
Methods...............................................................................................................83
Results.................................................................................................................83
Conclusions/interpretation ..................................................................................83
5.2.2 Introduction....................................................................................................83
5.2.3 Subjects and methods.....................................................................................85
5.2.4 Statistical analysis ..........................................................................................86
5.2.5 Results............................................................................................................87
Ascertainment .....................................................................................................87
Overall incidence ................................................................................................88
Gender .................................................................................................................88
Age group............................................................................................................88
Seasonal variation ...............................................................................................90
5.2.6 Discussion ......................................................................................................91
5.3 Follow-up study: Incidence of childhood Type 1 diabetes in Western Australia
(1985 - 2010)...............................................................................................................95
5.3.1 Introduction....................................................................................................95
5.3.2 Aims ...............................................................................................................96
5.3.3 Methods..........................................................................................................96
xi
5.3.4 Statistical Analysis .........................................................................................96
5.3.5 Results............................................................................................................97
Additional cases ..................................................................................................97
Overall incidence ................................................................................................97
Gender .................................................................................................................98
Age group............................................................................................................98
5.3.6 Discussion ....................................................................................................102
5.3.7 Acknowledgements......................................................................................108
5.4 Chapter Summary................................................................................................108
Chapter 6: Incidence of childhood Type 1 diabetes in Western
Australia by place of residence and socioeconomic status .................. 111
6.1 Preface.................................................................................................................112
6.1.1 Hypotheses ...................................................................................................113
6.1.2 Initial study period: 1985 - 2002..................................................................113
6.1.3 Follow-up study period: 1985 - 2010...........................................................114
6.2 Initial study: Independent effects of socioeconomic status and place of residence
on the incidence of childhood Type 1 diabetes in Western Australia (1985 – 2002)116
6.2.1 Abstract ........................................................................................................116
Objective ...........................................................................................................116
Methods.............................................................................................................116
Results...............................................................................................................116
Conclusions.......................................................................................................117
6.2.2 Introduction..................................................................................................117
6.2.3 Methods........................................................................................................119
Analysis by place of residence..........................................................................120
xii
Analysis by socioeconomic status.....................................................................121
6.2.4 Statistical analyses .......................................................................................122
6.2.5 Results..........................................................................................................123
Ascertainment ...................................................................................................123
Place of residence..............................................................................................123
Socioeconomic status........................................................................................125
Independent effects of place of residence and socioeconomic status ...............126
6.2.6 Discussion ....................................................................................................127
6.3 Follow-up study: Regional variation in incidence of childhood Type 1 diabetes in
Western Australia (1985 – 2010) ..............................................................................131
6.3.1 Introduction..................................................................................................131
6.3.2 Aims .............................................................................................................131
6.3.3 Methods........................................................................................................132
6.3.4 Results..........................................................................................................132
Additional cases ................................................................................................132
Mean incidence by area.....................................................................................132
Incidence rate trends by area.............................................................................133
6.3.5 Discussion ....................................................................................................135
6.4 Chapter Summary................................................................................................137
Chapter 7: Perinatal risk factors for childhood Type 1 diabetes in
Western Australia ................................................................................... 139
7.1 Preface.................................................................................................................140
7.1.1 Hypotheses ...................................................................................................141
7.2 Perinatal risk factors for childhood Type 1 diabetes in Western Australia - a
population based study (1980 – 2002) ......................................................................143
xiii
7.2.1 Abstract ........................................................................................................143
Aims ..................................................................................................................143
Methods.............................................................................................................143
Results...............................................................................................................143
Conclusions.......................................................................................................144
7.2.2 Introduction..................................................................................................144
7.2.3 Patients and Methods ...................................................................................146
Study cohorts.....................................................................................................147
Place of residence & Socioeconomic status at time of birth.............................147
7.2.4 Statistical Analysis .......................................................................................149
7.2.5 Results..........................................................................................................149
Data linkage ......................................................................................................149
All live births ....................................................................................................150
Pre-existing maternal diabetes, gestational diabetes, ethnicity, plurality .....150
Month of Birth ..................................................................................................151
Singleton, Caucasian live births, no maternal diabetes.....................................151
Place of residence at time of birth.................................................................154
Socioeconomic status at time of birth ...........................................................154
7.2.6 Discussion ....................................................................................................155
7.3 Chapter Summary................................................................................................159
Chapter 8: Synopsis ................................................................................ 161
8.1 Preface.................................................................................................................162
8.2 Succinct summary of findings.............................................................................162
8.3 Comparison with other studies............................................................................163
8.3.1 Temporal trends ...........................................................................................163
xiv
8.3.2 Gender ..........................................................................................................165
8.3.3 Age at diagnosis ...........................................................................................166
8.3.4 Regional variation ........................................................................................167
8.3.5 Perinatal risk factors.....................................................................................170
8.4 Strengths and limitations.....................................................................................172
8.4.1 Based on whole population at risk over a long study period .......................172
8.4.2 Accurate classification of Type 1 diabetes...................................................173
8.4.3 Use of linked data.........................................................................................173
8.4.4 Socioeconomic status measured at area, not individual level ......................174
8.4.5 Heterogeneity within non-urban areas of Western Australia.......................176
8.4.6 No information available on interstate or overseas migration .....................177
8.5 Discussion ...........................................................................................................177
8.6 Implication of Findings.......................................................................................187
8.7 Recommendations for Future Research ..............................................................189
8.7.1 Cyclical variation in incidence.....................................................................189
8.7.2 Regional variation in incidence....................................................................190
8.7.3 Childhood obesity ........................................................................................192
8.7.4 Smoking in pregnancy..................................................................................192
8.7.5 Mode of delivery..........................................................................................193
8.7.6 Genetic factors .............................................................................................193
8.7.7 Clinical and basic research...........................................................................194
8.7.8 Continued monitoring of incidence..............................................................195
xv
List of Tables
Table 5.1: Mean annual age- and sex- specific incidence rates per 100,000 population
(95% CI) in Western Australia from 1985 to 2002.........................................................90
Table 5.2: Mean annual age- and sex- specific incidence rates per 100,000 population
(95% CI) in Western Australia from 1985 to 2010.........................................................99
Table 5.3: Goodness of fit measurements for different cyclical models fitted to Western
Australian Type 1 diabetes incidence data (1985 – 2010) ............................................101
Table 6.1: Number of cases, total person years at risk, non-Indigenous person years at
risk and mean incidence rate per 100,000 person years (95% CI) in Western Australia,
from 1985 to 2002, by geographical area .....................................................................124
Table 6.2: Number of cases, total person years at risk, non-Indigenous person years at
risk, mean incidence rate per 100,000 person years (95% CI) in Western Australia, from
1985 to 2010, by geographical area ..............................................................................134
Table 7.1: Number of births by case, gender and age group at diagnosis for the cohort
of all live births in Western Australia between 1980 and 2002, and the sub-cohort of
singleton, Caucasian live births with no maternal pre-existing or gestational diabetes
.......................................................................................................................................150
Table 7.2: Number of cases, total person years at risk, incidence of T1DM (per 100,000
person years), unadjusted and adjusted incidence rate ratios by maternal age, birth
weight, gestational age and birth order .........................................................................153
xvi
List of Figures
Figure 2.1: Mean incidence of Type 1 diabetes in children under 14 years of age (per
100 000 per year), comparing countries worldwide........................................................18
Figure 2.2: A schematic diagram of the natural history of Type 1 diabetes..................25
Figure 3.1: Population distribution of Western Australia showing urban, rural and
remote demographic zones as defined by the Western Australia Department of Health
.........................................................................................................................................53
Figure 3.2: Data collections of the Western Australian Data Linkage System ............57
Figure 4.1: The capture-recapture method for estimating completeness of case
ascertainment...................................................................................................................67
Figure 4.2: Place of birth for cases not linked to Midwives’ Notification System........72
Figure 5.1: The trend in the incidence of Type 1 diabetes in 0-14 year olds in Western
Australia from 1985 to 2002 ...........................................................................................89
Figure 5.2: Seasonal variation in incidence of Type 1 diabetes in 0-14 year olds in
Western Australia............................................................................................................91
Figure 5.3: Trends in the age-specific annual incidence of Type 1 diabetes in 0-14 year
olds in Western Australia from 1985 to 2002 .................................................................92
Figure 5.4: Annual age standardised incidence of Type 1 diabetes in 0-14 year olds in
Western Australia from 1985 to 2010...........................................................................100
xvii
Figure 5.5: The trend in incidence of Type 1 diabetes in Western Australia from 1985
to 2010...........................................................................................................................100
Figure 5.6: Annual age standardised incidence of Type 1 diabetes in Western Australia
by gender.......................................................................................................................101
Figure 5.7: Trends in the age-specific annual incidence of Type 1 diabetes in 0-14 year
olds in Western Australia from 1985 to 20100 .............................................................102
Figure 6.1: Population distribution of Western Australia ............................................119
Figure 6.2: Annual age-standardised incidence of Type 1 diabetes in 0-14 year olds in
urban and non-urban areas of Western Australia..........................................................125
Figure 6.3: Mean incidence of type 1 diabetes in 0-14 year olds in Western Australia
from 1985 to 2002 by socioeconomic group.................................................................126
Figure 6.4: Observed annual age-standardised incidence of Type 1 diabetes in 0-14
year olds in urban and non-urban areas of Western Australia from 1985 to 2010 with
estimated incidence rate trends ....................................................................................134
Figure 7.1: Relationship between gestational age and birth weight and the incidence of
Type 1 diabetes .............................................................................................................152
xviii
List of Appendices
Appendix A: Consent for data collection……………………………………..259
1. Information sheet for parents/guardians of children aged 0 to 9 years
2. Information sheet for children aged 10 to 14 years
3. Consent form
Appendix B: Midwives’ Notification System………………………………..265
1. Data collection form
2. List of perinatal data variables stored in the Midwives’ Notification System
Appendix C: Perinatal data validation………………………………………...275
1. Variables not recorded for the whole study period
2. Variables with improbable values
3. Coding changes made in 1997
4. ICD Codes for Diabetes and related conditions
Appendix D: Outputs arising from Chapter 5………………………………..285
1. Conference and seminar presentations by Dr Aveni Haynes
2. PDF of published journal article:
Haynes A, Bower C, Bulsara MK, Jones TW and Davis EA. Continued increase
in the incidence of childhood Type 1 diabetes in a population-based Australian
sample (1985-2002). Diabetologia, 2004. 47(5): 866-870
3. Co-author declarations for published journal article
4. PDF of invited commentary:
Haynes A and Davis EA. Why is Type 1 diabetes becoming more common in
children? Diabetes Management Journal, 2006. 17: 7
5. Co-author declaration for invited commentary
6. In Press (Accepted for publication May 2012) - Brief report:
Haynes A, Bulsara MK, Bower C, Jones TW and Davis EA. Cyclical variation
in the incidence of childhood Type 1 diabetes in Western Australia (1985 –
2010). Diabetes Care
7. Co-author declarations for brief report
xix
Appendix E: Outputs arising from Chapter 6………………………………..319
1. Conference and seminar presentations by Dr Aveni Haynes
2. PDF of published journal article:
Haynes A, Bulsara MK, Bower C, Codde JP, Jones TW and Davis EA.
Independent effects of socioeconomic status and place of residence on the
incidence of type 1 diabetes in Western Australia. Pediatric Diabetes, 2006. 7:
94-100
3. Co-author declarations for published work
Appendix F: Outputs arising from Chapter 7………………………………..337
1. Conference and seminar presentations by Dr Aveni Haynes
2. PDF of published journal article:
Haynes A, Bower C, Bulsara MK, Finn J, Jones TW and Davis EA. Perinatal
risk factors for childhood Type 1 diabetes in Western Australia--a population-
based study (1980-2002). Diabetic Medicine, 2007. 24(5): 564-570
3. Co-author declarations for published work
xx
List of Abbreviations
AGEs Advanced glycation end-products
AIC Aikake Information Criterion
BABYDIAB Study of offspring of parents with Type 1 diabetes
CI Confidence Interval
DAISY Diabetes Autoimmunity Study of the Young (Denver, USA)
DCCT Diabetes Control and Complications Trial
DIAMOND Multinational Project for Childhood Diabetes
(DIAbetes MONDiale)
DIPP Type 1 Diabetes Prediction and Prevention Project (Finland)
DLU Data Linkage Unit
ENDIA Environmental Determinants of Islet Autoimmunity (Australia)
EURODIAB EUROpe and DIABetes Study (Collaborative European study)
HMDS Hospital Morbidity Data Linkage System
IRR Incidence Rate Ratio
PMH Princess Margaret Hospital (Perth, Western Australia)
MIDIA Environmental Triggers of Type 1 Diabetes (Norway)
miRNAs MicroRNAs
MNS Midwives’ Notification System
NIP Nutritional Intervention to Prevent Type 1 Diabetes (Denver)
OECD Organization for Economic Co-operation and Development.
SEIFA Socioeconomic Indexes for Area
T1DGC Type 1 Diabetes Genetics Consortium
TEDDY The Environmental Determinants of Type 1 Diabetes
T1DM Type 1 diabetes
xxi
TRIGR Trial to Reduce IDDM in the Genetically at Risk
WA Western Australia
WACDD Western Australia Childhood Diabetes Database
WADLS Western Australia Data Linkage System
xxii
Acknowledgements
Thank you to Professor Tim Jones (Head of Department, Diabetes & Endocrinology,
Princess Margaret Hospital) who many years ago when I was trying to find a way of
applying myself, gave me the opportunity to discover what the world of medical
research was all about. Since then he has continued to be very supportive of my work
and I have enjoyed working with him and the fabulous research team he has built.
Sincere thanks go to my supervisors Professor Carol Bower and Dr Elizabeth Davis
who have both been very encouraging and inspiring. I have learnt so much from them
both and hope that I can continue to build on these skills and contribute to the
knowledge of childhood Type 1 diabetes long into the future. In addition, Professor
Max Bulsara has been an invaluable support in terms of providing me with statistical
advice and mentoring.
Thanks also go to all the patients who willingly consent to having their details recorded
on the Western Australian Children's Diabetes Database, without which this work
would not have been possible. Several other people helped to make the studies done in
this research possible whom I would like to acknowledge and thank: Maree Grant for
management of the Western Australian Children's Diabetes Database; Dr Jim Codde
for providing me with population data and advice on the use of socioeconomic index
values; Diana Rosman for her assistance with record linkage to the Western Australian
Data Linkage System; Hoan Nguyen for his assistance with mapping cases to census
districts and providing me with the socioeconomic data; Peter Cosgrove, Margaret
Wood, Vivien Gee and Alan Joyce for providing me with the perinatal data and being
available for advice regarding this data collection.
xxiii
In addition, I am grateful to the Juvenile Diabetes Research Foundation and Diabetes
Australia who partially funded the research presented in this thesis.
Finally, but most importantly, thank you to my husband and life partner, Matthew
Haynes and my mother-in-law Tricia Haynes. Without their endless support,
encouragement and time spent looking after our beautiful young children, I would never
have been able to complete this research or write this thesis.
1
Chapter 1: Introduction
2
1.1 Introduction
Diabetes is a heterogeneous group of disorders which occur when the body’s energy
metabolism is disturbed by a deficiency of insulin action, insulin secretion or both [1,
2]. There are two main types of diabetes in childhood. Type 1 diabetes, which used to
be known as insulin-dependent or juvenile-onset diabetes, and Type 2 diabetes which
used to be known as non-insulin dependent or adult-onset diabetes.
Type 1 diabetes is generally accepted to be the result of immune mediated destruction of
the insulin producing pancreatic beta cells in genetically susceptible individuals. It can
occur at any age but usually starts in people younger than 30. Some children with Type
1 diabetes have a rapid onset of symptoms and present within days, whilst others have a
slow onset of symptoms over several months [1-3]. In contrast, Type 2 diabetes is
thought to be due to an insensitivity to insulin action rather than insulin deficiency. The
symptoms of Type 2 diabetes normally appear gradually and diagnosis may not occur
until some time after the onset of disease. The onset of Type 2 diabetes is usually in
middle age [1, 2], however in recent years there has been an increase in the number of
children being diagnosed with Type 2 diabetes associated with the increasing
prevalence of obesity in childhood [4].
Type 1 diabetes accounts for over 90% of diabetes occurring in childhood [5] and is the
disease of interest in the research studies presented in this thesis. It affects over 400,000
children worldwide [6] and in Australia an average of two children are diagnosed with
Type 1 diabetes every day [7].
People diagnosed with Type 1 diabetes require lifelong insulin replacement in order to
3
survive. Before the 1920s and the discovery of insulin by Banting and Best [8],
children diagnosed with Type 1 diabetes would have died quickly, within days of the
clinical onset of the disease. Nowadays, people diagnosed with Type 1 diabetes are
treated with insulin replacement either by multiple insulin injections a day, or by insulin
pumps which provide a continuous subcutaneous insulin infusion. In conjunction with
daily insulin therapy, people with Type 1 diabetes have to undertake four to eight finger
pricks a day to check that their blood glucose levels are within specified limits. Whilst
trying to avoid their blood glucose level from getting too high, there is also the constant
fear of hypoglycaemia, or blood glucose levels falling too low. For example, parents of
children with Type 1 diabetes report not being able to sleep well at night for fear of their
child’s blood glucose level dropping too low overnight and resulting in the child having
a seizure or coma.
In addition to these daily issues facing children with Type 1 diabetes, many of them go
on to suffer long term complications of the disease, the severity of which are increased
by longer duration of the disease and poor glycaemic control. The Diabetes Control and
Complications Trial (DCCT) reported that after having Type 1 diabetes for 30 years,
47% of patients developed diabetic eye disease, 17% developed kidney disease and 14%
developed cardiovascular disease [9]. Children diagnosed with Type 1 diabetes have a
three to four times higher mortality rate compared to the general population [10-12].
Therefore, Type 1 diabetes is a serious, chronic and lifelong condition which affects a
substantial number of children and their families and is a significant cause of morbidity
and mortality. Despite many years of research, the aetiology and natural history of
Type 1 diabetes remain poorly understood and there is still no known prevention or cure
for the disease.
4
In an effort to further the understanding of Type 1 diabetes, standardised, prospective
population-based diabetes incidence registers were established in many countries,
including Australia, in the mid-1980s [13, 14]. In most countries, data from these
registers have shown a continued increase in the incidence of childhood Type 1 diabetes
over the past few decades but the cause of this increase has not yet been identified [15-
18].
In Western Australia, a diabetes register called the Western Australian Children's
Diabetes Database was established at Princess Margaret Hospital in 1987, with data
being available for cases diagnosed from 1985 onwards . All children diagnosed with
diabetes in Western Australia are managed by hospital-based physicians at the time of
diagnosis and Princess Margaret Hospital is the only tertiary referral centre for children
diagnosed with diabetes in the State. The multidisciplinary diabetes team based at
Princess Margaret Hospital provide ongoing care to children diagnosed with Type 1
diabetes until their transition to adult services from the age of 16 years. Diabetes clinics
are held at Princess Margaret Hospital and in several regional and remote centres
around Western Australia. Therefore, complete and accurate data are available on all
children diagnosed with Type 1 diabetes in Western Australia and the case
ascertainment level of the Western Australian Children’s Diabetes Database is estimated
to be >99% complete [19].
As well as complete data being available on all children in Western Australia diagnosed
with Type 1 diabetes, there are several other State-wide population-based data
collections that are available via the Western Australian Data Linkage System [20, 21],
including data on all hospital admissions and perinatal data on all births in the State.
5
The availability of such complete population-based data resources in Western Australia
provides a unique opportunity to undertake epidemiological studies of childhood Type 1
diabetes in this population. By using the data available for all cases in Western
Australia for this research, the studies will be a comparison of risk factors between all
cases in the State and the remainder of the total population at risk. This enables a
powerful and accurate analysis of risk factors, which is rarely achievable in other
populations. The ultimate goal of identifying risk factors for childhood Type 1 diabetes
is to be able to contribute knowledge towards the early intervention and prevention, or
cure of this serious and lifelong condition.
Therefore, the overall aims of this research were to determine the trends in childhood
Type 1 diabetes in Western Australia from 1985 onwards, and to identify risk factors for
the disease in this population. The epidemiology of childhood Type 1 diabetes in
Western Australia has previously only been described from 1985 to 1992, when the
incidence was found to have increased significantly in both sexes and across all age
groups [19, 22]. So the first question to answer in this research was whether or not the
incidence of childhood Type 1 diabetes in Western Australia has continued to increase
since 1992? If so, has it been increasing in both genders, all age groups, all areas of
Western Australia and in all years, or are there specific patterns in incidence which we
can use to identify potential risk factors for the disease?
Prior to the research presented in this thesis, no previous studies had been undertaken in
Western Australia to examine the incidence of childhood Type 1 diabetes by place of
residence or socioeconomic status, or to investigate the association between perinatal
factors and the incidence of childhood Type 1 diabetes. In addition, no record linkage
had previously been undertaken between the Western Australian Children’s Diabetes
6
Database at Princess Margaret Hospital and other State-wide data collections linked by
the Western Australian Data Linkage System.
Therefore, the specific study aims of the research presented in this thesis were as listed
below.
1.2 Specific research aims
1. To determine the incidence of Type 1 diabetes in children aged 0-14 years in
Western Australia and to analyse the incidence rate trends by calendar year, gender,
age at diagnosis and for seasonal variation.
2. To examine the incidence of Type 1 diabetes in children aged 0-14 years in Western
Australia by place of residence and socioeconomic status.
3. To investigate the association between perinatal factors and the incidence of
childhood Type 1 diabetes in children aged 0-14 years in Western Australia.
1.3 Thesis Approach
This thesis is presented as a series of publications, together with additional findings
obtained from follow-up studies that have been undertaken since the published studies
were performed. Following on from the first research aim of determining the incidence
and incidence rate trends of childhood Type 1 diabetes, the consecutive publications
demonstrate the logical development of research ideas examined in this thesis.
Chapter 2 (Literature review) which follows this Introduction chapter, provides a review
of the literature on childhood Type 1 diabetes that is relevant to the research presented
7
in this thesis. An overview of the classification of childhood Type 1 diabetes is
provided to differentiate childhood Type 1 diabetes, which is the outcome of interest in
this study, from other types of diabetes in childhood. A review is then provided of the
literature on epidemiological studies of childhood Type 1 diabetes which have been
previously undertaken in Western Australia and worldwide. Finally, a review is
provided of the literature regarding the pathophysiology and aetiology of childhood
Type 1 diabetes including genetic and environmental factors that have been previously
associated with this disease and which were of interest in this research.
Chapter 3 (Background) describes the context in which the research presented in this
thesis was undertaken. A brief description of the demographic features of Western
Australia is provided, including its climate and population. A more detailed description
is then given of the total population-based data resources available in Western Australia
that were used in this study, and of the data linkage processes used.
To avoid repetition of content, Chapter 4 (Methods) contains a general overview of the
study design and methods used in this research. The subjects, specific methods and
statistical analyses used for achieving the individual research aims are described in more
detail in the relevant section of each chapter relating to the publications.
Chapters 5, 6 and 7 comprise the results section of this thesis and are structured around
three publications. Each of these chapters contains a published journal article in its
original format. The numbering of tables, figures and references in the publications has
been modified to fit with the flow of the entire text and to enable a single bibliography
to be included at the end of the thesis. The original PDF version of each publication has
been included in the Appendix.
8
Chapters 5 and 6 are divided into two main sections. The first section contains the
unedited version of a published journal article as detailed above. The second section of
these two chapters contains details on the follow-up studies which were undertaken to
update the research findings of the original, published studies. The follow-up studies
used the same methods as the original studies, but with an extended study period. This
is explained in greater detail in the preface to both chapters 5 and 6.
Chapter 5 relates to the first study aim, which was to analyse the incidence of childhood
Type 1 diabetes in Western Australia by calendar year, gender and age at diagnosis.
The first section of chapter 5 consists of the publication titled "Continued increase in
the incidence of childhood Type 1 diabetes in a population-based Australian sample
(1985 - 2002)"[23] (Appendix D). This publication details the background, specific
methods and results relating to the investigation of the incidence of Type 1 diabetes in
children aged 0-14 years in Western Australia from 1985 to 2002. The second section
of chapter 5 presents findings related to the same research aim but with the study period
extended from the end of 2002 to the end of 2010. A brief report describing the
findings from this follow-up study has been accepted for publication in the journal
Diabetes Care and is currently in press.
The first section of chapter 6 consists of the publication titled "Independent effects of
socioeconomic status and place of residence on the incidence of childhood Type 1
diabetes in Western Australia" [24] (Appendix E). This publication details the
background, methods and results relating to the second specific research aim, which was
to examine the incidence of Type 1 diabetes in children aged 0-14 years in Western
Australia by place of residence and socioeconomic status. Similar to chapter 5, the
second section of chapter 6 presents the findings from a follow-up study which was
9
undertaken to re-examine the incidence of childhood Type 1 diabetes in Western
Australia by place of residence at the time of diagnosis. Except for extending the study
period from the end of 2002 to the end of 2010, the methods used for this follow-up
study were similar to those used for the original, published study.
Chapter 7 relates to the third specific research aim, which was to investigate the
association between perinatal factors and the incidence of childhood Type 1 diabetes in
children aged 0-14 years in Western Australia. Details of the background, specific
methods and results relating to this research aim are presented in chapter 7 as a
publication titled "Perinatal risk factors for childhood Type 1 diabetes in Western
Australia - a population based study (1980 - 2002)" [25] (Appendix F).
The final chapter, Chapter 8 (Synopsis), brings together the research findings presented
in chapters 5, 6 and 7 and discusses them in relation to each other as well as in relation
to findings in other populations. A discussion is provided on how the findings
presented in this thesis may contribute to the understanding of childhood Type 1
diabetes and the implications of these findings for estimating future disease rates and
planning of health services and budgets. The various strengths and limitations of the
studies are discussed, as well as the steps taken to minimise the effect of such
limitations. Finally, recommendations are made for future research into childhood Type
1 diabetes epidemiology in Western Australia so that epidemiological research can
continue to contribute valuable information to the global understanding of this
significant disease, whose cause is still not known.
10
11
Chapter 2: Literature Review
12
2.1 Introduction
This chapter provides a review of the literature relevant to the research presented in this
thesis. Key findings from studies related to the epidemiology of childhood Type 1
diabetes are described and analysed and any gaps in knowledge that this thesis aims to
address are highlighted.
The review is divided into subsections, each pertaining to a main topic relevant to the
studies undertaken in this thesis. The first section describes the classification of
childhood diabetes, differentiating Type 1 diabetes, which is the outcome of interest in
this thesis, from other types of childhood diabetes. The subsequent sections provide a
review of various studies into childhood Type 1 diabetes epidemiology and how they
have contributed to furthering the understanding of the pathogenesis and aetiology of
this disease. This includes a review of genetic and environmental factors that have been
investigated in relation to childhood Type 1 diabetes in other populations.
2.2 What is diabetes?
Diabetes is a heterogeneous group of disorders, which occur when the body’s energy
metabolism is disturbed by a deficiency of insulin secretion, insulin action, or both [1,
2, 26, 27]. Insulin is a hormone produced by the pancreatic beta-cells that is required
for glucose uptake into cells. A deficiency in insulin secretion or action results in high
blood glucose levels, as well as disturbances in fat and protein metabolism.
There are 2 main types of diabetes in childhood: Type 1 diabetes, which used to be
known as insulin-dependent or juvenile-onset diabetes and Type 2 diabetes which used
to be known as non-insulin dependent or adult-onset diabetes. Although over 90% of
13
people with diabetes in the population as a whole have Type 2 diabetes [5, 28], Type 1
diabetes is the commonest type of diabetes in childhood and accounts for approximately
90% of children diagnosed with diabetes under the age of 15 years [1, 29, 30].
2.2.1 Type 1 diabetes
Type 1 diabetes is generally thought to be due to cellular-mediated autoimmune
destruction of the pancreatic beta-cells which results in absolute insulin deficiency [1,
5]. Patients with Type 1 diabetes need lifelong treatment with insulin for survival.
Although Type 1 diabetes can occur at any age [31], the disease commonly occurs in
childhood and adolescence with the peak age at onset being around puberty. The
clinical presentation of Type 1 diabetes can vary from non-emergency symptoms such
as polydipsia, polyuria, weight loss and enuresis to emergency symptoms such as severe
dehydration, shock and diabetic ketoacidosis [3]. In Australia, most children diagnosed
with Type 1 diabetes are admitted to hospital at the time of diagnosis. Patients with
Type 1 diabetes can be identified using a combination of clinical and biochemical
features, together with serological evidence of the autoimmune process and genetic
markers [30]. Idiopathic Type 1 diabetes is a rare form of Type 1 diabetes that is more
common in individuals of African or Asian origin. It is a strongly inherited disease in
which patients have an absolute insulin deficiency but no evidence of autoimmunity.
Autoimmune Type 1 diabetes diagnosed in children aged 0 to 14 years is the outcome of
interest in this thesis.
2.2.2 Type 2 diabetes
Type 2 diabetes is characterised by insensitivity of the target cells to insulin action as
well as relative insulin deficiency [1, 5]. The symptoms of Type 2 diabetes normally
appear gradually and diagnosis may not occur until some time after the onset of disease.
14
People with Type 2 diabetes do not necessarily need insulin replacement for survival.
The onset of Type 2 diabetes is usually in middle age (>40 years). However, there has
been a worrying increase in the number of children being diagnosed with Type 2
diabetes, which is associated with the increasing prevalence of childhood obesity [29,
32, 33]. Many people who have Type 2 diabetes are overweight or obese and they often
have a family history of the disease. In addition to the higher risk of Type 2 diabetes in
children who are overweight, there is an increased risk of Type 2 diabetes in children of
Indigenous descent [4, 34, 35].
2.2.3 Less common types of childhood diabetes
Other less common causes of diabetes in childhood include monogenic diabetes,
neonatal diabetes, drug-induced diabetes and secondary diabetes [1, 5]. Monogenic
diabetes refers to a group of disorders characterised by genetic defects of beta-cell
function or insulin action [30]. Neonatal diabetes is a rare condition characterised by
insulin-requiring hyperglycaemia occurring in the first three months of life. Other rare
causes of childhood diabetes include drug-induced diabetes, stress-induced diabetes and
mitochondrial diabetes [30]. Secondary diabetes occurs as a consequence of
degenerative damage to the pancreas associated with other medical conditions. For
example, the pancreas can be affected in patients with cystic fibrosis, resulting in
impaired insulin secretion and secondary diabetes [36]. In Japan, a subtype of Type 1
diabetes called fulminant diabetes accounts for approximately 20% of acute-onset cases
of diabetes [37]. Fulminant diabetes is characterised by a very acute onset, destruction
of the pancreatic beta-cells and an absence of islet autoantibodies [38].
2.3 Significance of Type 1 diabetes
The outcome of interest in this thesis is childhood Type 1 diabetes which is a lifelong,
15
chronic condition with serious and sometimes life-threatening short and long term
consequences. It places an enormous burden on the individual diagnosed with the
disease as well as their family, and adds significantly to the health burden to society in
general.
As described in chapter 1 (Introduction), children diagnosed with Type 1 diabetes need
lifelong insulin replacement therapy to survive. Without insulin, life-threatening
hyperglycaemia occurs, resulting in diabetic ketoacidosis or non-ketotic hyperosmolar
syndrome and death. In addition to either multiple daily injections of insulin or a
continuous subcutaneous insulin infusion, the children have to undertake numerous
daily finger prick tests to check that their blood glucose levels are within specified
limits and carefully monitor their diet and activity levels. Whilst trying to avoid their
blood glucose levels from getting too high, there is the constant fear of hypoglycaemia
or blood glucose levels falling too low, as this can result in a seizure or coma.
In the longer term, children with Type 1 diabetes are at significantly increased risk of
developing serious complications of the disease including eye, kidney and heart disease
[9, 39] and their mortality rate is three to four times higher than the general population
[11, 40]. In addition to this, they are at an increased risk of developing other
autoimmune conditions that are associated with Type 1 diabetes, such as coeliac disease
and autoimmune thyroid disease [41, 42]. Not surprisingly, together with the increased
health burden experienced by people with Type 1 diabetes, they also experience greater
social and economic burdens throughout the course of their life [43]. Children
diagnosed with Type 1 diabetes need lifelong care from a multidisciplinary team of
health professionals, and additional care has to be taken to ensure that their illness has a
minimal impact on their other physical, emotional and social development.
16
As well as the significant health and social costs experienced by children with Type 1
diabetes and their families, there is a large financial cost associated with having the
disease. These financial costs include direct costs related to the cost of medication and
hospitalisation, as well as the indirect costs of carer support and loss of productivity. In
2006, the estimated cost of Type 1 diabetes in Australia was between $430 to $570
million [44]. The average annual cost per person with Type 1 diabetes in Australia was
estimated to be $4,669, and this increased substantially for patients who also had
complications of the disease [44].
The International Diabetes Federation estimates that there are 480,000 children living
with Type 1 diabetes worldwide, and every year an additional 76,000 children are
diagnosed with this lifelong, serious disease [45]. In Australia, diabetes is one of the
most common chronic diseases in children. Between 2000 and 2008, an average of 2-3
children aged 0-14 years were diagnosed with Type 1 diabetes in Australia every day
[7]. The crude estimated prevalence of Type 1 diabetes in Australian children aged 0-
14 years in 2008 was 137.8 per 100,000 and this is predicted to rise [46]. Amongst the
OECD countries, Australia has the sixth highest prevalence of childhood Type 1
diabetes [46].
Therefore, Type 1 diabetes is an important cause of morbidity and mortality in
populations throughout the world. Despite several decades of research, the natural
history and aetiology of Type 1 diabetes are still not fully understood and there is no
known cure or prevention for this disease. The aim of this research is to analyse the
incidence and incidence rate trends of childhood Type 1 diabetes in Western Australia
to try and identify potential risk factors for the disease, thereby contributing to the
17
global understanding of who gets childhood Type 1 diabetes and why, so that
ultimately, the search for a prevention or cure to this serious disease can be accelerated.
2.4 Epidemiology of childhood Type 1 diabetes
2.4.1 Incidence
In order to further the understanding of childhood Type 1 diabetes, diabetes researchers
in many developed countries worked together in the 1980s to establish population-based
diabetes incidence registers [13] to monitor the incidence of childhood Type 1 diabetes
around the world. The registers were standardised according to the criteria of the
Diabetes Epidemiology Research International Group to enable the comparison of
incidence rates between different populations [13]. The Western Australia Children’s
Diabetes Database is a register that conforms to these standards and was established at
Princess Margaret Hospital in Western Australia in 1987. Retrospective data was
included in the register for cases diagnosed in 1985 and 1986 [19].
Large collaborative studies such as the EURODIAB Study in Europe [47] and the
worldwide DIAMOND Project [14, 48] have used data from these standardised registers
to greatly advance the understanding of childhood Type 1 diabetes epidemiology. For
example, data from these registers have shown that the incidence of childhood Type 1
diabetes varies widely across the world [15, 16, 18, 49, 50], ranging from 0.1/100,000
person years in China and Venezuela to over 40.9/100,000 person years in Finland [17]
(Figure 2.1). In Australia, the mean incidence of childhood Type 1 diabetes in 2008
was 22.8 per 100,000 person years (95% CI: 22.3–23.3) [7] which is considered to be in
the “very high” incidence category when compared to other countries around the world
[15].
18
Figure 2.1: Mean incidence of Type 1 diabetes in children under 14 years of age
(per 100 000 per year), comparing countries worldwide (Source: Definition,
epidemiology and classification of diabetes in children and adolescents. Paediatric
Diabetes, 2009[30])
19
The wide variation in incidence of childhood Type 1 diabetes around the world is thought to
be a reflection of the variation in ethnicity, and therefore genetic susceptibility to Type 1
diabetes, between different populations. However, large differences in incidence have been
reported in some neighbouring populations that are genetically similar [16, 51, 52] and
between different regions within the same country [53, 54], which indicates that differences
in environmental factors also contribute to the differences in incidence observed around the
world.
2.4.2 Gender
When analysed for variation by gender, in many populations no difference in the
incidence of childhood Type 1 diabetes has been observed between boys and girls [17,
55, 56]. In those that have reported a difference, in general a male excess in incidence
has been observed in countries with a high incidence, and a female excess in incidence
observed in countries with a low incidence [47, 57-59].
In Australia, the findings are inconsistent. A nationwide study reported no difference in
the incidence of childhood Type 1 diabetes between boys and girls [7, 60], whereas in
New South Wales, a higher mean incidence was observed in girls but a greater rate of
increase in incidence observed in boys [61]. One nationwide study reported a higher
mean incidence in boys aged 0-4 and 10-14 years old compared to girls the same age
[62], whereas a more recent study reported a higher mean incidence in boys compared
to girls in just the 0-4 year old age group [7]. So, the difference in incidence between
boys and girls in Australia remains unclear.
2.4.3 Age
In general, the incidence of childhood Type 1 diabetes increases with increasing age, the
20
highest incidence being observed in 10-14 year olds [15, 17]. This is consistent with
findings in Australia [62]. The peak age of onset for the disease observed in 10-14 year
olds corresponds with puberty and the significant metabolic changes that occur during
this developmental stage. As the onset of puberty is usually earlier in girls, the peak age
of onset of Type 1 diabetes is also usually earlier in girls compared to boys [55].
2.4.4 Temporal trends
Over the past few decades, there has been a steady increase in the incidence of
childhood Type 1 diabetes in many countries worldwide [16, 63, 64]. Between 1990
and 1999 the average rate of increase observed in countries around the world was 2.8%
a year [17]. A more recent study of 17 European countries reported an average annual
increase in incidence of 3.9% between 1989 and 2003 [18]. A much higher rate of
increase was observed in countries in Central-Eastern Europe which experienced major
political and economic changes during this time period. For example, the average
annual rate of increase in Poland over this time period was 9.3% compared to 1.3% in
Norway [18].
In Australia, one nationwide study reported an average increase of 2.8% a year between
2000 and 2006 [62]. A more recent study reported an average annual increase of 1.7%
between 2000 and 2008 [7], with the majority of the increase in incidence occurring
between 2000 and 2004, and the incidence appearing to have reached a plateau between
2005 and 2008 [7]. A marked increase in incidence of 9.3% a year was observed in
Victoria between 1999 and 2002 [65]. However, the very high rate of increase observed
in Victoria [65] may have been due to increasing ascertainment levels during the study
period as the National Diabetes Register, which was used as a secondary source of cases
in this study, was established in 1999 and was known to have incomplete data. A study
21
in New South Wales reported an average increase in incidence of childhood Type 1
diabetes of 2.8% a year between 1990 and 2002, with some suggestion of a plateau in
the incidence between 1997 and 2002 [61]. Evidence for a levelling off in the rate of
increase in incidence of childhood Type 1 diabetes in other populations includes a
recent study in Sweden [66]. In contrast, a steep non-linear increase in incidence has
been reported in other European populations including Finland [55], Switzerland [67]
and Austria [68].
In many countries, the incidence of childhood Type 1 diabetes has increased in both
genders [55] and all age groups, but more recently several countries have reported the
worrying trend of a greater rate of increase in children under the age of 5 years [18, 55,
68-71]. Based on current estimates, the number of cases of Type 1 diabetes in children
under the age of 5 years in Europe is expected to double between 2005 and 2020 [18].
Several studies have shown a corresponding decrease in the incidence of Type 1
diabetes in patients aged over 15 years [72-76], suggesting that the increase in
childhood Type 1 diabetes may be a reflection of more rapid disease progression and an
earlier age of onset, rather than more frequent disease initiation [77] or an increased
overall lifetime risk of the disease. In contrast to this, other studies have observed an
increase in both childhood and adult Type 1 diabetes [78, 79] showing that the
incidence is increasing in all age groups.
In Australia, a greater rate of increase in the incidence of childhood Type 1 diabetes in
children under the age of 5 years between 1999 and 2002 was reported in Victoria [65],
but no disproportionate increase was observed in this age group between 1990 and 2002
in New South Wales [61]. An increase in incidence was observed in both genders and
22
all age groups in an Australia-wide study from 2000 to 2008 [7] and in Western
Australia from 1985 to 1992 [22].
2.4.5 Childhood Type 1 diabetes in Western Austr alia so far
The first study to describe the epidemiology of childhood diabetes in Western Australia
was based on a school survey undertaken in 1984 [80]. This study estimated the
incidence to be 12.3 per 100,000 person years, an excess was observed in boys and no
cases of Type 1 diabetes were found in children of Indigenous descent [80]. Between
1985 and 1989, the mean incidence of childhood Type 1 diabetes in Western Australia
was 13.2 per 100,000 person years and no significant increase in incidence was
observed during this time [19]. The first significant increase in the incidence of
childhood Type 1 diabetes in Western Australia was reported in 1992 when it was found
to have increased from 15.5 per 100,000 person years in 1991 to 22.2 per 100,000
person years in 1992 [22]. The incidence was found to have increased significantly in
both genders and all age groups at this time [22]. No studies into the epidemiology of
childhood Type 1 diabetes have been undertaken in Western Australia since 1992.
Given the increases in incidence observed more recently in other populations, it is
important to determine any changes that may have occurred in Western Australia since
1992. Therefore, the first aim of the research presented in this thesis was to determine
the incidence and incidence rate trends of childhood Type 1 diabetes in Western
Australia from 1992 onwards, and to analyse the incidence rate trends by calendar year,
gender and age at diagnosis to see if the incidence in Western Australia is increasing at
a similar rate to that observed in other populations and to see if there has been a
disproportionate rate of increase in the youngest age group.
23
By continuing to investigate the epidemiology of childhood Type 1 diabetes, factors
important in the aetiology of childhood Type 1 diabetes that account for the continued
rise in this serious disease may be identified [81].
2.5 Aetiology of childhood Type 1 diabetes
Epidemiological studies can be used to describe who develops Type 1 diabetes and
analysis of such data may help to identify genetic and environmental factors that are
associated with disease risk [47, 82-85] and the development of islet autoimmunity [86-
88]. In conjunction with epidemiological studies, understanding of the aetiology and
natural history of Type 1 diabetes has been greatly increased by studies in animal
models of the disease and molecular biology.
Experimental studies in animal models of diabetes have been undertaken to investigate
the role of various genetic and environmental factors in the development of diabetes.
Although findings from animal studies cannot necessarily be directly extrapolated to
humans, such studies have greatly increased our knowledge regarding the pathogenesis
and immune mechanisms of diabetes [89-91].
In addition to knowledge gained from animal studies of diabetes, the understanding of
Type 1 diabetes has greatly increased over the past two decades by advances in
molecular biology. Studies investigating the role of humoral and cellular components
of the innate and adaptive immune responses have led to a much greater understanding
of the immunopathogenesis of Type 1 diabetes [92-95] [96]. It is now well established
that Type 1 diabetes, known to be an autoimmune condition, is the result of T-cell
mediated destruction of pancreatic beta-cells [97].
24
The first clue to Type 1 diabetes being an immune-mediated disease was in the 1960s,
when Gepts et al. described the infiltration of lymphocytes in pancreatic islets of
children who died within a few days of developing diabetes [98]. In the mid-1970s, the
presence of antibodies to islet cells was detected in the blood of people diagnosed with
insulin-dependent diabetes [99], indicating that it was an autoimmune condition.
Around the same time as this discovery, an association between childhood diabetes and
HLA genes on the short arm of chromosome 6 was reported [100]. Since these major
discoveries, it has become generally accepted that Type 1 diabetes is the result of
autoimmune mediated destruction of the insulin producing cells of the pancreas, called
pancreatic islet beta-cells [1, 2, 97].
Increasingly, Type 1 diabetes is thought to be a heterogeneous disorder with multiple
phenotypes of the disease [101]. This might explain why, when analysing aggregated
data on groups of patients, it has thus far been difficult to identify risk factors for the
disease. As described earlier in this chapter, both genetic and environmental factors
have been implicated in the aetiology of Type 1 diabetes and there are ongoing studies
investigating the role that such factors may play individually, or in combination, at
different time points in the natural history of the disease. Further complexity is likely as
the association between some environmental factors and Type 1 diabetes may depend
on the presence of certain genetic variants i.e. the presence of gene-environment
interactions.
The working model of the natural history of Type 1 diabetes is based on stages, starting
with a genetic susceptibility to the disease, followed by autoimmunity without clinical
symptoms of the disease, and finally clinical diabetes [102] (Figure 2.2). It is still not
25
known exactly which risk factors are important in the aetiology of Type 1 diabetes, and
whether or not they acts as initiators, propagators or precipitators of the disease process.
Figure 2.2: A schematic diagram of the natural history of Type 1 diabetes
Another difficulty in identifying factors that may be contributing to the continued global
increase in childhood Type 1 diabetes is that the time scale of the disease process that
results in clinical Type 1 diabetes varies and is not yet fully understood. This makes it
difficult to know at what time point factors may be acting to cause or propagate the
disease, and therefore what time point to measure risk factors in.
The autoimmune processes which cause a progressive destruction of the pancreatic
beta-cells may start months or years prior to the clinical onset of disease [103].
Similarly, prospective studies have shown that metabolic changes are also evident more
than two years before clinical onset of the disease [104]. However, the rate of beta-cell
destruction is highly variable [84, 105-107], and there is no definitive time period
Genetic Susceptibility Autoimmunity Clinical diabetes
Initiators Genes Environment
Primary prevention Prevent disease onset
Secondary prevention Halt disease
progression
Propagators/Precipitators Genes Environment
Tertiary prevention Preserve or regenerate beta-cell mass/ function & prevent complications
26
between initiation of the disease process and patients presenting unwell with clinical
symptoms of the disease. And finally, not all individuals who have evidence of islet
autoimmunity go on to develop clinical symptoms of the disease.
In order to try and understand the disease process of Type 1 diabetes more clearly, large
population-based longitudinal cohort studies have been undertaken in several, mainly
European, countries and collaborative studies established [108] to enable the analysis of
factors which may be associated with the pathogenesis of Type 1 diabetes. Examples of
such studies include the Denver Diabetes Autoimmunity in the Young Study (DAISY)
[109], the German BABYDIAB Study [110], the Finnish Type 1 Diabetes Prediction
and Prevention Study (DIPP) [111] and the multi-national The Environmental
Determinants of Diabetes in the Young Study (TEDDY) [112]. These study groups
have been investigating the role of multiple environmental and genetic factors in the
natural history and development of childhood Type 1 diabetes in children with an
increased genetic risk of Type 1 diabetes. Babies identified as having an increased
genetic risk of Type 1 diabetes have been followed up prospectively from birth and
measurements taken at regular intervals for islet autoantibodies and multiple
environmental factors thought to be associated with childhood Type 1 diabetes.
The autoantibodies measured include islet cell antibodies (ICA), antibodies to protein
tyrosine phosphotase (IA-2), insulin (IAA) and glutamic acid decarboxylase (GAD65).
More recently, it has been shown that measuring antibodies to the zinc transporter 8
islet autoantigen (ZnT8A) [113] increases the sensitivity of detecting diabetes-related
autoimmunity [114, 115]. Data from such longitudinal studies have shown that
antibodies to the islet cells can be detected in the blood of individuals many years prior
to clinical onset of the disease, and in some individuals antibodies may even be present
27
in utero or in very early life [111] [116]. In children diagnosed with Type 1 diabetes
under the age of 10 years, the majority show the first signs of autoimmunity before the
age of 2 years [111, 117]. More recently, results published from the BABYDIAB and
Finnish DIPP studies demonstrate that islet autoimmunity may be initiated very early in
life, including during pregnancy and the perinatal period. Therefore, genetic and
environmental factors acting early on during immune development and immune
maturation are likely to play significant roles in the initiation of islet autoimmunity and
the development of Type 1 diabetes in later life [86, 118].
In addition to the variable rate of autoantibody appearance, the pattern of autoantibody
seroconversion varies between individuals, and antibodies seem to appear sequentially
rather than all at the same time [119, 120]. Recently it has been reported that the
immunophenotype, or pattern of islet autoantibodies found at the time of diagnosis, has
changed over time with a more intense humoral autoimmune response evident in
patients diagnosed in later years [121].
The presence of islet autoantibodies without clinical symptoms of disease is sometimes
referred to as pre-diabetes. Data from longitudinal studies have shown that the number
and combination of islet autoantibodies can be used to predict which individuals are
more likely to go on to develop clinical Type 1 diabetes [122-128]. Individuals who
remain positive for multiple autoantibodies have a greater risk of being diagnosed with
Type 1 diabetes [126, 129-131], but the rate of progression to clinical Type 1 diabetes in
such individuals is highly variable [129]. Additional factors such as the titre and
affinity of islet autoantibodies and the age at which they appear, as well as the
individual’s HLA genotype and family history of Type 1 diabetes, increase the accuracy
of predicting which individuals are more likely to go on and develop the disease and
28
how long this is likely to take [125, 132-135]. Children who develop autoantibodies
before the age of 2 years are more likely to develop multiple autoantibodies and
progress to clinical Type 1 diabetes [127]. In contrast, children who first develop
autoantibodies after 2 years of age have a slower progression to multiple autoantibodies
and clinical diabetes [127]. However, not all individuals with detectable islet
autoantibodies go on to develop clinical diabetes and it is still not known why this is the
case [129-131, 136].
Therefore, despite recent advances in knowledge, the pathogenesis of Type 1 diabetes is
not fully understood and research into the natural history of the disease, as well as the
genetic and environmental factors which may be involved in its cause, is ongoing [83,
137]. Multiple genetic and environmental factors are thought to be involved in the
aetiology of Type 1 diabetes, and these are described below.
2.6 Genetic factors
It has been known for several decades that genetic factors play an important role in the
aetiology of childhood Type 1 diabetes. Evidence that the risk of childhood onset Type
1 diabetes is at least partly determined by genetic factors includes familial clustering of
the disease, concordance rates in monozygotic twins of 30 – 65% [138-140], and
differences in risk between different ethnic groups [35, 49],[80, 141].
Although ~85% of individuals who develop Type 1 diabetes have no family history of
the disease [138, 142], there is a heritable component of Type 1 diabetes and familial
clustering of the disease [143, 144]. The risk of developing childhood Type 1 diabetes
in individuals who have an affected first degree relative is ~6%, compared to 0.4% in
those without an affected first degree relative [145]. The pattern of disease transmission
29
may offer some insight into the mechanism of the disease. There is evidence of
differential transmission of the risk of Type 1 diabetes in offspring of affected fathers
compared to affected mothers [146]. The risk of developing childhood Type 1 diabetes
is 5-7% in offspring of affected fathers and 2-3% in offspring of affected mothers [147-
151]. Children with an affected sibling have an ~8% risk of developing the disease
[151]. In addition, the risk of Type 1 diabetes is higher in the relatives of individuals
who developed Type 1 diabetes at a younger age [148, 152]. No precise explanation for
these differences in disease transmission has yet been found. Perhaps affected mothers
are less fertile or suffer a greater number of pregnancy losses reducing the likelihood of
them having affected offspring [153].
Another possibility is that the lower risk of Type 1 diabetes in the offspring of mothers,
compared to fathers with Type 1 diabetes, may be due to immunological or metabolic
protection that occurs during pregnancy and reduces the risk of the disease in later life
[154]. Recently, findings from the longitudinal BABYDIAB study have shown that
there is a lower incidence of islet autoimmunity at the age of 9 months in the offspring
of mothers compared to fathers with Type 1 diabetes [155]. Pre-existing maternal Type
1 diabetes is known to influence the in utero environment. How such influences reduce
the risk of developing islet autoimmunity and Type 1 diabetes in their offspring is not
yet known, but factors such as infant birth weight, maternal HbA1c and fetal exposure
to maternal islet autoantibodies have been associated with changes in the risk of islet
autoimmunity in their offspring [118].
Over the past four decades, several genetic variants that confer susceptibility to, or
protection from, Type 1 diabetes have been identified. However, variations of the Class
30
II HLA haplotypes in the major histocompatibility complex account for approximately
40% of the genetic contribution to Type 1 diabetes [156, 157], and for 30-60% of the
familial clustering of the disease [31].
The association between genes in the HLA region and Type 1 diabetes was first
observed in the 1970s [100, 158]. Since that time, the association between HLA
genotypes and the risk of Type 1 diabetes has been intensively investigated. The Type
1 Diabetes Genetic Consortium (T1DGC) is a worldwide study of the genetics of Type
1 diabetes which confirmed that specific combinations of the HLA Class II DR-DQ
alleles are the major genetic contributors to the risk of childhood Type 1 diabetes.
Some of the specific DR-DQ combinations confer increased susceptibility, or an
increased risk of Type 1 diabetes, whilst others confer protection from the disease [156].
In Caucasians, the highest genetic risk is conferred by the heterozygous genotype
DR3/DR4 [143, 152], whereas DR2 and DR6 are protective haplotypes [159, 160].
Children with high risk HLA genotypes have a higher risk for early and multiple
autoantibody development, and subsequent clinical Type 1 diabetes [161]. Therefore,
the known high risk HLA genotypes can be used as a useful predictor of future disease,
although only <10% of individuals with high risk HLA genotypes for Type 1 diabetes
actually go on to develop the disease [162, 163]. Also, it is important to note that the
DR3/DR4 high risk haplotypes identified in predominantly European populations are
absent in some populations, e.g. Japan, and high risk haplotypes may vary according to
ethnicity [164]. In addition, some haplotypes increase susceptibility in some
populations and confer protection in others, indicating that the genotypic context in
which the haplotype occurs may modify its effect on the risk of Type 1 diabetes.
31
The low penetrance of susceptibility genes for Type 1 diabetes is illustrated by the
finding that only 5-10% of individuals in the population who are heterozygous for the
HLA genotype D3/DR4, that is known to confer the highest risk, actually go on to
develop Type 1 diabetes [163, 165]. This shows that, despite genetic factors clearly
playing a role in the aetiology of Type 1 diabetes, they are not sufficient on their own to
cause the disease.
Other genetic associations which have been identified include genes within other parts
of the HLA region such as those encoding tumour necrosis factor [166], complement C4
and class I alleles, as well as genes outside the HLA region, such as the insulin gene
[167] and PTPN22 gene [168, 169]. Over 60 loci have been identified as contributing
to the susceptibility of developing Type 1 diabetes [170-172]. However, the odds ratio
or effect size of these loci were relatively low, ranging from 1.1 – 1.3 [157], compared
to between 0.02 and > 11 for HLA class II haplotypes [156]. In general, the genes that
have been identified as being associated with Type 1 diabetes are thought to modify the
risk of disease by the effect of their end-products on either immune function, insulin
expression or beta-cell function [173].
Although a large number of genetic loci have been associated with Type 1 diabetes risk,
HLA haplotypes remain the strongest predictor of disease susceptibility [174]. The
HLA haplotypes refer to a region of the genome that includes multiple genes that
encode class II and class I proteins which are involved in antigen presentation to T-
helper cells. Class II (A, B and C) and Class I (DP, DQ and DR) proteins are cell-
surface proteins which bind exogenous and endogenous peptides and present them to T
cells. Class II proteins are mainly expressed on antigen presentation cells such as
dendritic cells, B-lymphocytes and thymus epithelium. The Class II proteins on these
32
cells present antigens to CD4+ T cells which promote inflammation by secreting
cytokines. Proteins encoded by Class I alleles of the HLA region have been found to
influence the risk of Type 1 diabetes independently and to influence the age of disease
onset. The influence of Class I proteins on the age of disease onset may be mediated
via their presentation of antigens to CD8+ cytotoxic T-cells, which are involved in the
autoimmune destruction of pancreatic beta-cells [174].
The HLA region is highly polymorphic, with thousands of alleles existing for each of
the HLA genes . Polymorphism in these genes affects the shape of the peptide-binding
groove in the Class I and Class II they encode, thereby influencing the range of
peptides/antigens that can bind to these proteins and the specificity of the immune
response [174].
Genes outside of the HLA region that influence the risk of Type 1 diabetes include the
promoter region of the insulin gene which influences insulin expression levels [175],
and the PTPN22 gene which encodes an enzyme called tyrosine phosphatase that is
important in down-regulation of the immune response [168].
Although the odds ratios conferred by most of the genes identified by GWAS range
from 1.1 – 1.3, many of the identified candidate genes have immunological or metabolic
functions which supports their potential role in the aetiology of Type 1 diabetes [157,
176]. Rather than having an individual moderate effect, multiple variants of such genes
may combine together to create a phenotype with altered susceptibility to Type 1
diabetes, and this is currently being investigated using new technologies such as next-
generation DNA sequencing [174, 176].
33
Despite significant advances that have been made in the understanding of the role of
genetic factors in the aetiology of Type 1 diabetes, the exact mechanisms in which
genetic effects confer risk or protection from childhood Type 1 diabetes either directly,
or via interaction with other genetic or environmental factors, are still not clearly
understood.
The identification of genes that influence the risk of Type 1 diabetes is the first step in
trying to understand the molecular basis of Type 1 diabetes [172]. Further investigation
using a systems genetic approach, involves analysing genetic variation in candidate
genes with end-products of its expression and phenotypic traits of susceptibility to Type
1 diabetes [172]. In this way, one can try and understand the molecular basis or
biological pathway by which these genes contribute to the risk of Type 1 diabetes.
For example, a recent study analysed data from GWAS with gene expression and
protein-protein interactions to construct biological pathways to try and translate genetic
associations with Type 1 diabetes risk into possible disease mechanisms [177].
Longitudinal cohort studies, such as the Diabetes Autoimmunity In the Young (DAISY)
study are investigating the role of HLA and non-HLA genes to try and determine how
they influence the biological pathway that results in Type 1 diabetes. Findings from this
study so far have shown that the high risk HLA DR3/4 genotype influences both the
development of islet autoimmunity and progression to clinical Type 1 diabetes [178]. ,
but the role of non-HLA genes in the disease pathway have been inconsistent [178-180].
As described above, by far the strongest genetic predictor of Type 1 diabetes risk is
conferred by HLA genotypes. With this in mind, an interesting finding over the past
two decades is that there has been a decrease in the prevalence of high risk HLA
34
genotypes in children diagnosed with Type 1 diabetes in some populations [181-184],
indicating reduced penetrance of these genotypes over time. This provides strong
evidence that environmental pressures have increased over time, resulting in Type 1
diabetes occurring more frequently in individuals with low-risk HLA haplotypes, who
previously may not have developed the disease in childhood. The following section
provides further evidence for the role of environmental factors in the aetiology of Type
1 diabetes.
2.7 Environmental factors
Additional evidence for the role of environmental factors comes from the rapid rate of
increase in incidence of childhood Type 1 diabetes observed in many countries around
the world [15-18, 71]. The rate of increase is too rapid to be accounted for by changes
in the population gene pool, and provides strong evidence for the role of environmental
factors in the aetiology of Type 1 diabetes [31, 77, 81, 185].
Twin studies have shown that the concordance rate for Type 1 diabetes in monozygotic
twins, who are genetically identical, ranges from 30-65% depending on the length of
follow-up [138-140], indicating that environmental factors must play a role in the
aetiology of the disease. Similarly, migrant studies have shown that the incidence of
Type 1 diabetes in migrant populations converges with that of the Indigenous
population. For example, the incidence of childhood Type 1 diabetes in South Asians
living in England is higher than that found in their country of origin and more similar to
the incidence observed in Caucasians born in England [186], indicating that factors in
the environment must be having an influence on disease risk.
35
2.7.1 Geographical variation in incidence of chi ldhood Type 1 diabetes
There is a large geographical variation in the incidence of childhood Type 1 diabetes
[13], with a greater than 350-fold difference in the incidence observed between
countries worldwide [15, 17]. The incidence varies from ~0.1 per 100,000 in Asian
countries such as China [187] and Japan, to ~40 per 100,000 in Sardinia and >50 per
100,000 in Finland [15, 55].
The variation in incidence between countries could partly be explained by differences in
genetic susceptibility to Type 1 diabetes in their respective populations. For example,
an ecological analysis found a positive correlation between the incidence of Type 1
diabetes in different countries and the prevalence of high risk HLA genotypes in their
population [188]. However, some neighbouring countries, such as Finland and Russian
Karelia, have populations that are genetically similar but have very different incidence
rates of Type 1 diabetes [17, 52]. This suggests that the geographical variation in the
incidence of Type 1 is more likely to be due to differences in environmental factors in
the two populations.
As well as differences observed in the incidence of childhood Type 1 diabetes between
countries, within country regional variation in the incidence of Type 1 diabetes has also
been reported. In Australia, the mean incidence of childhood Type 1 diabetes between
2000 and 2006 was 28.9 per 100,000 in Tasmania compared to 10.2 per 100,000 in the
Northern Territories [7]. Elsewhere, the incidence of childhood Type 1 diabetes has
been found to be higher in urban compared to rural areas of mainland Italy [189],
Lithuania [190] and North-East England [191]. In contrast to this, a higher incidence of
childhood Type 1 diabetes has been observed in the less populated rural compared to
36
urban areas of Finland [192], Northern Ireland [193], Austria [194] and the United
Kingdom [195].
Factors which vary between different countries or different regions within a country that
may be related to Type 1 diabetes include ethnicity [141, 187], genetic factors [196],
occupational exposures, diet and lifestyle factors, population density, infections,
environmental pollution and toxins [197, 198], general living conditions and
accessibility to health services.
2.7.2 Socioeconomic status
Some studies have suggested that the association between Type 1 diabetes and
geographical areas may be due to differences in the socioeconomic status of the areas.
Type 1 diabetes has been shown to be associated with higher social class or
socioeconomic status [199-202], with just a few studies reporting the converse [203].
An ecological analysis in Europe found that a higher incidence of Type 1 diabetes was
associated with national markers of increased prosperity, such as higher gross domestic
product and lower infant mortality [204]. In keeping with this, following the socio-
political changes in Central Europe during the 1990s and the resulting improvements in
the socioeconomic status of these countries, the incidence of Type 1 diabetes increased
at a greater rate in Central Europe compared to the rest of Europe during this time [16,
56, 205, 206].
2.7.3 Hygiene hypothesis
A possible explanation for the association with higher socioeconomic status is the
“hygiene hypothesis” which was first postulated as an explanation for the rising
prevalence of allergic diseases in Westernised countries [207]. This hypothesis
37
proposes that improvements in hygiene, and the concomitant reduction in exposure to
infections in childhood, predisposes individuals to immune-mediated diseases later in
life [208-212]. Exposure to infections in early childhood and the perinatal period is
thought to be necessary for adequate stimulation and development of the immature
immune system [213]. Without exposure to such infections, the immune system may
not develop appropriately, resulting in the development of immune-mediated diseases
such as asthma, atopy or Type 1 diabetes [210, 211, 213]. Several studies using proxy
measures of exposure to infections such as birth order, sharing a room with siblings, day
care attendance and use of antimicrobials have provided some support for the hygiene
hypothesis in the aetiology of childhood Type 1 diabetes [214-216].
With this in mind, the introduction of vaccinations against infectious diseases in many
Western countries was considered as a potential risk factor for Type 1 diabetes
[217, 218]. However, following the results of several large population-based studies,
the use of vaccinations is no longer thought to be associated with an increased risk of
Type 1 diabetes [216, 219-222].
2.7.4 Seasonal variation
A seasonal variation in the incidence of childhood Type 1 diabetes has been
demonstrated in many populations around the world [223]. A higher number of
children are diagnosed with Type 1 diabetes in the cooler autumn and winter months of
the year in countries in both the northern and southern hemispheres
[16, 17, 71, 224, 225]. Interestingly, a seasonal variation has also been observed in the
appearance of islet autoantibodies [111] suggesting that environmental factors which
vary in a seasonal pattern are involved in the initiation of islet autoimmunity and not
just in precipitating clinical Type 1 diabetes.
38
The seasonal variation in the incidence of childhood Type 1 diabetes suggests that
environmental factors which vary seasonally, such as viral infections and exposure to
sunlight, may be involved in the aetiology or clinical onset of the disease.
2.7.5 Viral infections
Early evidence for the role of viruses in the aetiology of childhood Type 1 diabetes
came from the finding of viral antibodies in patients with Type 1 diabetes
[226-229]. In support of this, an increased risk of the Type 1 diabetes was observed in
children with congenital rubella infection [230], although this association has been
recently questioned [231]. Over the past few decades, several other viruses have been
implicated, including cytomegalovirus [232], mumps virus [233], rotavirus [234, 235]
and enteroviruses such as Coxsackie B virus [236-239].
Enteroviral infections have been associated with both the development of islet
autoimmunity and Type 1 diabetes in different study populations [234, 240-246]. An
Australian study detected enterovirus RNA in 30% of children at the time of diagnosis
with Type 1 diabetes, compared to 4% of controls [247]. A recent meta-analysis of 24
observational studies found that enteroviral infections were associated with an increased
risk of developing islet autoimmunity and Type 1 diabetes [248]. The prospective
Finnish birth cohort study (DIPP) found an association between enteroviral infections
and the development of islet autoantibodies, as well as some evidence for a temporal
relationship between infection and autoimmunity [239, 243]. However, in contrast to
this, several prospective cohort studies including the DAISY study in Denver [249], the
BABYDIET study in Germany [244] and the MIDIA study in Norway [250] have found
39
no association between evidence of enterovirus infection and the risk of islet
autoimmunity.
Therefore, evidence from various studies supports the role of viruses in the aetiology of
Type 1 diabetes and there are several mechanisms by which viruses may exert their
effect. Viral infections could initiate or accelerate islet autoimmunity via mechanisms
such as molecular mimicry or cross-reactivity between viral peptides and islet antigens
[251, 252], by direct cytolytic effects on the pancreatic beta-cells [253, 254] or via
proinflammatory mediators that are produced in response to having a viral infection
[255, 256]. In Europe, an inverse correlation between the incidence of Type 1 diabetes
and the prevalence of enterovirus infections in the background population has been
observed [257], leading to the suggestion that, as per the polio hypothesis [256], the
diabetogenic complications of enterovirus infections may be more common in
environments with a lower rate of infections. There is also some evidence to suggest
that the effect of viral infections on the risk of childhood Type 1 diabetes varies
according to HLA genotypes and that future studies need to take this into account [247,
258].
Prospective, longitudinal studies such as The Environmental Determinants of Diabetes
in the Young study (TEDDY)[112], will help to determine whether enteroviral
infections initiate islet autoimmunity, accelerate the disease process, precipitate the
clinical onset of Type 1 diabetes or act at multiple points in the disease pathway.
2.7.6 Gut microbiome
More recently, there has been an increased focus on the role that the human microbiome
may play in the aetiology of Type 1 diabetes. The word microbiome refers to the
40
genetic catalogue of microorganisms within a defined environment. Advances in gene
sequencing and molecular biology technology have enabled genetic evaluation of the
human microbiome and identification of commensal microorganisms which previously
could not be cultured by traditional methods.
There is increasing evidence that disruptions in the commensal gut microflora can lead
to immune dysfunction and autoimmunity in humans [259, 260]. A prospective study
of children at risk of Type 1 diabetes observed a reduced diversity and a higher
proportion of Bacteroides phyla in the intestinal flora of children who developed Type 1
diabetes. Children who did not develop Type 1 diabetes had a more stable intestinal
microbiome and a greater proportion of Firmicutes phyla [261].
How could changes in the intestinal microbiome be influencing the risk of developing
Type 1 diabetes? Numerous animal studies have demonstrated that the onset of
autoimmune diabetes in diabetes-prone mouse models can be prevented or delayed by
neonatal exposure to bacteria and their by-products [262]. Animal studies have also
shown that the microbiome is important for the development, maturation and
maintenance of the immune system. For example, studies in germ free (GF) mice have
shown these mice to have anatomical and functional deficiencies in the immune system
which can be altered by introduction of specific bacterial species or products [263].
The relationship between the intestinal microbiome and aetiology of Type 1 diabetes
requires further investigation. An ongoing prospective study in Finland is underway
which is investigating the effect of probiotic treatment on the appearance of
autoantibodies in children genetically at risk of Type 1 diabetes [264]. Other studies
have recently reported an interaction between the gut microflora and dietary risk factors
41
for Type 1 diabetes, adding further complexity to how these environmental factors
influence the risk of disease [265, 266].
Therefore, as previously discussed, multiple genetic and multiple environmental factors
are involved in the aetiology of childhood Type 1 diabetes. Together with viral
infections and the gut microbiome, the other main areas of research into environmental
factors associated with childhood Type 1 diabetes are Vitamin D deficiency, dietary
factors and growth.
2.7.7 Vitamin D deficiency
A north-south gradient in the incidence of childhood Type 1 diabetes across Europe [16,
49, 71], combined with the seasonal variation in the incidence of childhood Type 1
diabetes [224, 267, 268] led to the investigation of Vitamin D as a potential
environmental factor in the aetiology of the disease [269]. Support for a role of Vitamin
D came from the observation that children with rickets, which results from Vitamin D
deficiency, have a three-fold higher risk of developing Type 1 diabetes in the first year
of life [270]. Several studies have found a high prevalence of Vitamin D deficiency in
children diagnosed with Type 1 diabetes [271-274].
A multi-centre European case-control study found that early supplementation of
Vitamin D in infancy was associated with a reduced risk of childhood Type 1 diabetes
[275]. This finding was replicated in a Finnish birth cohort study [270], and a case-
control study in Norway found that the use of cod liver oil, which is high in Vitamin D,
during the first year of life reduced the risk of childhood Type 1 diabetes [276].
However, no association was found between Vitamin D intake or serum 25-
hydroxyvitamin D3 levels during childhood and the risk of islet autoimmunity or Type
42
1 diabetes in the DAISY prospective cohort study of children genetically at risk of the
disease [277].
Other studies have focussed on the relationship between maternal Vitamin D levels and
the risk of Type 1 diabetes in their offspring. The Finnish prospective birth cohort
study (DIPP) reported no association between maternal Vitamin D intake and the
development of islet autoimmunity in their genetically at risk offspring in their first year
of life [278]. However, a nested case-control study undertaken in Norway found an
increased risk of Type 1 diabetes in the offspring of mothers with lower serum levels of
Vitamin D during pregnancy [279].
It has been suggested that changes in Vitamin D levels over the past few decades may
have contributed to the changing incidence of childhood Type 1 diabetes [280].
Prospective data on Vitamin D levels in children over time, taking seasonal variation of
Vitamin D levels into account, would be needed to determine whether or not there has
been a decreasing trend in population levels over time. With the increased use of
ultraviolet B absorbing sunscreens and sun-protective clothing, increased sedentary
lifestyle and associated decrease in physical activity and time spent outdoors, it is
feasible that there has been a decrease in Vitamin D levels of populations in developed
nations. Evidence supporting a role for Vitamin D in the changing epidemiology
includes the increasing incidence of childhood Type 1 diabetes observed in Finland
following reductions in the recommended daily dose of Vitamin D [281].
If Vitamin D does have a role in the aetiology of childhood Type 1 diabetes, how could
it affect the pathogenesis of this disease? As well as its role in calcium and bone
metabolism, Vitamin D is now recognised to have important immune modulatory
43
functions. Receptors for the metabolically active form of Vitamin D have been
identified on beta-cells and Vitamin D deficiency has been shown to be detrimental to
beta-cell function [282]. Vitamin D receptors have also been identified on cells in the
immune system [282]. However, a recent meta-analysis concluded that there was no
association between Vitamin D receptor polymorphisms and Type 1 diabetes risk [283]
as had previously been reported. Other findings include an association between Type 1
diabetes risk and low levels of Vitamin D binding protein [284], and with defects in
genes encoding the Vitamin D 25-hydroxylase enzyme which is responsible for
converting Vitamin D to its metabolically active form [285, 286]. Therefore, Vitamin D
may be acting via various mechanisms which are not yet fully understood.
The difficulty with interpreting current data available on the association between
Vitamin D and Type 1 diabetes is the differences in study design and measures of
Vitamin D that have been used. In addition, very little Vitamin D in humans is obtained
from the diet as most of it is made in the skin from UVB radiation [287]. Therefore
studies looking at oral Vitamin D intake and those measuring serum levels of Vitamin D
may not be comparable. A large scale, prospective randomised placebo-controlled trial
is needed to answer the question of whether or not Vitamin D supplementation can be
used to prevent islet autoimmunity or childhood Type 1 diabetes [287].
2.7.8 Dietary factors
Dietary factors other than Vitamin D intake that have been implicated in the aetiology
of childhood Type 1 diabetes include short duration of breastfeeding and early
introduction of cow’s milk protein and cereals [288] [289, 290]. Although
breastfeeding has been found to reduce the risk of Type 1 diabetes in some studies
[291-293], a lack of association between breastfeeding and Type 1 diabetes [220, 294],
44
as well as islet autoimmunity [295], has also been reported. Early introduction of
gluten-containing cereals before 3 months of age has been found to increase the risk of
developing islet autoantibodies [296, 297], as has the early introduction of cow’s milk
protein [298] although again, the findings are inconsistent [289, 294]. Therefore,
several dietary intervention trials have been undertaken to try and clarify the role of
these dietary factors in the pathogenesis of Type 1 diabetes.
The Trial to Reduce IDDM in the genetically at risk (TRIGR) is an ongoing multi-
centre intervention trial to establish whether weaning to a highly hydrolysed casein-
based formula, rather than a standard cow’s milk formula in infancy reduces the risk of
islet autoimmunity and Type 1 diabetes in newborn infants who have first-degree
relatives with Type 1 diabetes [299, 300]. A reduced risk of islet autoimmunity in
infants followed to the age of 10 years who were fed the hydrolysed casein-based
formula compared to standard cow’s milk formula was found in the TRIGR pilot trial,
supporting a role for cow’s milk protein in the aetiology of Type 1 diabetes [301].
In Germany, the BABYDIET study was established to determine whether delaying the
introduction of dietary gluten in infants genetically at risk of Type 1 diabetes from 6
months of age to 12 months of age, reduced their risk of islet autoimmunity [302]. A
recent report from this study showed that delaying the introduction of gluten in these
infants from 6 to 12 months of age did not alter the risk of them developing islet
autoimmunity [303], and therefore was not an effective dietary intervention for
preventing Type 1 diabetes in these children.
The Nutritional Intervention to Prevent (NIP) Type 1 Diabetes study in Denver has been
established to investigate the use of omega-3-fatty acids to prevent islet autoimmunity
in infants genetically at risk of Type 1 diabetes [304]. Another area of ongoing research
45
relates emerging evidence that dietary advanced glycation end-products (AGEs) may
have a role in the aetiology of Type 1 diabetes. Recent animal and in vitro studies have
shown that AGEs directly influence beta-cell function and insulin secretion [305].
Dietary AGEs are present in foods that are processed under high temperature, stored for
long periods or contain food additives, all of which are likely to have increased in
prevalence over time. Dietary restriction of AGEs and AGE-lowering therapies have
been shown to reduce the incidence of autoimmune diabetes in animal models of the
disease [305]. In addition, genetic polymorphisms in the AGE receptor have been
associated with genetic susceptibility to Type 1 diabetes [306]. Further investigation
required to determine whether or not dietary AGEs may be a modifiable environmental
factor for the prevention or delay of Type 1 diabetes onset.
Dietary factors are clearly important in the aetiology of childhood Type 1 diabetes and
studies are currently ongoing with the aim of clarifying the role that nutritional factors
play in the disease pathways and pathogenesis of childhood Type 1 diabetes [307, 308].
2.7.9 Rapid growth, overweight and obesity
As well as the specific dietary factors identified above, general changes in diet and
lifestyle factors, which have resulted in a significant increase in the prevalence of
childhood overweight and obesity in the developed world, may also be contributing to
the increasing incidence of childhood Type 1 diabetes [309]. In Australia, the
prevalence of obesity among 7-15 year olds tripled between 1985 and 1995 [310], and
approximately 25% of 5-15 year olds are overweight [311].
Overweight and obesity are normally associated with an increased risk of Type 2
diabetes [141, 312, 313], however they may also be contributing to the increase in
46
childhood Type 1 diabetes [204, 314, 315]. Several ecological studies have related
changes in the incidence of childhood Type 1 diabetes to changes in the prevalence of
overweight and obesity. For example, the rapid non-linear increase in incidence
observed in Finland in recent years may be due to the increase in overweight and
obesity observed in this population since the mid 1990s [55, 316]. In Sweden, could a
decrease in the prevalence of childhood overweight and obesity explain the levelling off
in incidence of childhood Type 1 diabetes that has been observed in more recent years
[66]? In Austria, an association was found between body mass index of children
diagnosed with Type 1 diabetes and the year and age at which they were diagnosed
[317]. In other studies, rapid early growth and weight gain in childhood has been
associated with an increased risk of both childhood Type 1 diabetes [318-323]286-291),
and islet autoimmunity [324]. A recent systematic review and meta-analysis concluded
that both high birth weight and increased weight gain in early life are risk factors for
Type 1 diabetes in later life [325].
Perhaps, rapid early growth and weight gain are markers of dietary and lifestyle factors
that influence the risk of childhood Type 1 diabetes [326], or could they indicate
metabolic factors that increase the risk of the disease [315]? The “accelerator
hypothesis” proposes that weight gain may account for the increasing incidence of both
Type 1 and Type 2 diabetes [27, 327]. It argues that excessive weight gain and the
associated increase in insulin resistance result in hyperinsulinism and increased beta-cell
activity and, in individuals with a genetic susceptibility to autoimmunity, an immune-
mediated acceleration of beta-cell damage which results in Type 1 diabetes [27, 327].
Several studies have found that a higher body mass index is associated with a younger
age of onset of childhood Type 1 diabetes, suggesting that being overweight/obese and
the higher insulin resistance associated with this, may accelerate the disease progression
47
in children at risk of Type 1 diabetes [328-332]. Others have not found this association
[333, 334], and the premise of the accelerator hypothesis remains controversial [335-
338].
2.7.10 Perinatal factors
Factors in early life are clearly important in the aetiology of childhood Type 1 diabetes.
Prospective birth cohort studies have shown that early signs of islet autoimmunity, or
pre-diabetes, can be seen within the first few months of life [111]. Islet autoantibodies
are detectable in the first 2 years of life in the majority of children diagnosed with Type
1 diabetes under the age of 10 years. In addition, many populations have reported a
greater rate of increase in the incidence of childhood Type 1 diabetes in children under
the age of 5 years, suggesting that the disease is presenting earlier in life [18, 55, 70].
Therefore, environmental exposures early on in life, including the perinatal period, are
likely to be important in initiating the autoimmune process that results in clinical Type 1
diabetes in some children.
There have been many studies examining the relationship between the incidence of
Type 1 diabetes and perinatal factors such as infant birth weight, gestational age, birth
order and maternal age at delivery. Other than with maternal age at delivery, the
associations between perinatal factors and childhood Type 1 diabetes are not consistent.
Older maternal age at delivery has been associated with an increased risk of Type 1
diabetes in several studies [339-343], including a recent pooled analysis of 30
observational studies [344]. Higher birth weight has been associated with an increased
risk of childhood Type 1 diabetes in several populations [345-347], although some have
found no association between the risk of Type 1 diabetes and birth weight [340, 348].
One study found no difference in the birth weights of twins that were discordant for
48
Type 1 diabetes but who would have shared the same intrauterine environment [349].
However, 2 recent meta-analyses both concluded that a higher birth weight is associated
with an increased risk of childhood Type 1 diabetes [325, 350]. Similarly, preterm birth
has been associated with an increased risk of Type 1 diabetes in some studies [341,
346], but other studies have found no association between the risk of Type 1 diabetes
and gestational age [345, 351] and both a decreased [351] and increased [346, 352] risk
of Type 1 diabetes has been reported in first born children, as well as no association
with birth order [353]. A pooled analysis of 31 observational studies reported a
reduced risk of Type 1 diabetes with increasing birth order [354], supporting the earlier
findings of a higher risk of Type 1 diabetes in first born compared to later born children
[340, 346].
So, why have the findings on the association of perinatal factors and childhood Type 1
diabetes been inconsistent? Some of the earlier case-control studies [352] may not have
been representative of the whole population at risk, and studies with lower case
numbers may not have had the statistical power to detect the small effect size of some
perinatal factors [353, 355], resulting in inconsistent findings from different studies.
Large, population-based studies which have been undertaken in several populations and
the analysis of pooled data from such observational studies more recently, have helped
to clarify the associations [344, 354, 356]. In future, studies that take into account
genetic factors that are associated with both Type 1 diabetes and factors such as birth
weight [357-359] may help to further clarify the relationships. However, it is possible
that true differences in the effect of perinatal factors on the risk of childhood Type 1
diabetes may exist between populations with different genetic and environmental
characteristics.
49
No studies have been undertaken into the association of perinatal factors and childhood
Type 1 diabetes in Western Australia. Therefore, it was of interest to determine how
perinatal factors are associated with childhood Type 1 diabetes in this population, and
whether or not the findings were consistent with those reported in other predominantly
Caucasian populations.
2.8 Chapter summary
In summary, childhood Type 1 diabetes is a chronic medical condition requiring
lifelong insulin replacement therapy and is associated with significant morbidity and
mortality. The incidence of childhood Type 1 diabetes has been steadily increasing
worldwide and the cause for this continued increase is not yet known. Multiple genetic
and environmental factors are thought to be involved in the aetiology of Type 1 diabetes
but no causal factors have yet been identified. The increase in incidence is most likely
to be due to an increase in environmental exposures that initiate or accelerate the disease
process, or a reduction in exposures protective against the disease. The understanding
of the aetiology, natural history and epidemiology of this disease is an area of ongoing
research.
Western Australia presents a unique opportunity to study the epidemiology of childhood
Type 1 diabetes as complete and accurate population-based data are available over an
extended period. The findings from such research will be a valuable contribution to the
local, national and global data available on childhood Type 1 diabetes and may help to
further the understanding of this important disease.
The incidence of childhood Type 1 diabetes in Western Australia has only been
described from 1985 to 1992. Therefore the first aim of the research presented in this
50
thesis is to determine the incidence and incidence rate trends of childhood Type 1
diabetes in Western Australia from 1985 to the latest year with data available. No
previous studies have been undertaken in Western Australia to investigate the incidence
of childhood Type 1 diabetes for variation by region and socioeconomic status, or to
examine the association between childhood Type 1 diabetes and factors in the perinatal
period. Therefore, this research also aims to analyse the incidence of childhood Type 1
diabetes in Western Australia by place of residence and socioeconomic status. The third
aim of this research is to analyse the relationship between perinatal factors and the
incidence of childhood Type 1 diabetes.
The following chapter, Chapter 3 (Background chapter), describes the population of
Western Australia and provides details on the data sources that were used to achieve the
study aims described above.
51
Chapter 3: Background
52
3.1 Introduction
This chapter describes the context in which the research presented in this thesis was
undertaken. A brief description of the demographic features of Western Australia is
provided, including its population and climate. A more detailed description is then
given of the total population-based data resources available in Western Australia and the
data linkage processes that were used for the studies undertaken during this research.
3.2 Western Australia
3.2.1 Size
Australia is the sixth largest country in the world by total land area and is comprised of
6 States and several Territories [360]. Western Australia is the largest State in
Australia, making up the entire western third of the country (Figure 3.1). With a land
area of over 2.5 million square kilometres, Western Australia covers an area similar in
size to that of the whole of continental Europe.
3.2.2 Climate
As Western Australia encompasses such a large land mass, there is a wide variation in
the climate experienced across the State [361]. The northern parts of the State have a
tropical climate with heavy seasonal rains in the summer ‘wet’ season followed by a
cooler ‘dry’ season. Most of the interior parts of the State are classified as being arid or
semi-arid areas and experience intense summer heat with little rainfall. The climate is
moderated by coastal influences and there is a gradual decrease in rainfall and increase
in temperature ranges experienced as one moves further inland, away from the coast.
The climate in the capital city of Perth and the South Western part of the State where
the majority of the population live is of a Mediterranean type [361].
53
3.2.3 Population
Despite being the largest state of Australia, only ~10% of the total Australian population
lives in Western Australia. In June 2010, the population of Western Australia was
estimated as 2.29 million with males slightly outnumbering females [362]. Children
under the age of 15 years made up ~20% of the total population.
Figure 3.1: Population distribution of Western Australia showing urban, rural
and remote demographic zones as defined by the Western Australia Department of
Health. Each dot represents 100 persons, White background = Urban, Light grey
background = Rural, Dark grey background = Remote. (Produced by GIS Unit,
Department of Health, Western Australia in April 2012. Source: Australian Bureau of
Statistics 2010 Estimated Resident Population)
54
3.2.4 Ethnicity
In Western Australia, the majority of the population is made up of Caucasians
[363], with Indigenous Australians accounting for just 3% of the total population [364].
However, due to differing age distributions between the Indigenous and non-Indigenous
populations, Indigenous Australians account for 6% of the population of 0-14 year olds
in Western Australia [365].
3.2.5 Population distribution
The Western Australian Department of Health uses factors such as population density,
community infrastructure and service availability to define three main demographic
zones in Western Australia, called urban or metropolitan, rural and remote ( Figure 3.1).
Almost 75% of the total population lives in the urban area surrounding the capital city
of Perth and another 15% of the population lives in the South Western corner of the
State. The remaining rural and remote areas of the State are very sparsely populated
[362].
In contrast to the overall population who predominantly live in the urban areas of
Western Australia, the majority of Indigenous Australians live in the rural and remote
parts of the State and not in and around the capital city of Perth [364].
3.3 Data sources
3.3.1 Western Australian Children’s Diabetes Dat abase
Princess Margaret Hospital for Children is situated in the capital city of Perth. It is the
only tertiary paediatric hospital in Western Australia and the only referral centre for
children diagnosed with diabetes. In addition to providing services in the urban areas of
55
Western Australia, the diabetes team at Princess Margaret Hospital has been running
outpatient clinics in rural and remote areas of the State since 1989. Almost all children
diagnosed in Western Australia with Type 1 diabetes under the age of 15 years are
managed by the diabetes team at Princess Margaret Hospital.
The Western Australian Children’s Diabetes Database is a prospective population-based
register that was established at Princess Margaret Hospital in 1987. When it was
established, retrospective data on children diagnosed with Type 1 diabetes from 1985 to
1987 was obtained using multiple sources and also recorded in the database.
The Western Australian Children’s Diabetes Database meets the criteria for diabetes
incidence registers that were established by the Diabetes Epidemiology Research
International Group [13, 366]. These criteria were established to standardise diabetes
registers worldwide, and therefore enable comparisons between different populations.
To be included in the register, cases had to satisfy the following inclusion criteria [13,
366]:
• Diagnosed with Type 1 diabetes by a physician
• Treated with insulin
• Aged between 0 – 14 years at the time of diagnosis
• Resident in the defined area/population at the time of diagnosis
Using these criteria to identify eligible cases for the register, a minimum amount of data
was then to be collected on each case registered, including variables such as name, sex,
date of birth, date of diagnosis and place of residence at time of diagnosis.
56
Since 1987, every newly diagnosed case of Type 1 diabetes at Princess Margaret
Hospital has been registered on the Western Australian Children’s Diabetes Database
after obtaining written informed consent (Appendix A) from the child’s guardian. Data
stored include name, date of birth, date of diagnosis, age at diagnosis, sex, address at
diagnosis, results from autoantibody assays and family history. Using the capture-
recapture method [367], the case ascertainment level of this database has been estimated
to be over 99% complete [19]. The capture-recapture method is described in more
detail in section 4.4.1 of Chapter 4 (Methods). Additional data on these cases can be
obtained from the Western Australian Data Linkage System.
3.3.2 Western Australian Data Linkage System
The Western Australian Data Linkage System was established in Western Australia in
1995 [20] as a way of bringing together data from multiple sources that relate to the
same individual, family, event or place. The Western Australian Data Linkage System
currently consists of links between nine “core” State-wide health data collections
(Figure 3.2) [21] and multiple other “satellite” datasets.
The process used to identify records relating to the same individual, place or event from
different databases is called record linkage [368]. Record linkage involves the use of
computerised probabilistic matching algorithms with variables such as name, sex, date
of birth and address to identify the records of the same individual from multiple
databases. The algorithms are based on searching for identical variables, as well as
variables that are similarly spelt or that sound phonetically similar.
Following this computerised matching process, an extensive manual review is
undertaken of any doubtful links. For example, a manual review may be needed for
57
records with one day or one month’s difference in the date of birth. If a data entry
mistake is identified then the doubtful link would be considered correct. However, if
the records are of two separate individuals with similar names or addresses then the
doubtful link made would be considered incorrect. Another example of where the
record linkage process may need a manual review is where all variables for records
match apart from the family name. The doubtful link may be correct if the discrepancy
is due to the family name of a child having changed since its birth record due to a
divorce or remarriage of his/her parents later on in its life.
Figure 3.2: Data collections of the Western Australian Data Linkage System [21]
The record linkage methods used by the Western Australian Data Linkage Unit result in
very accurate linking of records from multiple data sources [20]. The estimated rate of
false positives, or incorrectly linked records, is <=0.1%. Similarly, the estimated rate of
false negatives, or records which are not linked but should be, is also ~0.1% (Personal
58
communication, Diana Rosman, Manager of Western Australian Data Linkage Unit,
November 2011).
The links in the Western Australian Data Linkage System are dynamic and are
continually updated with new or edited records. Therefore, the Western Australian Data
Linkage System can be used to readily identify up-to-date records stored in different
data collections that relate to the same individual or event. Linked data from these data
collections can be released for research purposes, after obtaining all the required ethics
approvals [369].
For the research studies presented in this thesis, linked data were released from the
Hospital Morbidity System, Midwives’ Notification System and Western Australia
Deaths Registry. These data collections are described below and the ways in which the
data were used is explained in Chapter 4 (Methods).
3.3.3 Hospital Morbidity Data System
The Hospital Morbidity Data System is a mandatory data collection maintained by the
Western Australia Department of Health that contains individual patient records for all
in-patient admissions to public and private sector hospitals in the State since 1970. Data
recorded include patient demographic information such as name, sex, age and address,
date of admission, date of separation and diagnosis and procedure codes relating to the
reason for admission to hospital. Data are obtained via a direct feed from computerised
patient administration systems at all public hospitals, and yearly audits are performed at
all private hospitals to ensure complete ascertainment. Therefore, the data are 99.99%
complete (Personal communication, Paul Stevens, Acting Manager, Hospital Morbidity
Data Collection, May 2012).
59
3.3.4 Midwives’ Notification System
The reporting of data on all births attended by midwives is a statutory requirement in
Western Australia (Health Act 1911, Section 335). Perinatal data are stored in the
Midwives’ Notification System and have been available in a computerised format since
1980. The data collection is maintained by the Western Australian Department
of Health and contains data on >99% of all births in Western Australia (Personal
communication, Alan Joyce, Manager Midwives’ Notification System, Department of
Health, August 2011).
Perinatal data are collected using the Notification of Case Attended form and recorded
for every hospital and community birth attended by a midwife (~ 30,000 births/year)
(Appendix B). The variables collected can be broadly categorised into those relating to
maternal and infant characteristics, pregnancy, labour and delivery (Appendix B).
3.3.5 Deaths registry
All deaths in Western Australia are required to be registered by the Registrar of Births,
Deaths and Marriages and since 1969 have been recorded in the Deaths registry.
Variables recorded include the date, place and cause of death.
Although the data stored in the Hospital Morbidity Data System, Midwives’
Notification System and Deaths Registry are collected for administrative purposes, they
have been found to be of a high quality. There are multiple checks made of the data
quality for all data collections and periodic audits are performed on random samples of
the data to ensure their accuracy [20].
60
Data from the Western Australian Data Linkage System were used for this research as
they are population-based and have the advantage of being prospectively collected and
therefore free of selection bias. By using linked data from these administrative data
collections it was possible to undertake relatively low cost research studies on the whole
population at risk. This is important for a rare and complex disease such as Type 1
diabetes.
3.4 Chapter Summary
This chapter describes the geography, population and climate of Western Australia to
illustrate the context in which the research undertaken for this thesis took place. In
addition, the data collections that were used to investigate the epidemiology of
childhood Type 1 diabetes in Western Australia and the process of record linkage are
described. Record linkage was used to link all records relating to each individual from
the multiple data collections used for this research.
The following chapter, Chapter 4 (Methods), provides an overview of the study design,
subjects and methods used during this research.
61
Chapter 4: Methods
62
4.1 Introduction
This thesis comprises several studies undertaken to address the study aims described in
Chapter 1 (Introduction). This chapter provides a general overview of the study design
and methods used in this research. To avoid repetition of content, a description of
common methods is provided in this chapter and will be referred to in the relevant
section of later chapters. The subjects, specific methods and statistical analyses used for
the individual studies are described in more detail in the methods section of Chapters 5,
6 and 7.
4.2 Ethical Issues
4.2.1 Ethics Approval
Ethics approval for all studies undertaken as part of the research presented in this thesis
was obtained from the Princess Margaret Hospital Ethics Committee and the Western
Australia Department of Health’s Confidentiality of Health Information Committee.
The Confidentiality of Health Information Committee has since been decommissioned
and its function taken over by the Western Australia Department of Health’s Human
Research Ethics Committee, which was established in April 2008.
4.2.2 Consent
Data on children diagnosed with childhood Type 1 diabetes in Western Australia were
obtained from the Western Australian Children’s Diabetes Database at Princess
Margaret Hospital in Perth. Clinical staff from the Diabetes and Endocrinology
department at Princess Margaret Hospital obtain written informed consent from patients
and/or their guardians prior to their data being registered on the Western Australian
Children's Diabetes Database. Copies of the information sheet provided to patients
63
and/or their guardians and the consent form used are provided in Appendix A.
4.2.3 Confidentiality
The confidentiality of health information collected primarily for administrative purposes
is of utmost importance. Where possible, only de-identified data were used. Identified-
data such as name, date of birth and address were obtained from the Western Australian
Children's Diabetes Database and provided to the Western Australia Data Linkage Unit
for record linkage purposes. Records for children diagnosed with Type 1 diabetes in
Western Australia were linked to the Hospital Morbidity Data System and the
Midwives’ Notification System, which are part of the Western Australia Data Linkage
System. The perinatal data obtained from the Midwives’ Notification System for all
remaining births in Western Australia were de-identified. No named or individual data
have been included in any of the reports emanating from this research.
4.3 Outcome of Interest - Type 1 diabetes
The outcome of interest for all of the studies presented in this thesis was the diagnosis
of Type 1 diabetes in children under the age of 15 years.
In Western Australia, children diagnosed with diabetes are usually admitted to hospital
at the time of diagnosis and treated by hospital-based paediatricians and physicians.
The diagnostic criteria used to determine whether or not a child has Type 1 diabetes are
in accordance with the position statements of the World Health Organization [1] and
American Diabetes Association [5, 30].
Some children have a rapid onset of Type 1 diabetes whilst others have a slow onset
over several months [30]. Symptoms that may be present at clinical onset of Type 1
64
diabetes include extreme thirst, frequent urination, recent weight loss, sudden vision
changes, lethargy, vomiting and occasionally drowsiness or coma. On investigation,
they are found to have very high blood glucose levels, together with high levels of
glucose and ketones in their urine. In addition, in many cases the presence of islet
autoantibodies in their blood provides evidence of autoimmune destruction of their
pancreatic beta-cells, which is thought to be the pathophysiological basis of Type 1
diabetes. One or more autoantibodies are present in 85-90% of individuals at the time
of diagnosis [5].
In Western Australia, the diagnosis of Type 1 diabetes is therefore based on a
combination of the clinical features described above, together with biochemical and
autoantibody assay findings. Autoantibodies that are tested for include islet cell
antibodies (ICA), antibodies to glutamic acid decarboxylase (GAD65) and antibodies to
tyrosine phosphotases (IA-2 and IA-2β). Antibodies have been assessed routinely on all
children with diagnosed with diabetes at Princess Margaret Hospital since 1985.
Antibodies to ZnT8 are not currently routinely tested for at Princess Margaret Hospital.
With the increase in obesity and prevalence of Type 2 diabetes in childhood [4],
assigning a type of diabetes to children at the time of diagnosis is becoming
increasingly complex, as some cases do not easily fit into one class [5, 26, 27, 370,
371]. As children diagnosed with diabetes in Western Australia are managed by the
same team at Princess Margaret Hospital until they are at least 16 years old, this
provides the opportunity to review and reconsider the type of diabetes after the initial
diagnosis is made and to update the Western Australian Children’s Diabetes Database
with any changes. Children diagnosed with Maturity Onset Diabetes of the Young
65
(MODY), Type 2 diabetes, or diabetes secondary to other causes, such as cystic fibrosis
or pancreatitis, were excluded from the studies undertaken in this research.
4.4 Study Cohorts
4.4.1 Incidence and trends of childhood Type 1 d iabetes in Western
Australia
To investigate the incidence and trends of childhood Type 1 diabetes in Western
Australia, cases were defined as all children newly diagnosed with Type 1 diabetes
under the age of 15 years, who were resident in Western Australia at the time of
diagnosis. At the start of this research, the study period used to investigate the
incidence and trends of Type 1 diabetes in Western Australia was from 1985 to 2002.
Due to an additional 8 years of data being available since the publication of these initial
findings, follow-up studies were undertaken using data for the study period of 1985 to
2010. The case definition and denominator population used for these follow-up studies
was the same as those used for the initial study, and are described below. The results
from the initial and follow-up studies using these study cohorts are presented in Chapter
5 and Chapter 6.
Case identification
Eligible cases of childhood Type 1 diabetes were identified using the Western
Australian Children's Diabetes Database at Princess Margaret Hospital. In order to
ensure complete case ascertainment, the Hospital Morbidity Data System was used as
an independent secondary source of cases. The Western Australian Children's Diabetes
Database and Hospital Morbidity Data System are described in more detail in the
Background chapter, Chapter 3.
66
A list of all patients whose diagnoses at discharge included any reference to Type 1
diabetes, and whose age at admission was less than 15 years, was obtained from the
Hospital Morbidity Data System. This was then cross-checked with the list of cases
obtained from the Western Australian Children’s Diabetes Database. As the reporting
of cases to these two databases is independent of each other, the capture-recapture
method was used to estimate completeness of ascertainment [367].
The capture-recapture method
The core to an epidemiological study is the identification and counting of cases of the
disease being studied. The aim is to count as close to 100% of the cases as possible so
that accurate rates of the disease can be calculated for the population. Using two
independent sources of cases, the capture-recapture method can be used to estimate the
total number of cases in a population using the formulae shown below [372] (Figure
4.1). If the degree of ascertainment is not found to be close to 100% then the rates of
disease can be corrected for this undercount so that they represent the true rate of
disease in the population more accurately. The capture-recapture method has been used
in numerous studies of the epidemiology of diabetes [15].
For this research, the Western Australian Children’s Diabetes Database at Princess
Margaret Hospital was the primary source of cases (Sample 1) and the Hospital
Morbidity Data System was used as an independent secondary source of cases (Sample
2). Case ascertainment for 1985 to 1992 had been previously estimated as 99.6%
complete [22] so the capture-recapture method described above was applied for cases
diagnosed between 1992 and 2002.
67
From 1992 to 2002, of the total of 801 cases, 786 were ascertained from both the
register at Princess Margaret Hospital and the Hospital Morbidity Data System, 10 cases
were ascertained only from the Hospital Morbidity Data System and 15 cases were
ascertained only from the register at Princess Margaret Hospital.
Figure 4.1: The capture-recapture method for estimating completeness of case
ascertainment [372].
Sample 1 Sample 2
M = Number in first sample
n = Number in second sample
m = Number common to both samples
N = Estimate of total number of cases
N = (M+1)(n+1) -1 (m+1)
Variance (N) = (M+1)(n+1)(M-m)(n-m) (m+1)(m+2)
95% CI = + 1.96 Variance (N)
The degree of case ascertainment from the sources used can then be calculated:
Degree of ascertainment (%) = M + n - m x 100 N
Using the capture- recapture method, the case ascertainment from 1992 to 2002 was
estimated to be 99.98%. There were no known differences in case ascertainment levels
n
M m
68
within this period. Estimates of case ascertainment for the follow-up study period of
2002 – 2010 are currently being undertaken, but will not be available until after
submission of this thesis. Based on previous estimates of case ascertainment being
>99% complete, and the fact that there have been no known changes in the referral
practices of children diagnosed with Type 1 in Western Australia, it is unlikely that
ascertainment levels have changed significantly during this time. Due to the high
estimated case ascertainment rate, no corrections were applied to the disease rates
calculated in this research.
Population data and calculation of incidence rates
To calculate the incidence of Type 1 diabetes in Western Australian, cases ascertained
from all sources were used as the numerator data, and population data published by the
Australian Bureau of Statistics were used as the denominator data. The Australian
Bureau of Statistics publishes annual estimates of the resident population in Western
Australia [362, 373, 374]. These data are based on the five yearly population census
counts, and take into account changes to population numbers that result from births,
deaths, and inward and outward migration.
The Epidemiology Branch of the Western Australian Department of Health provide a
Rates Calculator [365, 375], which is a tool that can be used to obtain population data
published by the Australian Bureau of Statistics. The Rates Calculator was used in this
research to obtain annual population counts for children aged 0-14 years in Western
Australia categorised by calendar year, age, sex, Indigenous status and geographical
area.
When calculating the incidence of childhood Type 1 diabetes, each individual in the
69
particular population being analysed contributed to the person time at risk. For
example, the sum of all children aged 0-14 years for each year in the study period made
up the total person years at risk for the overall incidence of Type 1 diabetes in 0-14 year
olds over the whole study period.
4.4.2 Perinatal factors and childhood Type 1 dia betes in Western
Australia
To investigate the association between perinatal factors and the incidence of childhood
Type 1 diabetes in children aged 0-14 years in Western Australia, a retrospective birth
cohort study design was used.
Case definition and identification
Eligible cases were defined as all children born in Western Australia between 1980 and
2002 who were newly diagnosed with Type 1 diabetes under the age of 15 years up to
the 31st December 2003. Cases were identified using the Western Australian Children’s
Diabetes Database, which is described in more detail in Chapter 3 (Background).
Population data
The birth cohort used to investigate the association between perinatal factors and the
incidence of childhood Type 1 diabetes in Western Australia was all live births in
Western Australia between 1980 and 2002. In Australia, the definition of a live birth is
≥20 weeks gestation or birth weight ≥400g [376, 377]. Data on live births in Western
Australia were obtained from the Midwives’ Notification System which is a database
containing perinatal data on all births in Western Australia from 1980 on. More
information on the Midwifes’ Notification System can be found in the Background
chapter, Chapter 3.
70
All live births in Western Australia between 1980 and 2002 contributed to person time
under observation from the date of birth until the date of diagnosis of Type 1 diabetes,
reaching the age of 15 years or the 31st December 2003 - whichever occurred earliest.
Deaths occurring in children under the age of 15 years, occurring during the study
period were censored at the date of death.
Calculation of individual person time at risk
CASES Date of birth ----------------x Date of diagnosis with T1DM NON-CASES Whichever is earlier of: Date of birth ---------------o Date of 15th birthday Date of birth ---------------o End of follow-up (31/12/2003) Date of birth ---------------o Date of death if before diagnosis/15th birthday/
study end date
Calculation of incidence rates
Incidence rates were calculated using all newly diagnosed cases of Type 1 diabetes as
the numerator data and total person-time under observation of all live births as the
denominator data. Incidence rates and their 95% confidence limits were estimated
assuming a Poisson distribution of cases.
Record linkage and perinatal data extraction
A list of all eligible cases identified from the Western Australian Children’s Diabetes
Database was provided to the Data Linkage Unit at the Western Australian Department
of Health. The Data Linkage Unit were able to identify the perinatal records of these
cases in the Midwives’ Notification System by using the record linkage techniques
described in the Background chapter, Chapter 3. In addition, record linkage was used to
identify the records of any children born in Western Australia between 1980 and 2002
71
who died in Western Australia between 1980 and 2003 before the age of 15 years. This
was to enable accurate estimation of individual time under observation.
Following record linkage, place of birth was obtained for both cases and non-cases from
their perinatal records extracted from the Midwives’ Notification System. As non-cases
were sampled from the Midwives’ Notification System, which is the statutory record of
all births in Western Australia, all non-cases were by definition born in Western
Australia. For the few cases which did not link to the Midwives’ Notification System,
additional steps were taken to obtain their place of birth using alternate data sources,
thereby verifying the accuracy of the record linkage process. If the record linkage
process was accurate, alternate data sources should confirm that these cases were not
born in Western Australia between 1980 and 2002.
Alternate data sources for identifying the place of birth for cases that did not link to the
Midwives’ Notification System included the TOPAS Hospital Information System and
the Western Australian Children’s Diabetes Database at Princess Margaret Hospital.
The TOPAS Hospital Information System contains information on all children admitted
to Princess Margaret Hospital, including their country and State of birth. The Western
Australian Children’s Diabetes Database contains data on country and State of birth, but
only for patients diagnosed after 1999. Cases with no data on country of birth in either
of these two sources were contacted directly to obtain this information firstly by letter,
and then by telephone. The outcomes of this validation process were as follows.
Validation of record linkage process
Using the Western Australian Children’s Diabetes, 1181 cases were identified who were
born in Western Australia between 1980 and 2002 and diagnosed with Type 1 diabetes
72
up to 31st December 2003 aged <15 years at the time of diagnosis. Of these 1181 cases,
926 (78%) cases linked to the Midwives’ Notification System. Using place of birth
information obtained from either the TOPAS Hospital Information System, the Western
Australian Children’s Diabetes Database or directly from patients, 221 of the 255 (87%)
cases that did not link to the Midwives’ Notification System were confirmed to have
been born outside of Western Australia (Figure 4.2). 14 of the 255 cases that did not
link to the Midwives’ Notification System were confirmed as being born in Western
Australia but no data was available for them in the Midwives’ Notification System.
Possible explanations given for this included the loss or misplacement of their
Notification of Case Attended data collection forms or being born in Western Australia
but the birth being registered in a different Australian State or Territory.
Figure 4.2: Place of birth for cases not linked to Midwives’ Notification System
Perinatal data were therefore obtained for 926 of the 940 (98.5%) cases known to be
born in Western Australia that were successfully linked to the Midwives’ Notification
1181 cases from Western
Australian Children’s diabetes database
926/1181 (78%) linked to
Midwives’ Notification System
255/1181 (22%) did not link to
Midwives’ Notification System
222/225 (87%) not born in
Western Australia
14/255 (6%) born in
Western Australia
19/255 (7%) place of birth
not known
73
System, and the remaining 558,271 live births occurring in Western Australia between
1980 and 2002.
Perinatal data validation
The perinatal data that are stored in the Midwives’ Notification System are statutorily
collected data. As the data are collected mainly for administrative rather than research
purposes, only small random samples are usually validated by the data custodians. In
addition, over time, there have been changes to the type of variables for which data have
been collected and the format in which the data are stored. Therefore, it was important
to check the data for completeness and validity prior to using them for research
purposes. The data validation steps taken are described below.
Missing values
Each variable obtained from the Midwives’ Notification System was analysed
individually to assess the proportion of missing values and whether or not the missing
values occurred in a non-random manner. The frequency of each variable by birth year
was analysed to identify those variables which may not have been collected for the
whole study period, and were therefore missing for certain birth years. Although the
data were collected prior to the knowledge of disease onset, the proportion of missing
values was compared in cases compared to non-cases, to check whether this was similar
in both groups. As the proportion of missing values for the variables analysed in this
study was very low, records with missing data were not included in analyses using those
variables.
74
Improbable values
The frequency distribution for each variable was analysed to identify records with
improbable values (e.g. maternal age > 100). For variables with values clearly likely to
be errors, the value was set to missing. More detailed information on the data validation
steps taken for variables used in this study which had missing or improbable values can
be found in Appendix C (Perinatal data validation).
4.5 Measuring Socioeconomic Status
Part of the aim in each study was to investigate the incidence of Type 1 diabetes and its
association with socioeconomic status at the time of diagnosis, or at the time of birth,
respectively. Socioeconomic status is a complex notion and can be measured in
numerous ways. No individual level measures of socioeconomic status (e.g. parental
income, parental education level) were available from any of the data collections
available in Western Australia. However, the Australian Bureau of Statistics publishes
the Socioeconomic Indexes for Area (SEIFA) [378] which can be used as a measure of
socioeconomic status at the area level.
4.5.1 Socioeconomic Indexes for Area (SEIFA)
The SEIFA scores provide a summary measure of the socioeconomic status for an area
in which an individual lives and are not a measure of socioeconomic status at the
individual level [378]. Each index summarises different socioeconomic aspects of the
geographic area. The indexes are based on the socioeconomic characteristics of people
living in a geographically defined area, such as their income, education, occupation,
economic resources and the percent of people living in an area that are of Indigenous
origin or non-English speaking backgrounds. The data are collected during the census
every five years. The census collection district is the smallest geographical area for
75
which SEIFA scores are available and represents the area usually covered by one census
officer – in the urban area this equates to about 250 households [378].
At the time when this research was undertaken, the Australian Bureau of Statistics had
released four sets of SEIFA scores using data from the censuses conducted in 1986,
1991, 1996 and 2001. One of the indexes available is the Index of Disadvantage, which
is a general index derived from attributes of an area such as low income, low
educational attainment, high unemployment and people with jobs in low skilled
occupations [378, 379]. To investigate the association between socioeconomic status
and childhood Type 1 diabetes in this research, cases were assigned a SEIFA value from
the Index of Disadvantage using the address at the time of diagnosis or time of birth.
The address was used to determine the census collection district in which the case was
living at the time of diagnosis or time of birth. Following this, the SEIFA score for this
census collection district from the index released closest in time to the year of diagnosis
or year of birth was assigned to the case. The SEIFA value for the census collection
district was used to minimise heterogeneity occurring within geographical areas. By
using the smallest geographical area available, the SEIFA score was more likely to
represent the area in which the case lived.
Population counts of 0-14 year olds for the whole study period were obtained for each
census collection district, together with its Index of Disadvantage value. Cases were
then categorised into 5 groups of socioeconomic status using SEIFA scores that divided
the total population of 0-14 year olds in Western Australia into quintiles of
disadvantage.
76
4.6 Statistical analysis
To avoid repetition of content, this section only describes the general statistical methods
used in this research. Statistical analyses specific to the individual studies are detailed
in the statistical methods section of each publication or study presented in Chapters 5, 6
and 7.
Poisson regression with aggregated data was used to analyse incidence rates and
incidence rate trends. Statistical adjustment of potential confounders was carried out
using multivariate Poisson regression models. Covariate selection for the multivariate
analyses was based on known biological inter-relationships and optimal model fit. A
two-tailed p value of less than 0.05 was considered statistically significant. Data
analysis was performed using the STATA statistical software [380].
4.7 Chapter Summary
This chapter describes methods common to the studies undertaken in this research to
investigate various aspects of the epidemiology of childhood Type 1 diabetes in
Western Australia. Detailed information is given in this chapter on how cases of
childhood Type 1 diabetes were identified and the application of the capture-recapture
method to estimate the completeness of case ascertainment. In addition, a detailed
description is given of the steps taken to confirm accurate record linkage between the
Western Australian Children’s Diabetes Database and the Midwives’ Notification
System and the accuracy and completeness of the perinatal data obtained. Finally, a
detailed description is provided on SEIFA, a measure of socioeconomic status at the
area level, and how it was used in this research.
77
Information on the subjects, methods and statistical analyses specifically used for
individual studies is provided in the relevant section of Chapters 5, 6 and 7. The
following chapter, Chapter 5, consists of a publication describing the epidemiology of
childhood Type 1 diabetes in Western Australia from 1985 to 2002 [23], and the results
from a similar follow-up study which used an extended study period of 1985 to 2010.
78
79
Chapter 5: Incidence and incidence rate trends of
childhood Type 1 diabetes in Western Australia
80
5.1 Preface
The incidence of childhood Type 1 diabetes has been increasing in many populations
over the past two decades, including in Australia [18, 61, 65]. Due to the unique
demography of Western Australia, all children diagnosed with childhood diabetes are
managed by the diabetes team at the State’s only tertiary paediatric hospital, Princess
Margaret Hospital. This enables accurate epidemiology studies to be undertaken as
complete population-based data are available over an extended period of time. The
incidence in Western Australia was last reported in 1992, when it was found to have
increased significantly compared to previous years [22]. Therefore, the first aim of this
research project was to determine the incidence of Type 1 diabetes in children aged 0-14
years in Western Australia, and to analyse the incidence and incidence rate trends by
calendar year, gender, age at diagnosis and for seasonal variation.
5.1.1 Hypotheses
The specific research hypotheses being tested were:
1. There is no trend in the incidence of childhood Type 1 diabetes in Western Australia
by calendar year
2. There is no difference in the incidence or incidence rate trends of childhood Type 1
diabetes in Western Australia by gender or age group at diagnosis
3. There is no seasonal variation in the incidence of childhood Type 1 diabetes in
Western Australia by month of diagnosis
This chapter describes the background, subjects, methods and results specifically
relating to the investigation of the research hypotheses listed above. More detailed
descriptions of the general population and climate of Western Australia, the data
81
sources used in the study and the capture-recapture method that was used to estimate the
completeness of case ascertainment are provided in the Background and Methods
chapters (Chapters 3 and 4).
5.1.2 Initial study period: 1985 - 2002
When this research was initially started, data were available on the number of newly
diagnosed cases of Type 1 diabetes in Western Australia from 1985 up to the end of
2002. The first half of this chapter describes this initial study which had a study period
of 1985 to 2002, inclusive. The findings from this study were published in 2004 as a
journal article titled "Continued increase in the incidence of childhood Type 1 diabetes
in a population-based Australian sample (1985 - 2002)" in the journal Diabetologia [23]
(Appendix D). The unedited journal article has been reproduced in this chapter in its
entirety.
As well as being published in the journal Diabetologia, the results were presented at
several local, national and international conferences (Appendix D). In addition, some of
the findings were published in an invited commentary in the Diabetes Management
Journal [381] (Appendix D).
5.1.3 Follow-up study period: 1985 - 2010
Due to a further 8 years of data being available since the completion of this initial study,
a follow-up study was undertaken to extend the study period from 2002 to the end of
2010. Of particular interest, was the question of whether or not the incidence of
childhood Type 1 diabetes had continued to increase in Western Australia since 2002,
and if so, in which sub-groups and at what rate had it been increasing?
82
Therefore, the aim of this follow-up study was to update the information available on
the incidence rate and incidence rate trends of childhood Type 1 diabetes in Western
Australia from the end of 2002 to the end of 2010. The research hypotheses being
tested in this follow-up study and the methods used to investigate them were the same
as those used for the initial study described above, the only difference being an
extended study period of 1985 to 2010, rather than 1985 to 2002.
The results from this follow-up study are described and discussed in the second half of
this chapter. A brief report of these findings has been accepted for publication in the
journal Diabetes Care, and is currently in press (Appendix D). The results have also
been recently presented at some local and national conferences (Appendix D), and to
various local research groups in order to help formulate the logical next steps for further
investigation.
5.2 Initial study: Continued increase in the inci dence of
childhood Type 1 diabetes in a population-based Aus tralian
sample (1985-2002)
[This section is an unedited version of the journal article published in Diabetologia [23].
A PDF copy of the original journal article is included in Appendix D]
5.2.1 Abstract
Aims/hypothesis
Our aim was to determine the incidence of Type 1 diabetes in children who were 0 to 14
years of age in Western Australia from 1985 to 2002, and to analyse the trends in
incidence rate over the same period.
83
Methods
Primary case ascertainment was from a prospective population-based diabetes register
that was established in 1987, and secondary case ascertainment was from the Western
Australia Hospital Morbidity Data System. Denominator data were obtained from the
Australian Bureau of Statistics. Poisson regression was used to analyse the incidence
rates by calendar year, sex and age at diagnosis.
Results
There was a total of 1144 cases (560 boys, 584 girls). Using the capture–recapture
method, case ascertainment was estimated to be 99.8% complete. The mean age
standardised incidence from 1985 to 2002 was 16.5 per 100,000 person years (95% CI:
14.7–18.2), ranging from 11.3 per 100000 in 1985 to 23.2 per 100,000 in 2002. The
incidence increased on average by 3.1% (95% CI 1.9%–4.2%) a year over the period
(p<0.001). No significant difference was found between boys and girls. A significant
increase in incidence was found in all age groups, with no disproportionate increase
found in the 0 to 4 year olds.
Conclusions/interpretation
The incidence of childhood-onset Type 1 diabetes in Western Australia has increased
significantly over the past 18 years and shows no signs of abating. In contrast to other
studies, a higher rate of increase was not found in the youngest children.
5.2.2 Introduction
The changing epidemiology of childhood-onset Type 1 diabetes has been reported in
many countries and remains unexplained. A significant increase in the incidence of
Type 1 diabetes over the past few decades has been reported worldwide
84
[16, 64, 71, 382]. More recently, several studies in Europe have found that the highest
rate of increase is occurring in children under the age of five years [67, 69, 70].
There is a greater than 350-fold variation in the incidence of childhood-onset Type 1
diabetes worldwide [15], ranging from 0.1 per 100,000 per year in China to 45 per
100,000 per year in Finland [383]. Epidemiological studies of Type 1 diabetes are
important in identifying possible aetiological factors which might explain the large
geographical variation and changing epidemiology of the disease.
Western Australia presents a unique opportunity to characterise every child diagnosed
with Type 1 diabetes under the age of 15 years, enabling the epidemiology of
childhood-onset Type 1 diabetes in this State to be accurately described.
Western Australia is the largest State in Australia with a land area of over 2.5 million
square kilometres and an estimated population of 1.9 million. Aboriginal people
account for 3.5% of the State’s total population [384] and the remainder is mainly
composed of people of European descent.
Princess Margaret Hospital for Children is the only tertiary paediatric hospital in the
State and the only referral centre for children diagnosed with diabetes. In Australia,
children with diabetes are usually treated by hospital based paediatricians and
endocrinologists. This enables case ascertainment levels of over 99% to be consistently
achieved and the complete population-based data set can be used to accurately describe
the characteristics of newly diagnosed patients.
85
The epidemiology of Type 1 diabetes in 0-14 year olds in Western Australia has
previously been described from 1985 to 1992, when it was found to have increased
significantly in both genders and across all age groups [22]. Therefore, the aims of this
study were to determine the incidence of Type 1 diabetes in Western Australia in 0-14
year old children from 1985 to 2002 and to analyse the incidence rate trends over the
same period.
5.2.3 Subjects and methods
The primary source of cases was the Western Australia Children’s Diabetes database,
which was established at Princess Margaret Hospital in 1987 [19]. This prospective
population-based register satisfies the criteria of the Diabetes Epidemiology
International Group [13], enabling an international comparison of incidence rates. The
database also contains retrospective data, obtained using multiple sources, on children
diagnosed with Type 1 diabetes from 1985 to 1987. All patients seen at Princess
Margaret Hospital with newly diagnosed Type 1 diabetes are registered on the database.
As Princess Margaret Hospital is the only referral centre for children diagnosed with
diabetes in Western Australia, the case ascertainment level of this database is over 99%
[19, 22].
Data collected includes name, date of birth, date of diagnosis, age at diagnosis, sex,
address at diagnosis, results from autoantibody assays and family history. Patients are
seen three-monthly at the diabetes clinic and the database is regularly updated with
results from investigations. Therefore, the records of any patients found to have been
incorrectly diagnosed with Type 1 diabetes, or to have diabetes secondary to other
causes, are changed to include this information.
86
Secondary case ascertainment was from the Western Australian Hospital Morbidity
Data System, which contains name-identified information on discharges from all public
and private sector, acute-care hospitals in the State. A list of all patients whose
diagnosis at discharge included any reference to Type 1 diabetes, and whose admission
age was less than 15 years was obtained, and then crosschecked with cases in the
diabetes database. The capture-recapture method was used to estimate completeness of
ascertainment [367].
All patients diagnosed with Type 1 diabetes by a physician, commenced on insulin
therapy before the age of 15 years and resident in Western Australia at the time of
diagnosis were included in the study. Patients with diabetes secondary to other causes
such as cystic fibrosis were excluded. Type 1 diabetes was diagnosed according to the
World Health Organisation’s definition [1] and the date of diagnosis was defined as the
date of commencement of insulin therapy.
Population estimates based on census data were obtained from the Australian Bureau of
Statistics.
Ethical approval for this study was obtained from the Princess Margaret Hospital Ethics
Committee and the Western Australia Department of Health’s Confidentiality of Health
Information Committee. Informed consent was obtained from all patients prior to their
data being stored on the diabetes database at Princess Margaret Hospital.
5.2.4 Statistical analysis
Incidence rates were calculated using cases from both sources as the numerator and
annual population estimates from the Australian Bureau of Statistics as the denominator
87
[373]. Age standardised rates were calculated using the direct method with the
developed world population, which has equal numbers in each subgroup defined by age
(0-4, 5-9 and 10-14 yrs) and gender, as the standard [57]. Confidence intervals were
estimated assuming a Poisson distribution of cases.
Poisson regression models were used to analyse the incidence rates by calendar year,
gender and age group at diagnosis (0-4, 5-9 and 10-14 years), and to estimate the
temporal trends. Models were fitted to test for differences in the incidence rate trends
between the sexes and between the three age groups.
To analyse the incidence for seasonal variation, sine and cosine functions in a logistic
regression analysis were used [385]. This method also allows for covariate adjustment.
Annual incidence rates were calculated from 1985 to 2002. Importantly, the rates
calculated from 1985 to 1992 were found to be almost identical to those previously
published [22]. However, for consistency, the most recently obtained data are presented
here and were used to analyse the incidence rate trends over the study period.
A p value of less than 0.05 was considered statistically significant. Data analysis was
performed using the STATA statistical software package [386].
5.2.5 Results
Ascertainment
From 1985 to 2002, there were a total of 1144 cases (560 boys, 584 girls). Case
ascertainment for 1985 to 1992 has been estimated as 99.6% [22]. From 1992 to 2002,
of the total of 801 cases, 786 were ascertained from both the register at Princess
88
Margaret Hospital and the Hospital Morbidity Data System, 10 cases were ascertained
only from the Hospital Morbidity Data System and 15 cases were ascertained only from
the register at Princess Margaret Hospital. Using the capture-recapture method, the case
ascertainment from 1992 to 2002 was estimated to be 99.98%. Therefore, the case
ascertainment for the whole study period was estimated to be 99.8% complete.
Overall incidence
The mean age standardised incidence rate from 1985 to 2002 was 16.5 per 100,000
person years (95% CI: 14.7 – 18.2). The incidence ranged from 11.3 per 100,000
person years in 1985, to 23.2 per 100,000 person years in 2002 (Figure 5.1). The
incidence has increased by an average of 3.1% (95% CI: 1.9% – 4.2%) a year since
1985 and this was highly significant (p < 0.001).
Gender
From 1985 to 2002, the mean age standardised incidence was 15.6 per 100,000 person
years (95% CI: 13.7 – 17.5) in boys and 17.3 per 100,000 person years (95% CI: 15.3
– 19.4) in girls. The incidence has increased significantly in both genders, with an
average annual increase of 3.9% (95% CI: 2.2% – 5.6%, p < 0.001) in boys and 2.3%
(95% CI: 0.6% – 3.9%, p = 0.005) in girls. There was no significant difference between
the age standardised incidence (Incidence rate ratio = 1.10 (95% CI: 0.98 – 1.24, p =
0.10)) or incidence rate trends in boys compared to girls (p = 0.18).
Age group
The mean incidence in 0-4 year olds, from 1985 to 2002, was 11.0 per 100,000 person
years (95% CI: 9.2 – 12.8). This was significantly lower than the mean incidence in 5-9
and 10-14 year olds (Table 5.1). There was no significant difference between the
89
overall incidence of 18.8 per 100,000 person years (95% CI: 16.3 – 21.3) in 5-9 year
olds and 19.6 per 100,000 person years (95% CI: 17.6 – 21.6) in 10-14 year olds. The
incidence has increased significantly in all age groups since 1985 (Figure 5.3). The
incidence increased by an average of 3.3% a year (95%CI: 0.9% - 5.9%) in 0-4 year
olds, 3.7% a year (95% CI: 1.8% - 5.7%) in 5-9 year olds and 2.2% a year (95% CI:
0.4% - 4.0%) in 10-14 year olds. There was no significant difference in the incidence
rate trends between the age groups, and a disproportionate rate of increase in the
youngest age group was not found.
Figure 5.1: The trend in the incidence of Type 1 diabetes in 0-14 year olds in
Western Australia from 1985 to 2002 ( Annual age standardised incidence,
error bars show 95% confidence intervals; Estimated trend)
Year
1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
Inci
denc
e pe
r 10
0,00
0 pe
rson
yea
rs
0
5
10
15
20
25
30
90
Table 5.1: Mean annual age- and sex- specific incidence rates per 100,000
population (95% CI) in Western Australia from 1985 to 2002
Gender and age group No of Cases Population at risk Mean annual incidence (95% CI)
Boys
0-4 yrs 125 1,155,560 10.7 (8.5 – 13.0)
5-9 yrs 196 1,181,568 16.4 (13.3 – 19.5)
10-14 yrs 239 1,195,858 19.8 (17.3 – 22.4)
0-14 yrs 560 3,532,986 15.6 (13.7 – 17.5)
Girls
0-4 yrs 124 1,094,048 11.3 (9.2 – 13.4)
5-9 yrs 241 1,117,536 21.4 (18.0 – 24.7)
10-14 yrs 219 1,129,789 19.3 (16.5 – 22.1)
0-14 yrs 584 3,341,373 17.3 (15.3 – 19.4)
All
0-4 yrs 249 2,249,608 11.0 (9.2 – 12.8)
5-9 yrs 437 2,299,104 18.8 (16.3 – 21.3)
10-14 yrs 458 2,325,647 19.6 (17.6 – 21.6)
0-14 yrs 1144 6,874,359 16.5 (14.7 – 18.2)
Data are mean annual incidence rates per 100,000 per year (95% CIs) unless otherwise stated. Age standardised rates are shown for the 0-14 year age group.
Seasonal variation
There was significant seasonal variation in the incidence of Type 1 diabetes. A seasonal
pattern with one maximum and one minimum level per year was highly significant
(p<0.001). A higher number of cases were diagnosed from April to September which
are the autumn and winter months in Australia (Figure 5.2). A non-significant pattern
was found when investigating two minimum and maximum levels per year (p = 0.398).
There was no significant effect of gender (p = 0.098) however there was a highly
91
significant effect of age (p < 0.001). When analysing the individual age groups, a
significant seasonal pattern was only found in the 0-4 year olds (p = 0.046).
Figure 5.2: Seasonal variation in incidence of Type 1 diabetes in 0-14 year olds in
Western Australia
0
5
10
15
20
25
Jan Feb Mar April May June July Aug Sep Oct Nov Dec
Month
Inci
denc
e ra
te (
per
100,
000)
5.2.6 Discussion
The incidence of childhood-onset Type 1 diabetes in Western Australia is increasing at
a rate comparable to that reported in another State of Australia, New South Wales (355),
and in many European countries (3.2% a year) [16, 71]. This increase in incidence
cannot be explained by increased ascertainment levels as the completeness of the
population-based diabetes database in Western Australia has been estimated to be over
99% since its establishment in 1987 and case ascertainment methods have remained
unchanged since that time. The overall incidence in Western Australia is similar to that
in New South Wales [387] and when compared to other countries, is in the high
incidence category [15].
There has been a significant increase in the incidence in both boys and girls. No
significant difference was found in the overall incidence or incidence rate trends in boys
92
Figure 5.3: Trends in the age-specific annual incidence of Type 1 diabetes in 0-14
year olds in Western Australia from 1985 to 2002. ---- Estimated trends; Observed
rates ( 0-4 years; 5-9 years; 10-14 years)
Inci
denc
e pe
r 10
0,00
0 pe
rson
yea
rs
0
5
10
15
20
25
30
Inci
denc
e pe
r 10
0,00
0 pe
rson
yea
rs
0
5
10
15
20
25
30
Year
1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
Inci
denc
e pe
r 10
0,00
0 pe
rson
yea
rs
0
5
10
15
20
25
30
93
compared to girls. There are inconsistent findings on the difference in incidence of
Type 1 diabetes by sex. In general, a male excess has been found in countries with a
high incidence and a female excess has been found in countries with a low incidence of
Type 1 diabetes [16, 57, 58]. However, a higher incidence in girls was recently reported
in New South Wales [387] and similar to our findings, some studies have found no
difference in incidence when comparing boys to girls [388, 389], although both these
studies had ascertainment levels of 85%.
Many studies report an increase in incidence with increasing age, with the highest
incidence occurring in the 10-14 year old age group [15]. In this study, the incidence of
Type 1 diabetes in the youngest age group was found to be significantly lower than that
in 5-9 and 10-14 year olds. However, there was no significant difference in the
incidence between 5-9 and 10-14 year olds.
In contrast to reports from Europe [71], Oxford [69] and Finland [70], in which higher
rates of increase in the youngest age group have been documented, a disproportionate
rate of increase in 0-4 year olds was not found in Western Australia. There was no
significant difference in the incidence rate trends between the age groups, which is
similar to the findings in other countries including Scotland [63, 390]. Although a trend
towards a higher rate of increase in the youngest age group was found in New South
Wales, no statistically significant difference was found in the trend between the age
groups.
Significant seasonal variation in incidence was found with a higher number of cases
being diagnosed in the autumn and winter months. This is consistent with the findings
from many studies that have found a lower incidence of Type 1 diabetes occurring in
94
the warmer months [16]. Studies examining seasonal variation by gender and age group
have not had consistent findings. Some studies have found significant seasonal variation
occurring only in boys [224] and varying with age, with some studies showing a
seasonal effect in all age groups [16] and others, only in 10-14 year olds [372].
Seasonal variation in incidence may reflect the role of environmental factors such as
increased exposure to infections, lower physical activity and dietary intake in the
aetiology of Type 1 diabetes.
The rapid increase in incidence of Type 1 diabetes around the world is likely to be the
result of increased exposure to environmental risk factors in genetically susceptible
individuals. The non-differential increase in incidence in both genders and all age
groups, found in Western Australia, suggests that the increased exposure to risk factors
affects all of these groups.
The 18 years of complete population-based data on incident cases of Type 1 diabetes in
Western Australia means that the results presented here are an accurate description of
the epidemiology of Type 1 diabetes in this State. Continued monitoring of the
incidence in Western Australia is important for contributing to the knowledge about the
global patterns of this disease and for generating hypotheses for potential aetiological
factors.
95
5.3 Follow-up study: Incidence of childhood Type 1 diabetes
in Western Australia (1985 - 2010)
5.3.1 Introduction
From 1985 to 2002, the incidence of Type 1 diabetes in children aged 0-14 years in
Western Australia was found to have increased by an average of 3.2% a year, and to
have increased in both boys and girls, and in all age groups [23]. A rising incidence of
childhood Type 1 diabetes has also been reported in other parts of Australia including
New South Wales [61] and Victoria [65]. More recently, an increase in the national
incidence of childhood Type 1 diabetes was reported between 2000 and 2004, but no
marked increase in cases reported between 2005 and 2008 [7].
Around the world, the incidence of childhood Type 1 diabetes continues to increase in
many populations [17, 18, 55], with some countries reporting the greatest rate of
increase in the youngest age group [18, 55, 71]. A greater rate of increase in the
incidence of Type 1 diabetes in under 5 year olds has also been reported in Victoria
[65], however no difference was found in the rate of increase between different age
groups in the last study done in Western Australia [23] or in New South Wales [61].
With an additional 8 years of data being available on cases of Type 1 diabetes
diagnosed in Western Australia since the published findings above, it was important to
update the information available on the incidence and incidence rate trends of childhood
Type 1 diabetes in Western Australia, and to analyse any changes that may have
occurred since 2002.
96
Consistent with the research hypotheses listed in section 5.1.1 above, the specific
research questions being investigated in this follow-up study were: Has the incidence of
childhood Type 1 diabetes continued to increase in Western Australia since 2002? If so,
has it increased in both boys and girls, and in all age groups? Or, has there been a
disproportionate rate of increase in the youngest age group as has been reported in other
populations?
5.3.2 Aims
Therefore, the aim of this study was to analyse the incidence and incidence rate trends
of Type 1 diabetes in children aged 0-14 years in Western Australia from 1985 to 2010.
5.3.3 Methods
The methods used in this follow-up study were the same as those used for the initial
study and are described in section 5.2.3 above. To avoid repetition of content, these
methods have not been described again here. The only difference between the studies
was the study period used, which was 1985 to 2010 in this study, rather than 1985 to
2002 in the initial study.
More detailed explanations of the methods used to estimate the completeness of case
ascertainment, the population data that was used for the denominator data and the
calculation of person time and incidence rates are provided in the Methods chapter,
Chapter 4.
5.3.4 Statistical Analysis
The same statistical methods used in the initial study were used to determine the
incidence and analyse the incidence rate trends in this follow-up study. These are
described in section 5.2.4 above. In addition, in this follow-up study the incidence rate
97
trends were analysed for non-linear variation over time. To analyse the incidence for
non-linear variation, sine and cosine functions were applied to Poisson regression
models for 3, 4, 5, 6, and 7 year cycles and the Akaike Information Criterion (AIC) used
to assess goodness of fit [391, 392]. The AIC is one way of measuring the goodness of
fit of a statistical model. It provides a relative measure of the information lost when a
given model is used to describe data, estimating the trade off between bias and variance
in model construction. The smaller the AIC the better the model fit.
5.3.5 Results
Additional cases
There were 729 cases of Type 1 diabetes diagnosed in children under the age of 15
years in Western Australia from 2003 and 2010, and 1144 cases diagnosed between
1985 and 2002. Therefore, for this follow-up study, data were available on a total of
1873 cases (923 girls, 950 boys) diagnosed with childhood Type 1 diabetes in Western
Australia between 1985 and 2010 (Table 5.2).
Overall incidence
From 1985 to 2010, the mean age standardised incidence rate was 18.1 per 100,000
person years (95% CI: 17.5 – 19.2) (Table 5.2). The incidence ranged from 11.3 per
100,000 person years in 1985, to 25.5 per 100,000 person years in 2003 (Table 5.2)
(Figure 5.6). Using Poisson regression models with sine and cosine functions, the
incidence rate trend was found to have a statistically significant cyclical pattern, with a
5-year cycle providing the best model fit (AIC 7.518) (Table 5.3). The overall
incidence rate increased by an average of 2.3% a year (95% CI: 1.6% - 2.9%) with a
sinusoidal 5-year cyclical variation of 14% (95%CI: 7% – 22%) (Figure 5.5).
98
Gender
From 1985 to 2010, the mean age standardised incidence was 17.7 per 100,000 person
years (95% CI: 16.9 – 19.3) in boys and 18.5 per 100,000 person years (95% CI: 17.4 –
19.8) in girls (Table 5.2). The incidence increased significantly in both genders from
1985 to 2010, with an average annual increase of 2.9% (95% CI: 2.0% – 3.8%) in boys
and 1.5% (95% CI: 0.6% – 2.4%) in girls (Figure 5.6). There was no significant
difference in the overall mean incidence or incidence rate trends between boys and girls.
Age group
Incidence rates were analysed by 5-year age groups. The mean incidence in 0-4 year
olds from 1985 to 2010 was 11.0 per 100,000 person years (95% CI: 10.3 – 12.6). This
was significantly lower than the mean incidence in 5-9 and 10-14 year olds (Table 5.2).
The incidence in 5-9 yr olds was 84% higher than that in 0-4 yr olds (IRR 1.84
(95%CI:1.62 - 2.08)) and 95% higher in 10-14 year olds compared to 0-4 yr olds (IRR
1.95 (95%CI: 1.72 - 2.20)). There was no significant difference between the mean
incidence in 10-14 compared to 5-9 year olds (IRR 1.06 (95%CI: 0.96 – 1.17)) (Table
5.2). From 1985 to 2010, the lowest rate of increase was seen in the youngest age
group (Table 5.2). The incidence increased by an average of 1.3% a year (95%CI: 1.0%
-2.6%) in 0-4 year olds, 2.4% a year (95% CI: 1.4% - 3.5%) in 5-9 year olds and 2.5% a
year (95% CI: 1.5% - 3.5s%) in 10-14 year olds (Table 5.2) (Figure 5.7).
When examined for non-linear variation, a sinusoidal 5-year cyclical variation was
observed in both boys and girls and in all age groups.
99
Table 5.2: Mean annual age- and sex- specific incidence rates per 100,000
population (95% CI) in Western Australia from 1985 to 2010
Gender and age group No of Cases Person years at risk Mean annual incidence
(95% CI)
Boys
0-4 yrs 200 1,715,837 11.6 (10.1 – 13.4)
5-9 yrs 335 1,748,584 18.9 (17.2 – 21.3)
10-14 yrs 415 1,793,899 22.8 (21.0 – 25.5)
0-14 yrs 950 5,258,320 17.7 (16.9 – 19.3)
Girls
0-4 yrs 182 1,624,775 11.2 (9.6 – 13.0)
5-9 yrs 379 1,649,796 22.8 (20.7 – 25.4)
10-14 yrs 362 1,688,677 21.3 (19.3 – 23.8)
0-14 yrs 923 4,963,248 18.5 (17.4 – 19.8)
All
0-4 yrs 382 3,340,612 11.0 (10.3 – 12.6)
5-9 yrs 714 3,398,380 21.1 (19.5 – 22.6)
10-14 yrs 777 3,482,576 25.5 (20.8 – 23.9)
0-14 yrs 1,873 10,221,568 18.1 (17.5 – 19.2)
Data are mean annual incidence rates per 100,000 per year (95% CI) unless otherwise stated. Age standardised rates are shown for the 0-14 year age group.
100
Figure 5.4: Annual age standardised incidence of Type 1 diabetes in 0-14 year olds
in Western Australia from 1985 to 2010
0
5
10
15
20
25
30
1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010
Diagnosis year
Inci
denc
e (p
er 1
00,0
00 p
erso
n ye
ars)
Figure 5.5: The trend in incidence of Type 1 diabetes in Western Australia from
1985 to 2010 ( Observed annual incidence; Estimated 5-yearly cyclical
trend; ------ Estimated linear trend)
0
5
10
15
20
25
30
1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010
Year
Inci
de
nce
ra
te
(pe
r 1
00
,00
0 p
ers
on
ye
ars
)
101
Table 5.3: Goodness of fit measurements for different cyclical models fitted to
Western Australian Type 1 diabetes incidence data (1985 – 2010)
Model Fit Incidence Trend Cyclical variation
Cyclical Model AIC IRR (95% CI)
Sinusoidal variation
(95%CI)
3 years 8.218 1.022(1.016 – 1.029)* 2% (5%-8%)
4 years 8.129 1.023(1.016 – 1.029)* 5% (1% - 12%)
5 years 7.518 1.023(1.016 – 1.029)* 14%(7% - 22%)*
6 years 8.193 1.022(1.016 – 1.028)* 3%(1% - 10%)
7 years 8.149 1.023(1.017 – 1.029)* 4%(1% - 10%)
* p<0.001
Figure 5.6: Annual age standardised incidence of Type 1 diabetes in Western
Australia by gender ( Girls, ---- Boys)
0
5
10
15
20
25
30
1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010
Year
Inci
denc
e ra
te(p
er 1
00,0
00 p
erso
n ye
ars)
102
Figure 5.7: Trends in the age-specific annual incidence of Type 1 diabetes in 0-14
year olds in Western Australia from 1985 to 2010. Estimated trends (----);
Observed rates ( 0-4 years; 5-9 years; 10-14 years)
0
5
10
15
20
25
30
35
40
Inci
denc
e pe
r 10
0,00
0 pe
rson
yea
rs
0
5
10
15
20
25
30
35
40
1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010
Year
Inci
denc
e pe
r 10
0,00
0 pe
rson
yea
rs
0
5
10
15
20
25
30
35
40
Inci
denc
e pe
r 10
0,00
0 pe
rson
yea
rs
103
5.3.6 Discussion
Between 1985 and 2010, the incidence of Type 1 diabetes in children aged 0-14 years in
Western Australia increased by an average of 2.2% a year. This is comparable to the
rate of increase found in other populations in Australia [61] and around the world [17,
18]. In the initial study, which analysed the incidence of childhood Type 1 diabetes in
Western Australia between 1985 and 2002, the incidence rate was found to have
increased by an average of 3.1% a year [23]. The lower average rate of increase in
incidence found in Western Australia between 1985 and 2010 of 2.2% a year, compared
to 3.1% a year between 1985 and 2002, may reflect a slowing of the incidence of
childhood Type 1 diabetes in Western Australia. However, it is probably more likely to
be explained by the cyclical variation in incidence that was observed in this follow-up
study, as 2010 appears to be a low incidence or “trough” year, whereas the incidence
rate peaked in 2003, just after the end of the last study.
Non-linear, rather than cyclical patterns in the incidence rate trends of childhood Type 1
diabetes have been observed in Finland where the incidence continues to increase
rapidly over time [55], and in Sweden where a levelling off or slowing down in the
incidence has been reported [66]. In other populations, epidemics, or peaks in incidence
have been observed. For example, over the past few decades, epidemics in the
incidence of childhood Type 1 diabetes have been observed in Poland [393], Sweden
[394], Eastern Europe [395], Italy [396] and China [397]. A cyclical pattern in the
incidence of childhood Type 1 diabetes has been observed in the Yorkshire region of
England [398] and more recently in a neighbouring area of North-East England [399].
104
In Western Australia, a significant 5-year cyclical pattern in incidence was observed
between 1985 and 2010. Of particular interest, almost identical peak and trough
incidence years were found in Western Australia as those reported in North-East
England (Peak years in Western Australia: 1992, 1998, 2003 (Figure 5.5); peak years in
North-East England: 1992, 1997, 2003 [399]. The cyclical variation in incidence of
Type 1 diabetes reported in North-East England was based on a total of 526 cases
ascertained from several independent sources. However, no estimate was provided for
the completeness of ascertainment [399] or for possible variation in ascertainment levels
over time and validation of the data sources or details of disease classification were not
provided. Bearing these limitations in mind, the authors have been contacted to seek
clarification of their data sources, ascertainment estimates and methods. However, if
the peaks in incidence observed in North-East England have been accurately determined
it is of interest that similar peaks and troughs have been observed in Western Australia
and North-East England. Being on opposite sides of the globe, the two populations
have very different demographic and climatic conditions, including temporally opposite
seasons and infectious disease cycles.
Consistent with findings in many European populations, a seasonal variation in the
incidence of childhood Type 1 diabetes in Western Australia has been reported, with a
higher number of patients diagnosed in the cooler winter and autumn months [23]. The
data reported in North East England are the annual incidence rates so no comparisons
could be made for differences in seasonal variation in incidence that may occur due to
the populations experiencing opposite seasons.
What environmental factors have a cyclical nature and could modify the risk of
childhood Type 1 diabetes to result in peaks and troughs in its incidence? Viral
105
infections, in particular enterovirus infections, are thought to have a role in the aetiology
of Type 1 diabetes [248, 256]. Evidence supporting a role of viral infections in Type 1
diabetes aetiology include an early study which linked the cyclical pattern in incidence
of childhood Type 1 diabetes with the cyclical variation in number of reported mumps
virus infections [233]. In addition, an epidemic in the incidence of childhood Type 1
diabetes was reported to occur two years after a measles epidemic in Philadelphia [400]
and an association between echovirus infections and the induction of islet autoimmunity
has been reported more recently in Cuba [401]. A recent study investigating temporal
clustering of childhood Type 1 diabetes provides further evidence supporting the role of
infections in initiating, accelerating the disease process or precipitating the onset of
clinical symptoms [402].
As well as infections, the climate [403] and cyclical weather patterns may play a role by
altering lifestyle or environmental factors which modify the risk of childhood Type 1
diabetes. With increasing evidence of epigenetic effects, environmental factors acting
before conception or during pregnancy may also be relevant in modifying the future risk
of childhood Type 1 diabetes in the developing child.
An alternate explanation for the cyclical pattern observed is that a trough in incidence
may follow peak incidence years due to an exhaustion of individuals at risk of the
disease.
Some of the difficulty in identifying possible contributory factors is that such factors
could be acting to either increase or decrease the risk of Type 1 diabetes and influencing
the risk of disease at various or different points in the disease pathway. That is, some
factors may initiate Type 1 diabetes de novo, whilst others may accelerate the
106
progression of disease resulting in clinical Type 1 diabetes in those with established
autoimmune pre-diabetes. In addition, there are likely to be gene-environment
interactions, whereby the influence of some environmental factors varies according to
an individual’s genetic susceptibility. Type 1 diabetes is a complex disease and is
becoming increasingly recognised as being a heterogeneous condition. It is highly
likely that there are multiple causal pathways that result in childhood Type 1 diabetes,
with different environmental and genetic risk factors playing a role in different
subgroups of the population at risk of the disease.
Perhaps only a subgroup of the population at risk is susceptible to factors that vary in a
cyclical pattern and therefore display a cyclical variation in incidence, with the
remaining population at risk accounting for an underlying steady base rate of increasing
incidence. For example, an Australian case-control study found that enteroviral RNA
was detected in fewer children with increased genetic susceptibility to Type 1 diabetes
compared to those without the high risk HLA haplotype, suggesting that there may be a
subset of children diagnosed with Type 1 diabetes who are at low genetic risk of the
disease in whom enteroviral infections may increase the risk of disease onset [247].
In this follow-up study, the cyclical variation in incidence was observed in both boys
and girls and in all age groups and no other stratifications were made. In addition, no
differences were observed in the mean incidence or incidence rate trends between boys
and girls. This is consistent with findings from the initial study done in Western
Australia [23] and an Australia-wide study [60], but in contrast to findings in New
South Wales, where a higher incidence was observed in girls and a higher rate of
increase in incidence was observed in boys [61]. In other populations, some studies
have found a higher incidence in boys [240, 368, 369], and others have found no gender
107
difference in incidence [55, 56]. The reasons for these inconsistent findings are unclear.
Potential reasons for a gender difference in childhood Type 1 diabetes maybe the
differences in insulin sensitivity between boys and girls, particularly during puberty,
and differential transmission of familial Type 1 diabetes by fathers and mothers with
Type 1 diabetes to their male and female offspring [59].
Another interesting finding in our population is that the lowest, not greatest, rate of
increase in the incidence of childhood Type 1 diabetes was found in children under the
age of 5 years. This is similar to the findings in the initial study done in Western
Australia between 1985 and 2002 [23] but in contrast to findings in several other
populations [18, 68-70, 206], including one in Australia [65]. Studies done in some
populations suggest that rather than an increase in the overall lifetime risk of Type 1
diabetes, the disease is manifesting earlier and therefore presenting more frequently in
the youngest age groups [72, 73, 76, 77, 404].
This is not the case in Western Australia, where the overall incidence of Type 1 diabetes
in 0-4 year olds has remained significantly lower than the older age groups since 1985,
and where the rate of increase in incidence in the youngest age group is significantly
lower than that in the other age groups. A possible explanation for this finding in
Western Australia is that factors conferring protection from or susceptibility to Type 1
diabetes are acting in utero or in early childhood, which are either different to those
found in other populations or different in their prevalence in this population.
Continued monitoring of the incidence of childhood Type 1 diabetes in Western
Australia is important to determine whether or not the cyclical pattern in incidence
remains and to help identify environmental factors which may be related to Type 1
108
diabetes that display a similar cyclical pattern.
5.3.7 Acknowledgements
The Poisson regression models used to test the incidence rate trends for cyclical
variation were performed by Professor Max K Bulsara (Chair in Biostatistics, Institute
of Health and Rehabilitation Research, University of Notre Dame). The interpretation
of the results from the analyses of cyclical variation in incidence rate trends was
undertaken by myself with advice given by Professor Bulsara and the results from these
analyses have been included in this thesis with his knowledge and consent.
5.4 Chapter Summary
This chapter consists of two main sections, with both sections describing the
epidemiology of childhood Type 1 diabetes in Western Australia. Findings from the
initial study using a study period of 1985 to 2002 were described in the first section, and
findings from the follow-up study which used a study period of 1985 to 2010 were
described in the second section.
The incidence of Type 1 diabetes in children aged 0-14 years in Western Australia was
shown to have increased since 1985. No differences were found between boys and
girls. In contrast to other populations, the lowest rate of increase in incidence was seen
in the youngest age group. Of particular interest, a 5-year cyclical variation in the
temporal incidence rate trend was found in Western Australia, with almost identical
peak and trough years to a recently published study from North East England [399].
This provides strong evidence for the role of unidentified environmental factors in the
aetiology of childhood Type 1 diabetes.
109
The following chapter (Chapter 6) describes the geographical variation in the incidence
of childhood Type 1 diabetes in Western Australia and provides further support for the
role of environmental factors in the aetiology of childhood Type 1 diabetes.
110
111
Chapter 6: Incidence of childhood Type 1 diabetes in
Western Australia by place of residence and
socioeconomic status
112
6.1 Preface
The previous chapter (Chapter 5) described how the incidence of childhood Type 1
diabetes in Western Australia has been increasing since 1985. This increase in
incidence has been observed in both boys and girls, and in children of all ages. What
could explain this increasing incidence? Type 1 diabetes is a complex disease and
multiple genetic and environmental factors are thought to be involved in its aetiology
[77, 83, 405]. As discussed in the Introduction and Literature Review chapters
(Chapters 1 and 2), the increasing incidence of childhood Type 1 diabetes is most likely
due to changes in as yet unidentified environmental factors, as the rate of increase in
incidence is too rapid to be accounted for solely by changes in the population gene pool.
In the late 1990s, paediatric endocrinologists at Princess Margaret Hospital noticed that,
based on their clinical observations from earlier years, the number of children being
diagnosed with Type 1 diabetes in rural areas of Western Australia seemed to be higher
than expected (Personal communication, Professor Tim Jones, Head of Department of
Diabetes & Endocrinology, Princess Margaret Hospital, 2002). If the rate of childhood
Type 1 diabetes is increasing at a greater rate in rural areas of Western Australia, this
may help generate hypotheses regarding environmental factors which may be changing
in these areas that could be involved in the aetiology of childhood Type 1 diabetes.
Regional variation in the incidence of childhood Type 1 diabetes has been reported to
occur within several countries [31, 53, 195, 406] but has not been previously examined
in Western Australia. In some countries, regional variation in the incidence of
childhood Type 1 diabetes has been explained by differences in the socioeconomic
status between the areas [31, 191]. In Australia, the Australian Bureau of Statistics
113
publishes Socioeconomic Indexes for Area (SEIFA). These indexes are constructed
using the 5-yearly national census data and provide a summary measure of the
socioeconomic status of individuals living in a defined geographical area [378]. A
more detailed description of SEIFA is provided in Chapter 4 (Methods, section 4.5.1).
Having described the overall incidence and incidence rate trends of childhood Type 1
diabetes in Western Australia, the next step in this research was to analyse the incidence
of childhood Type 1 diabetes in Western Australia by place of residence at the time of
diagnosis, and to determine whether or not the clinical impression of more children
being diagnosed with Type 1 diabetes in rural areas of Western Australia could be
confirmed.
Therefore, the second aim of this research project was to examine the effects of place of
residence and socioeconomic status on the incidence of Type 1 diabetes in children aged
0-14 years in Western Australia.
6.1.1 Hypotheses
The specific research hypotheses being tested were:
1. There is no difference in the incidence or incidence rate trends of childhood Type 1
diabetes in Western Australia by place of residence at the time of diagnosis.
2. There is no difference in the incidence or incidence rate trends of childhood Type 1
diabetes in Western Australia by socioeconomic status at the time of diagnosis.
6.1.2 Initial study period: 1985 - 2002
As per the previous chapter, Chapter 5, when this research was initially done, data were
114
available on cases of Type 1 diabetes diagnosed in Western Australia from 1985 up to
the end of 2002. With a further 8 years of data now being available, a follow-up study
was undertaken, extending the study period from the end of 2002 to the end of 2010.
The first half of this chapter describes the initial study which had a study period of 1985
to 2002. The findings from this study were published in 2006 as a journal article titled
"Independent effects of socioeconomic status and place of residence on the incidence of
childhood type 1 diabetes in Western Australia (1985 – 2002)" in the journal Pediatric
Diabetes [24] (Appendix E). The unedited journal article has been reproduced in this
chapter in its entirety. As well as being published in Pediatric Diabetes, the results from
this initial study were presented at two international conferences (Appendix E). In
addition, a presentation was given to local researchers to discuss methodological issues
relating to the use of SEIFA for research purposes that were learnt during this research
study (Appendix E).
6.1.3 Follow-up study period: 1985 - 2010
Due to the additional 8 years of data being available since the initial study described
above, a follow-up study was undertaken to extend the study period from 2002 to the
end of 2010. Consistent with the hypotheses listed in section 6.1.1 above and following
on from findings of this initial study, the specific research questions of interest were:
Has the incidence of childhood Type 1 diabetes in non-urban areas of Western Australia
continued to increase at a higher rate than that in the urban areas of the State since
2002? And is the mean incidence of childhood Type 1 diabetes in the non-urban areas
now significantly higher than that in the urban areas of Western Australia? The results
from this follow-up study are described and discussed in the second half of this chapter.
115
Results from this follow-up study have also been recently presented at a local and
national conference (Appendix E).
This chapter describes the background, subjects, methods and results specifically
relating to the investigation of regional variation in the incidence of childhood Type 1
diabetes in Western Australia. More detailed descriptions of the general population and
climate of Western Australia and the data sources used in these studies are provided in
Chapters 3 and 4 (Background and Methods). A more detailed description of SEIFA,
the measurement of socioeconomic status used in the initial study, is provided in
Chapter 4 (Methods, section 4.5.1).
116
6.2 Initial study: Independent effects of socioec onomic status
and place of residence on the incidence of childhoo d Type 1
diabetes in Western Australia (1985 – 2002)
[This section is an unedited version of the journal article published in Pediatric Diabetes
[24]. A PDF copy of the original journal article is included in Appendix E]
6.2.1 Abstract
Objective
To analyse the incidence of Type 1 diabetes in 0-14 year olds in Western Australia,
from 1985 to 2002, by region and socioeconomic status.
Methods
Primary case ascertainment was from the prospective population-based Western
Australian Diabetes Register and secondary case ascertainment was from the Western
Australian Hospital Morbidity Data System. The address at diagnosis was used to
categorise cases into urban, rural and remote areas and into 5 socioeconomic groups
using the Index of Relative Socioeconomic Disadvantage. Denominator data were
obtained from the Australian Bureau of Statistics. Poisson regression was used to
analyse the incidence rates by area and socioeconomic status.
Results
There were a total of 1143α cases (904 urban, 190 rural, 49 remote). Case ascertainment
was estimated to be 99.8% complete. The mean annual age-standardised incidence
α When this study was undertaken, there was some question regarding the classification of the Type of diabetes for 1 of the total 1144 cases included in the earlier study presented in section 5.1.2
117
from 1985 to 2002 was 18.1 per 100,000 person years in urban (95% CI: 16.3 - 19.9),
14.3 per 100,000 in rural (95% CI: 11.4 - 7.3) and 8.0 per 100,000 in remote areas (95%
CI: 5.8 - 10.3). The incidence was significantly higher in urban compared to rural (Rate
ratio 1.27, p=0.001) and remote (Rate ratio 2.28, p<0.001) areas. The incidence
increased with higher socioeconomic status. The incidence in the highest
socioeconomic group was 56% greater than the lowest socioeconomic group (Rate ratio
1.56, p<0.001). These differences in incidence by socioeconomic status and region
were independent of each other.
Conclusions
Higher socioeconomic status and residence in the urban area are independently
associated with an increased risk of childhood Type 1 diabetes in Western Australia.
6.2.2 Introduction
There is a wide geographical variation in the incidence of childhood Type 1 diabetes
both between countries worldwide [15, 50, 71] and within some countries
[187, 189, 192]. The findings on regional variation in the incidence within countries
have been inconsistent, with some studies reporting a higher incidence in urban areas
[189, 407], others reporting a higher incidence in rural areas [192] and some reporting
no regional variation [408, 409].
In Scotland, the urban/rural difference in incidence of childhood Type 1 diabetes was
largely explained by differences in socioeconomic status between the areas [410].
Several studies in the early 1980s reported a greater incidence of Type 1 diabetes in
individuals with higher socioeconomic status [199, 411]. In contrast, a higher incidence
of Type 1 diabetes was observed in the most deprived areas of Northern England [412].
118
Environmental factors are thought to have an important role in the aetiology of Type 1
diabetes in genetically susceptible individuals [83, 413] and health behaviours and risk
factors are known to vary according to socioeconomic status in Australia [414].
Western Australia is the largest State in Australia with a land area of over 2.5 million
square kilometres and an estimated total population of 1.9 million. The population is
mainly composed of people of European descent and Indigenous Australians account
for just 3.5% of the State’s total population [384]. Over 70% of the total population
resides in and around the State’s capital city, Perth (Figure 6.1). The remainder of the
State is very sparsely populated and a greater proportion of the Indigenous population
resides in the rural and remote areas.
Princess Margaret Hospital for Children is the only tertiary paediatric hospital in the
State and the only referral centre for children diagnosed with diabetes. In Australia,
children with diabetes are usually treated by hospital-based paediatricians and
endocrinologists. The diabetes team at Princess Margaret Hospital has been running
outpatient clinics in rural and remote areas of the State since 1989. This enables case
ascertainment levels of over 99% to be consistently achieved [23], and the complete
population-based data set can be used to accurately describe the characteristics of all
newly diagnosed patients in Western Australia.
The epidemiology of Type 1 diabetes in 0-14 year olds in Western Australia has
previously been described from 1985 to 2002, when it was found to have increased
significantly in both sexes and across all age groups [23]. Preliminary analysis
suggested a greater increase in the incidence in rural areas. Therefore, the aims of this
119
study were to analyse the incidence of Type 1 diabetes in Western Australia in 0-14
year old children from 1985 to 2002 by region and socioeconomic status.
Figure 6.1: Population distribution of Western Australia. Each dot represents
1,000 people. (Produced by Epidemiology Branch, HIC, Department of Health,
Western Australia in August 2004. Source: Australian Bureau of Statistics 2003
Estimated Resident Population)
6.2.3 Methods
The primary source of cases was the Western Australian Children’s Diabetes Database,
a prospective population-based register, which was established at Princess Margaret
Hospital in 1987 and has been described previously [19]. Princess Margaret Hospital is
the only referral centre for children diagnosed with diabetes in Western Australia and
the case ascertainment level of this database is over 99% [22, 23].
Secondary case ascertainment was from the Western Australian Hospital Morbidity
120
Data System, which contains name-identified information on discharges from all public
and private sector, acute-care hospitals in the State. As the reporting of cases to these
two sources is completely independent, the capture-recapture method was used to
estimate completeness of ascertainment [367].
All patients diagnosed with Type 1 diabetes by a paediatric physician, according to the
World Health Organisation’s definition [1], under the age of 15 years and resident in
Western Australia at the time of diagnosis were included in the study. The diagnosis of
Type 1 diabetes was based on clinical presentation and the presence of islet cell
autoantibodies and/or autoantibodies to glutamic acid decarboxylase (GAD65). The
continuation of the clinical care of these patients under the same team until the age of 16
years provides the opportunity to review and reconsider the type of diabetes. Patients
with diabetes secondary to other causes such as cystic fibrosis, maturity-onset diabetes
of the young (MODY) and Type 2 diabetes were excluded from the study. The date of
diagnosis was defined as the date of commencement of insulin therapy.
Analysis by place of residence
The residential postcode at time of diagnosis was used to categorise cases into urban,
rural or remote areas, as defined by the Western Australian Department of Health. The
area definitions are based on population density, as well as factors such as community
infrastructure and service availability. In 2001, the population density in urban areas of
Western Australia was 172 people/km2, compared to 0.43 people/km2 in rural areas and
0.07 people/km2 in remote areas [414].
As there is a very low incidence of Type 1 diabetes in children of Indigenous origin in
Western Australia [80], and the Indigenous population is unevenly distributed
121
throughout the State, incidence rates for each area were calculated using both the total
population and the non-Indigenous population as denominators. Annual population
estimates, based on census data and published by the Australian Bureau of Statistics
[373] were obtained from the Western Australian Department of Health by area, and
used as the total population denominator data.
Non-Indigenous population estimates were derived by subtracting Indigenous
population estimates, obtained from the Department of Health’s Epidemiology Branch,
from the annual population estimates published by the Australian Bureau of Statistics.
Indigenous population estimates were made by the Epidemiology Branch based on 2001
census counts, with annual adjustments for births and deaths (Personal communication,
JP Codde, Epidemiology Branch, Western Australia, Department of Health).
Analysis by socioeconomic status
The Index of Relative Socioeconomic Disadvantage was used to analyse the incidence
of childhood Type 1 diabetes in Western Australia by socioeconomic status.
This index, which is published 5-yearly by the Australian Bureau of Statistics, is based
on census data and provides a summary measure of the socioeconomic status of a
geographical area. It is constructed from variables such as the proportion of individuals
in an area who are on a low income, are of Indigenous descent, unemployed or in
relatively unskilled occupations [379].
The census collection district is the smallest geographical area for which the Index of
Relative Socioeconomic Disadvantage is available. It represents the area covered by an
individual census officer, which is approximately 250 dwellings in urban areas and
122
proportionately less in rural and remote areas, as the distance between dwellings
increases [379].
The address at time of diagnosis was used to map each case to a census collection
district and obtain the Index of Relative Socioeconomic Disadvantage score for the
census year closest to the year of diagnosis. Cases were then categorised into five
socioeconomic groups using index scores that divided the total population of 0-14 year
olds in Western Australia, into quintiles of disadvantage.
Census population counts from the 1991, 1996 and 2001 censuses were obtained from
the Australian Bureau of Statistics by sex, age and census collection district. Linear
regression using the census counts was used to interpolate the annual inter-census
population counts, which were used as the denominator data to calculate the incidence
by socioeconomic status.
Ethics approval for this study was obtained from the Princess Margaret Hospital Ethics
Committee and the Western Australian Department of Health’s Confidentiality of
Health Information Committee. Informed consent was obtained from all patients prior
to their data being stored on the diabetes database at Princess Margaret Hospital.
6.2.4 Statistical analyses
Incidence rates were calculated for each area and socioeconomic group, using cases
from both sources as the numerator data. Confidence intervals were estimated assuming
a Poisson distribution of cases. Poisson regression models were used to analyse the
differences in incidence rates and incidence rate trends, using calendar year, area and
socioeconomic group in the models. To examine potential interactions, the analysis of
123
socioeconomic status was stratified by area. Due to the small number of cases in rural
and remote areas, cases from these areas were combined for some analyses and referred
to as being non-urban.
A p value of less than 0.05 was considered statistically significant. Data analysis was
performed using the STATA statistical software package [386].
6.2.5 Results
Ascertainment
From 1985 to 2002, there were a total of 1143 casesβ. As described previously [23], the
case ascertainment for the study period was estimated to be 99.8% complete.
Place of residence
Of the total 1143 cases in Western Australia, 904 occurred in the urban area, 190
occurred in rural areas and 49 occurred in remote areas. There were highly significant
differences in the mean age-standardised incidence of Type 1 diabetes between all areas
(Table 6.1). The overall incidence in the urban area was 27% higher than in rural areas
(Incidence rate ratio (IRR) 1.27 (95% CI: 1.09 - 1.48), p = 0.001), and 128% higher
than in remote areas (IRR 2.28 (1.71 - 3.05), p < 0.001)). The overall incidence in rural
areas was 80% higher than in remote areas (IRR 1.80 (95% CI: 1.31 - 2.47), p < 0.001)).
After adjusting for sex and age group, the incidence in the urban area was 44% higher
than in non-urban areas (IRR 1.44 (95% CI: 1.25 - 1.66), p < 0.001).
β When this study was undertaken, there was some question regarding the classification of the Type of diabetes for 1 of the total 1144 cases included in the earlier study presented in section 5.1.2
124
From 1992 to 2002, less than 0.01% of cases in this study were of Indigenous ethnic
origin. In 2001, in the urban area, 3% of 0-14 year olds were of Indigenous origin,
compared to 7% in the rural areas and 26% in remote areas [375]. To examine the
impact of this uneven distribution, the mean incidence was calculated using just the
non-Indigenous population as the denominator, and there was still a highly significant
difference in the mean incidence between all areas (Table 6.1).
Table 6.1: Number of cases, total person years at risk, non-Indigenous person
years at risk and mean incidence rate per 100,000 person years (95% CI) in
Western Australia, from 1985 to 2002, by geographical area
Urban Rural Remote
No. of Cases 904 190 49
Total person years at risk 4,943,792 1,318,302 612,259
Mean incidence in total population 18.1 (16.3 - 19.9) 14.3 (11.4 - 17.3) 8.0 (5.8 - 10.3)
Non-Indigenous person years at risk 4,798,106 1,228,998 450,985
Mean incidence in non-Indigenous population
18.7 (16.8 - 20.6) 15.4 (12.2 - 18.7) 10.9 (7.8 - 14.0)
From 1985 to 2002, the incidence increased by an average of 2.5% a year (95% CI:
1.2% - 3.8%) in the urban area (p < 0.001) compared to 5.0% a year (95% CI: 2.4% -
7.6%) in the non-urban area (p < 0.001) (Figure 6.2). This difference in incidence rate
trends over the whole study period was not statistically significant (p = 0.08). However,
analysed just over the last ten years, the rate of increase in incidence in the non-urban
areas was significantly higher than that in the urban areas (p = 0.02).
125
Figure 6.2: Annual age-standardised incidence of Type 1 diabetes in 0-14 year olds
in urban ( ) and non-urban ( ) areas of Western Australia
Socioeconomic status
Socioeconomic index values were obtained for 99% (900/904) of cases in the urban area
and 94% (225/239) of cases in non-urban areas. Cases for which a socioeconomic
index value could not be obtained included those with an onset address that could not be
mapped to a census collection district, and those that occurred in a census collection
district for which the Australian Bureau of Statistics did not publish an index value, a
greater proportion of which are in remote areas [379]. There was a significant increase
in the incidence of Type 1 diabetes with higher socioeconomic status (Figure 6.3).
For each increase in socioeconomic group, the incidence increased by an average of
9.9% (95% CI: 5.4% - 14.5%, p < 0.001). The incidence in socioeconomic group 5, the
Year
1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
Inci
denc
e R
ate
per
100,
000
pers
on y
ears
0
5
10
15
20
25
30
126
least disadvantaged group, was 56% higher than in socioeconomic group 1, the most
disadvantaged group (IRR 1.56 (95% CI: 1.29 - 1.88), p < 0.001).
When the 904 cases in the urban area were analysed separately, the same trend was
observed. In the urban area, for each increase in socioeconomic group, the incidence
increased significantly by an average of 10.7% (95% CI: 5.7% - 15.9%, p < 0.001).
However, in the non-urban area, no significant trend was observed (IRR 0.90 (95% CI:
0.81 – 1.01), p = 0.06).
Figure 6.3: Mean incidence of type 1 diabetes in 0-14 year olds in Western
Australia from 1985 to 2002, by socioeconomic group (1 = most disadvantaged
group, 5 = least disadvantaged group). Error bars represent the upper 95% CI
limit.
Independent effects of place of residence and socioeconomic status
The effect of area on the incidence of Type 1 diabetes in Western Australia was
independent of differences in socioeconomic status. After adjusting for socioeconomic
status, the incidence in the urban area was 48% higher than in non-urban areas (IRR
1.48 (95% CI: 1.27 - 1.71), p < 0.001). In addition, the effect of socioeconomic status
0
5
10
15
20
25
1 2 3 4 5Socioeconomic group
Inci
den
ce p
er 1
00,0
00
127
on the incidence of childhood Type 1 diabetes in Western Australia was independent of
place of residence. After adjusting for area, for each increase in socioeconomic group,
the incidence increased by an average of 7.3% (95% CI: 2.9% - 11.9%, p = 0.001) and
the incidence in socioeconomic group 5, the least disadvantaged group, was 42% higher
than in socioeconomic group 1, the most disadvantaged group (IRR 1.42 (95% CI: 1.17
- 1.72), p < 0.001).
Due to the small number of cases in the non-urban area, adjustment for sex and age
group was not possible.
6.2.6 Discussion
This study reports significant independent effects of place of residence and
socioeconomic status on the incidence of childhood Type 1 diabetes in the complete,
predominantly Caucasian, population of Western Australia.
The incidence in urban areas of Western Australia is significantly higher than that in
rural and remote areas. This is in keeping with findings in Italy [189] and Lithuania,
where the urban/rural difference was found to be most marked in 0-9 year olds [190].
In contrast, another Australian study in New South Wales from 1990 to 1991, reported
no significant geographical variation in the incidence of childhood Type 1 diabetes
[372]. The difference in findings between our study and that in New South Wales may
be due to the use of different definitions of urban/rural, or due to differences in the
urban/rural areas of the two States, which differ in their population distribution,
topography and climate.
In Western Australia, the incidence was highest in the urban areas, which have the
128
highest population density, and lowest in the sparsely populated remote areas.
Conversely, a higher incidence of childhood Type 1 diabetes in rural compared to urban
areas was reported in Finland, where an inverse relationship between the incidence and
population density was found [192].
Findings on the relationship between the incidence of childhood Type 1 diabetes and
population density are inconsistent, with a higher incidence found in the more densely
populated regions of Norway [407], but a lower incidence found in areas with higher
population density in Northern Ireland [193] and Yorkshire, England [191]. The
relationship between the incidence of childhood Type 1 diabetes and population density
may be confounded by socioeconomic status. In Scotland, for example, a lower
incidence in urban areas was explained by increased deprivation in these areas [410].
In Western Australia, the incidence of childhood Type 1 diabetes increased significantly
with increasing socioeconomic status, independent of place of residence. A higher risk
of childhood Type 1 diabetes in the least deprived, or most wealthy, social groups has
been reported by other studies [199, 411]. An ecological analysis of the incidence of
childhood Type 1 diabetes in Europe found that the incidence was positively correlated
with national indicators of prosperity such as gross domestic product and low infant
mortality [204]. Compared to other European countries, the incidence of childhood
Type 1 diabetes increased more rapidly in the former socialist countries in Central
Europe, possibly related to an increase in wealth [16]. No association between the
incidence of childhood Type 1 diabetes and social class was found in a study in South
West England [415] and Pittsburgh [200].
129
Comparison of the findings from different studies on the relationship between the
incidence of Type 1 diabetes and socioeconomic status is difficult, due to the use of
different measures of socioeconomic status that may reflect different underlying
environmental risk factors. The measure of socioeconomic status used in this study
(Index of Relative Socioeconomic Disadvantage) is a summary measure of the
socioeconomic status of a geographical area, based on the characteristics of individuals
living in that area, rather than a measure of socioeconomic status at the individual level
[379]. The index was assigned to individuals at the census collection district level,
which has been shown to provide a more accurate measure of socioeconomic status than
if assigned at the larger, postcode level [416]. Heterogeneity in the socioeconomic
status of individuals living in the same geographical area would, if anything, result in an
underestimation of the true relationship between socioeconomic status and the incidence
of Type 1 diabetes.
The difference in incidence between urban, rural and remote areas of Western Australia
is not due to differing ascertainment levels, as case ascertainment levels have been
comparable in all areas of Western Australia since the mid-1980s, and there are no
differences in accessibility to services provided by the diabetes department at Princess
Margaret Hospital.
In keeping with previous findings in Western Australia [80], this study reports a very
low incidence of Type 1 diabetes in children of Indigenous descent. This study
demonstrates that the regional variation in the incidence of childhood Type 1 diabetes in
Western Australia is not explained by the differing proportion of Indigenous
Australians, who have a reduced risk of the disease, in the population of these areas. In
contrast to the low incidence of Type 1 diabetes in Indigenous Australians, there is an
130
increased risk of childhood Type 2 diabetes in this ethnic group [4]. This may be
indicative of differences in genetic predisposition or environmental exposures between
Indigenous and non-Indigenous Australians.
The differences in incidence of childhood Type 1 diabetes in Western Australia found
by place of residence and socioeconomic status are likely to be the result of differences
in environmental factors. Health related behaviours have been shown to vary both by
socioeconomic group and by area in Australia [417]. Potential risk factors for Type 1
diabetes that may vary by socioeconomic status and place of residence include diet and
lifestyle, breast feeding practices, increased maternal age, as well as factors in support
of the hygiene hypothesis [214] such as smaller family size, differing child care
practices and differences in intestinal microflora [418].
There has been a significant increase in the incidence of Type 1 diabetes in both urban
and non-urban areas of Western Australia since 1985. Interestingly, there was some
evidence of a greater rate of increase in the incidence of Type 1 diabetes in non-urban
compared to urban areas, although this only reached statistical significance when
analysed over the last ten years. The greater rate of increase in the incidence of Type 1
diabetes in non-urban areas may reflect changes in the lifestyle or environment of these
areas, as urban lifestyle factors become more prevalent or accessible. Further studies
are required to investigate regional differences in potentially important environmental
exposures during childhood, including antenatal and perinatal risk factors.
Continued monitoring of the incidence of Type 1 diabetes in Western Australia is
important for evaluating variations in the incidence and incidence rate trends in this
complete population-based data set, and investigating potential aetiological factors.
131
6.3 Follow-up study: Regional variation in incid ence of
childhood Type 1 diabetes in Western Australia (198 5 – 2010)
6.3.1 Introduction
In the initial study described above (section 6.2), the incidence of childhood Type 1
diabetes was found to have increased in both urban and non-urban areas of Western
Australia [24] between 1985 and 2002. The results also suggested that in the last
decade of this time period, the rate of increase in non-urban areas was higher than that
in the urban areas [24]. The follow-up study described below was undertaken in order
to determine whether or not the incidence of childhood Type 1 diabetes had continued
to increase more rapidly in the non-urban areas of Western Australia between 2002 and
2010.
As the initial study found independent effects of place of residence and socioeconomic
status of the area on the incidence of childhood Type 1 diabetes [24], the analysis of
socioeconomic status was not repeated in this follow-up study. This was to avoid the
lengthy and time-consuming task of mapping cases and population counts to obtain
their appropriate SEIFA value, when the main questions of interest in the follow-up
study related to the place of residence at the time of diagnosis, and not to
socioeconomic status.
6.3.2 Aims
The aim of this follow-up study was to analyse the incidence and incidence rate trends
of Type 1 diabetes in children aged 0-14 years in Western Australia from 1985 to 2010,
by place of residence at the time of diagnosis.
132
6.3.3 Methods
The methods used to analyse the incidence of childhood Type 1 diabetes in Western
Australia by place of residence at the time of diagnosis were the same as those
described in section 6.2.3 above, the only difference being that the study period used
was 1985 to 2010, rather than 1985 to 2002. To avoid repetition of content, these
methods have not been described again here.
6.3.4 Results
Additional cases
There were 729 cases of children diagnosed with Type 1 diabetes in Western Australia
under the age of 15 years from 2003 and 2010 and 1144χ cases diagnosed between
1985 and 2002. Therefore, there were a total of 1873 cases diagnosed with Type 1
diabetes in Western Australia between 1985 and 2010. Of these 1873 cases, 5 cases had
insufficient data on the place of residence at the time of diagnosis so could not be
included in this study. Of the remaining 1868 cases that were included in the study,
1462 (78%) occurred in the urban area, 320 (17%) occurred in rural areas and 86 (5%)
occurred in remote areas (Table 6.2).
Mean incidence by area
There were significant differences in the mean age-standardised incidence of Type 1
diabetes between the areas (Table 6.2). The overall incidence in the urban area was 12%
higher than in rural areas (IRR 1.12 (95% CI: 0.99 - 1.26), p=0.07) although this was
not statistically significant, and 107% higher than that in remote areas (IRR 2.07 (95%
CI: 1.67 – 2.57)). The overall incidence in rural areas was 85% higher than that in
χ The 1 case that was excluded from the initial study of regional variation presented in section 6.2 due to an unclear diagnosis, was later confirmed as having Type 1 diabetes and was therefore included in this follow-up study
133
remote areas (IRR 1.85 (95% CI: 1.46 - 2.35)). After adjusting for sex and age group,
the incidence in the urban area was 32% higher (IRR 1.32 (95% CI: 1.28 - 1.47)) than in
the non-urban areas (rural and remote areas combined).
From 1985 to 2010, less than 0.01% of cases diagnosed with Type 1 diabetes in
Western Australia under the age of 15 years were of Indigenous ethnic origin. Over the
same time period, an estimated 3% of the total population of 0-14 year olds in the urban
area were of Indigenous origin, compared to 7% in the rural areas and 27% in remote
areas.
Similar to the initial study, the mean incidence of childhood Type 1 diabetes was
calculated for each area using just the non-Indigenous, rather than total, population as
the denominator to take the uneven distribution of children of Indigenous origin in the
population into account. When using the non-Indigenous population as the denominator
data, no significant difference was found in the mean incidence between urban and rural
areas, and the mean incidence in both urban and rural areas was still significantly higher
than that in remote areas (Table 6.2).
Incidence rate trends by area
Between 1985 and 2010, the incidence increased by an average of 1.7% a year (95% CI:
1.0% - 2.4%) in the urban areas of Western Australia compared to 3.8% a year (95% CI:
2.4% - 5.1%) in the non-urban areas. The temporal incidence rate trends in urban and
non-urban areas of Western Australia had a similar pattern in both areas (Figure 6.4).
However, there was a statistically significant difference in the incidence rate trends
between urban and non-urban areas (p=0.01), confirming that the rate of increase in
134
incidence in the non-urban areas has been significantly higher than that in the urban
areas over this time period.
Table 6.2: Number of cases, total person years at risk, non-Indigenous person
years at risk, mean incidence rate per 100,000 person years (95% CI) in Western
Australia, from 1985 to 2010, by geographical area
Urban Rural Remote
No. of Cases 1462 320 86
Total person years at risk 7,427,327 1,839,777 943,731
Mean incidence in total population (95% CI)
19.5 (17.9 - 21.1) 17.3 (14.7 - 19.9) 9.2 (7.0 - 11.3)
Non-Indigenous person years at risk 7,207,248 1,715,165 692,022
Mean incidence in non-Indigenous population (95% CI) 20.1 (18.4 - 21.8) 18.7 (15.8 – 21.5) 12.7 (9.6 - 15.8)
Figure 6.4: Observed annual age-standardised incidence of Type 1 diabetes in 0-14
year olds in urban ( ) and non-urban ( ) areas of Western Australia from 1985
to 2010 with estimated incidence rate trends (---)
135
6.3.5 Discussion
This follow-up study provides further evidence for regional differences in the incidence
and incidence rate trends of childhood Type 1 diabetes in Western Australia. In the
initial study, which examined the incidence of childhood Type 1 diabetes in Western
Australia for regional variation between 1985 and 2002, the mean incidence in urban
areas was found to be significantly higher than that in both rural and remote areas of
Western Australia [24]. In addition, there was some suggestion in this initial study that
the incidence was increasing at a greater rate in the non-urban (rural and remote areas
combined) compared to urban areas of Western Australia, although the difference in
incidence rate trends between the two areas was not statistically significant [24]. By
extending the study period to 2010, this study has confirmed that between 1985 and
2010, the incidence of childhood Type 1 diabetes has increased at a greater rate in the
non-urban, and in particular in rural, compared to urban areas of Western Australia.
The higher rate of increase in incidence in non-urban areas of Western Australia is
predominantly due to a higher rate of increase in rural areas of the State. The difference
in incidence of childhood Type 1 diabetes between rural and urban areas of Western
Australia has narrowed, with the incidence in rural areas catching up to that in urban
areas. The question is what changes have there been over the past two decades in rural
or urban areas of Western Australia which might account for this change in incidence of
childhood Type 1 diabetes?
As described in the Background chapter (Chapter 3), Western Australia has a very large
land area with diverse climatic and ecological conditions. The majority of its
population resides in and around the capital city of Perth. Over the last two decades
136
there has been considerable population growth, both in the urban areas surrounding
Perth, as well as some of the other main population centres in rural areas of the State
[363, 419]. Perhaps, the converging of incidence rates in rural and urban areas of
Western Australia reflects the increased urbanisation of some rural parts of the State.
Environmental exposures that may have changed in rural areas due to increasing
urbanisation which may be related to childhood Type 1 diabetes include dietary factors,
obesity, access to fast food outlets, physical activity and time spent outdoors, sun
exposure and Vitamin D levels, housing and population density, access to group
childcare facilities, frequency of contact with animals and environmental pathogens and
different occupations being available for parents. These changes may be resulting in
either an increase in predisposing factors, or a decrease in factors which protect children
living in non-urban areas of Western Australia from developing the disease.
In addition to changes in environmental factors in these areas, could there be changes in
genetic factors that may be contributing to the converging incidence? For example,
differences in immigration patterns of individuals from low-risk populations such as
Asia, or high-risk populations such as Northern Europe, to urban compared to non-
urban areas, may alter the pool of genetically susceptible individuals in the populations
of these areas.
Ongoing monitoring of the incidence and incidence rate trends of childhood Type 1
diabetes in Western Australia will be important to determine whether or not the
incidence in rural areas levels off at a similar level to that found in urban areas, or if it
continues to increase further. In addition, further investigation into possible
environmental or genetic factors which may be contributing to this regional variation in
the incidence of childhood Type 1 diabetes is needed. Possible future studies could
137
include analysing the incidence for clustering over space and time to identify particular
“hot spots”, as well as examining factors that may vary between regions such as
population density, obesity, and immigration patterns or differences in the non-
Indigenous ethnic make-up of the population at risk.
6.4 Chapter Summary
As per the previous chapter, Chapter 5, this chapter consisted of two main sections.
Both sections described the variation in the incidence of childhood Type 1 diabetes in
Western Australia by place of residence at the time of diagnosis. Findings from the
initial study using a study period of 1985 to 2002 were described in the first half of the
chapter, and findings from the follow-up study, which used a study period of 1985 to
2010, were described in the second half.
The incidence of childhood Type 1 diabetes in Western Australia was found to vary by
place of residence at the time of diagnosis. Key findings from the initial study included
a higher mean incidence of childhood Type 1 diabetes being associated with an urban
place of residence at the time of diagnosis and with living in areas of higher
socioeconomic status.
The follow-up study, which examined the incidence of childhood Type 1 diabetes for
regional variation between 1985 and 2010, found that the mean incidence in rural areas
of Western Australia was no longer lower than that found in urban areas. In addition,
the rate of increase in the incidence of childhood Type 1 diabetes in non-urban areas of
Western Australia has been greater than that in the urban areas. These findings provide
further evidence for the role of environmental factors in the aetiology or clinical onset
of childhood Type 1 diabetes. The next challenge is to attempt to identify potential risk
138
or protective factors which have changed in non-urban areas of Western Australia over
the past two decades, which may be related to the changing epidemiology of childhood
Type 1 diabetes seen in this population.
The following chapter, Chapter 7, describes a different epidemiological study that was
undertaken in Western Australia to investigate the association between the incidence of
childhood Type 1 diabetes and various factors in the perinatal period.
139
Chapter 7: Perinatal risk factors for childhood Ty pe 1
diabetes in Western Australia
140
7.1 Preface
The previous two chapters, Chapters 5 and 6, described the epidemiology of childhood
Type 1 diabetes in Western Australia from 1985 to 2010. The incidence was found to
have increased over time in Western Australia and a significant 5-year cyclical variation
in the temporal incidence rate trend was observed. In addition, significant regional
variation in the incidence of childhood Type 1 diabetes in Western Australia was
observed and a higher incidence was found in children living in areas with a higher
socioeconomic status.
These findings provide evidence for the role of environmental factors in the aetiology or
clinical onset of childhood Type 1 diabetes but do not identify what factors may
explain the observed associations. Also, it is not known whether or not the effect of
such factors on the subsequent risk of childhood Type 1 diabetes could vary according
to the timing of exposure. For example, does place of residence or socioeconomic
status at the time of birth have a similar association with the risk of childhood Type 1
diabetes that was found with these factors at the time of diagnosis?
Although not observed in Western Australia [23], a higher rate of increase in the
incidence of Type 1 diabetes in children under the age of 5 years compared to older
children has been reported in several populations [68, 69]. The earlier onset of
childhood Type 1 diabetes in these young children suggests that factors in early life and
the perinatal period could be important in the aetiology of the disease. Perinatal factors
which have been associated with a higher risk of developing childhood Type 1 diabetes
include older maternal age, higher birth weight and lower birth order [203, 347, 351,
355, 420]. In Western Australia, perinatal data collection is mandatory for all hospital
141
and home births and is available for all births since 1980. The data are stored in the
Midwives’ Notification System at the Western Australian Department of Health and
include variables relating to the mother, pregnancy, labour and infant. A more detailed
description of the Midwives’ Notification System, including the variables available for
research purposes and data validation steps taken to ensure their accuracy, is provided in
Chapters 3 and 4 (Background and Methods). The Methods chapter also describes in
more detail the record linkage process that was performed to identify the perinatal
records of children diagnosed with Type 1 diabetes in Western Australia under the age
of 15 years from the Midwives’ Notification System.
The third aim of the research presented in this thesis was to use record linkage with the
complete population data available in the Midwives’ Notification System to investigate
the association between perinatal factors and the incidence of childhood Type 1 diabetes
in children aged 0-14 years in Western Australia.
7.1.1 Hypotheses
The specific research hypotheses being tested were as follows:
1. There is no association between maternal and infant characteristics and the
incidence of childhood Type 1 diabetes in Western Australia.
Maternal characteristics to be examined:
Pre-existing diabetes
Gestational diabetes
Ethnicity
Maternal age at delivery
Place of residence at time of birth
142
Socioeconomic status at time of birth
Infant characteristics to be examined:
Plurality
Month of birth
Birth weight
Gestational age
Birth order
Year of birth
2. There is no difference in the effect of the maternal and infant characteristics and the
incidence of Type 1 diabetes by age of onset of the disease (comparing early onset
(< 5 years) versus later onset (5-14 years) disease).
This chapter provides the details on the subjects and specific methods that were used to
address these research hypotheses and the resulting findings. The chapter was
published in 2007 as a journal article titled "Perinatal risk factors for childhood Type 1
diabetes in Western Australia - a population based study (1980 - 2002)" in the journal
Diabetic Medicine [25] (Appendix F). The journal article has been reproduced in this
chapter in its entirety. The results reported in this publication have also been presented
at several local, national and international conferences (Appendix F). In addition, a
presentation was given to local researchers to provide feedback on the accuracy of the
record linkage process that was performed for this study (Appendix F). Data from this
research have also been used in meta-analyses of the association between several
perinatal factors and childhood Type 1 diabetes [344, 350, 354].
143
7.2 Perinatal risk factors for childhood Type 1 d iabetes in
Western Australia - a population based study (1980 – 2002)
[This section is an unedited version of the journal article published in Diabetic Medicine
[25]. A PDF copy of the original journal article is included in Appendix F]
7.2.1 Abstract
Aims
To investigate perinatal risk factors for childhood Type 1 diabetes in Western Australia,
using a complete population-based cohort.
Methods
Children born between 1980 and 2002 and diagnosed with Type 1 diabetes aged <15
years (n=940) up to 31st December 2003, were identified using a prospective
population-based diabetes register with a case ascertainment rate of 99.8%. Perinatal
data were obtained for all live births in Western Australia from 1980 to 2002
(n=558,633) and record linkage performed to identify the records of cases.
Results
The incidence of Type 1 diabetes increased by 13% for each 5-year increase in maternal
age (adjusted IRR 1.13 (95%CI: 1.05 – 1.21)), by 13% for every 500g increase in birth
weight (adjusted IRR 1.13 (95%CI: 1.04 – 1.23)). The incidence decreased with
increasing birth order (adjusted IRR 0.89 (95%CI: 0.82 – 0.96)) and increasing
gestational age (adjusted IRR 0.84 (95%CI: 0.77 – 0.93)). A higher incidence of Type 1
diabetes was associated with an urban versus non-urban maternal address at the time of
144
birth (adjusted IRR 1.38 (95%CI: 1.18 – 1.63)), but no association was found with
socioeconomic status of the area.
Conclusions
A higher incidence of Type 1 diabetes was associated with increasing maternal age,
higher birth weight, lower gestational age, lower birth order and urban place of
residence at the time of birth.
7.2.2 Introduction
Although the aetiology of Type 1 diabetes (T1DM) remains unknown, it is generally
accepted to be the result of immune-mediated destruction of pancreatic beta-cells
caused by complex interactions between genetic and environmental factors [136, 137].
Over the past two decades, the incidence of childhood T1DM in Western Australia has
increased by an average of 3% a year [23], similar to the rate of increase reported in
many countries worldwide [17, 71]. The increase in incidence is too rapid to be due to
shifts in the population gene pool and is more likely to be the result of changes in still
unknown environmental factors. Environmental exposures that may have a role include
viruses, reduced exposure to microbial agents in early childhood and early introduction
of cow’s milk protein [83, 413].
More recently, several studies have reported a greater rate of increase of T1DM in
children under 5 years of age [69, 71] suggesting that factors in early life, including the
perinatal period, may be important in the aetiology of T1DM [83, 413]. Many studies
have examined the relationship between the incidence of T1DM and various perinatal
factors, but findings between studies have been inconsistent.
145
Maternal age at delivery has been associated with an increased risk of T1DM in several
cohort studies [342, 346, 351]. Higher birth weight has also been associated with an
increased risk [345, 346] but some studies have found no association between the risk of
T1DM and birth weight [341, 349]. Preterm birth has been associated with an increased
risk of T1DM [341, 346], but several studies have found no association between the risk
of T1DM and gestational age [345, 351]. Similarly, a decreased [351] and increased
[346, 352] risk of T1DM has been reported in first born children, as well as no
association with birth order [353].
Western Australia presents a unique opportunity to investigate in detail, perinatal
factors associated with T1DM as complete, population-based data are available on all
children diagnosed with T1DM under the age of 15 years, and for all births in the State
since 1980. Perinatal data collection has been mandatory for all births in Western
Australia since 1980.
Princess Margaret Hospital is the only tertiary paediatric hospital in the State and the
only referral centre for children diagnosed with diabetes. Children diagnosed with
T1DM can be identified from the Western Australia Children’s Diabetes Database, a
prospective population-based diabetes register with a case ascertainment rate of 99.8%
[24]. Data from this register have shown an increase in the incidence of childhood
T1DM in Western Australia by 3% a year from 1985 to 2002 [23], and that higher
socioeconomic status and residence in urban areas at the time of diagnosis are
independently associated with an increased risk of T1DM in Western Australian
children [24]. This raises the question of what environmental factors may explain these
findings and whether the effect of such factors on the risk of childhood T1DM might
vary according to timing of exposure. For example, do socioeconomic status and place
146
of residence at the time of birth have a similar association with the risk of T1DM that
was found at the time of diagnosis?
Therefore, the aim of this study was to investigate the association between perinatal
factors and childhood T1DM in Western Australia, including socioeconomic status and
place of residence at the time of birth, using a complete population-based cohort born
between 1980 and 2002.
7.2.3 Patients and Methods
Cases were defined as all children born in Western Australia between 1980 and 2002,
who were diagnosed with T1DM up to the 31st December 2003, aged less than 15 years
at diagnosis. The diagnosis of T1DM was based on clinical presentation and the
presence of islet cell autoantibodies and/or autoantibodies to glutamic acid
decarboxylase (GAD65). As these patients are managed by the same team until they are
at least 16 years old, this provides the opportunity to review and reconsider the type of
diabetes. Patients with maturity-onset diabetes of the young (MODY), Type 2 diabetes
or diabetes secondary to other causes such as cystic fibrosis were excluded from the
study. Eligible cases were identified using the previously described prospective,
population-based Western Australian Children’s Diabetes Register at Princess Margaret
Hospital [23].
Perinatal data were obtained for all live births in Western Australia from 1980 to 2002
from the Midwives’ Notification System. The Midwives’ Notification System contains
data on over 99% of all births in Western Australia [421]. The data are collected using
a standard form, by midwives at the time of the birth, and include details on the mother,
infant, pregnancy, labour and delivery. In Australia, the definition of live birth is ≥20
147
weeks gestation or birth weight ≥400g [376]. Birth order was determined from the
number of previous births reported by the mother at the time of birth of the index child.
Data are available on the presence of maternal diabetes prior to pregnancy and
gestational diabetes occurring during the current pregnancy. However, no
differentiation is made between pre-existing maternal Type 1 and Type 2 diabetes.
Maternal ethnicity is self-reported at the time of birth. In the Midwives’ Notification
System, all people of Caucasoid descent (European heritage) are classified as being
Caucasian. The non-Caucasian population in Western Australia is principally
composed of Indigenous Australians and people of Asian descent.
Record linkage was performed by the Western Australia Data Linkage Unit to identify
perinatal records of the cases, using probabilistic matching of name, sex, date of birth
and address variables, combined with an extensive manual review of doubtful links.
Study cohorts
Due to the potential confounding effects of ethnicity, plurality and maternal diabetes
[422, 423], the effects of these factors on the incidence of T1DM were analysed for the
whole cohort, consisting of all live births occurring during the study period.
The analysis of the incidence of T1DM by year of birth, birth weight, gestational age,
maternal age and birth order was then restricted to singleton pregnancies of Caucasian
mothers, without pre-existing or gestational diabetes.
Place of residence & Socioeconomic status at time of birth
The maternal address at the time of birth was used to analyse the incidence by area and
socioeconomic status of that area. The postcode was used to categorise births into urban
148
and non-urban areas, as defined by the Western Australia Department of Health. In
Western Australia, the urban area consists of the metropolitan areas surrounding the
capital city of Perth, with the remainder of the State being classified as non-urban.
In addition, the maternal address was used to analyse the incidence by socioeconomic
status. The Index of Relative Socioeconomic Disadvantage is published every 5 years
by the Australian Bureau of Statistics based on census data, and provides a summary
measure of the socioeconomic status of a geographical area [23, 379]. It is constructed
from variables such as the proportion of individuals in an area, who are on a low
income, are of Indigenous descent and are unemployed or in relatively unskilled
occupations. Each record was mapped to its corresponding census collection district
and the Index of Relative Socioeconomic Disadvantage score for this geographical area
obtained from the census year closest to the year of birth. Births were then categorised
into quintiles of socioeconomic status based on the scores of the study cohort. Data on
socioeconomic status were only available for births between 1984 and 1998. These
methods are similar to those used previously for the analysis of these factors at the time
of diagnosis [24]δ.
Ethical approval for this study was obtained from the Princess Margaret Hospital Ethics
Committee and the Western Australia Confidentiality of Health Information Committee.
Written, informed consent was obtained from all patients prior to their data being stored
on the diabetes database at Princess Margaret Hospital.
δ A more detailed explanation of using the Index of Relative Socioeconomic Disdvantage as a measure of socioeconomic status is given in Chapter 4 (Methods)
149
7.2.4 Statistical Analysis
All live births in Western Australia between 1980 and 2002 contributed to person time
under observation from birth until diagnosis of Type 1 diabetes, the age of 15 years or
the 31st December 2003, whichever occurred earlier. Deaths under the age of 15 years,
occurring during the study period were censored.
Poisson regression modelling with aggregated data was used to analyse incidence rates
for differences between and trends across variable categories, and to test for
interactions. Covariate selection for the multivariate analyses was based on known
biological inter-relationships and optimal model fit. To test for differences in effect
according to age at disease onset, the interaction term between each perinatal factor and
age group at diagnosis (<5 years, 5–14 years) was also examined. Seasonality of month
of birth was analysed using the method described by Stolwijk et al [353] which is based
on rates and therefore controls for seasonal variation of births in the general population.
A two-tailed p value of less than 0.05 was considered statistically significant. Data
analysis was performed using STATA statistical software (Stata Corporation, version
9.0).
7.2.5 Results
Data linkage
There were 940 eligible cases identified from the diabetes register at Princess Margaret
Hospital. Of these, 926 (98.5%) were successfully linked to the Midwives’ Notification
System. Perinatal data were obtained for 926 cases and 557,707 non-cases (Table 7.1).
150
All live births
Pre-existing maternal diabetes, gestational diabetes, ethnicity, plurality
A significantly higher incidence of T1DM was found in babies born to mothers with
pre-existing diabetes (IRR 4.74 (95%CI: 2.93 – 7.66), p<0.001) and those with
Caucasian compared to non-Caucasian mothers (IRR 3.73 (95%CI: 2.61 – 5.34),
p<0.001). The increased incidence of T1DM in babies born to mothers with gestational
diabetes was not significant, even after adjusting for maternal age and year of birth (IRR
1.47 (95% CI: 0.89 - 2.42, p = 0.13)). No significant difference in the incidence of
T1DM was found between babies from singleton compared to multiple pregnancies
(IRR 0.86 (95%CI: 0.58 – 1.27), p=0.45).
Table 7.1: Number of births by case, gender and age group at diagnosis for the
cohort of all live births in Western Australia between 1980 and 2002, and the sub-
cohort of singleton, Caucasian live births with no maternal pre-existing or
gestational diabetes
All live births
Singleton Caucasian live births, no maternal diabetes
Cases n = 926 n = 840
Boys 0-4 yrs 147 135 5-9 yrs 163 148 10-14 yrs 143 132 Total 453 415 Girls 0-4 yrs 127 114 5-9 yrs 217 199 10-14 yrs 129 112 Total 473 425
Non-cases n = 557,707 n = 466,004
Boys 286,584 239,268 Girls 271,123 226,736
151
Month of Birth
No significant seasonal variation was found when the incidence of T1DM was analysed
by month of birth for both genders combined (Chi 2 (3) = 3.94, p=0.27), or for boys and
girls analysed separately.
Singleton, Caucasian live births, no maternal diabetes
Due to the effects of ethnicity, plurality and maternal diabetes on birth weight and
gestational age, the remainder of the analyses was restricted to singleton live births to
Caucasian mothers with no pre-existing or gestational diabetes (840 cases, 466,004 non-
cases) (Table 7.1). The adjusted rate ratios are based on the analysis of 835 cases and
462,184 non-cases for which complete data on all covariates were available. Without
adjusting for other factors, the incidence of T1DM increased with increasing maternal
age, and decreased with increasing gestational age (Table 7.2).
After adjusting for year of birth, maternal age at delivery, birth weight, gestational age
and birth order respectively, a higher incidence of T1DM was associated with older
maternal age, higher birth weight, lower gestational age and lower birth order (Table
7.2). The incidence of T1DM increased by an average of 13% for every 5 year increase
in maternal age, and by an average of 13% for every 500g increase in birth weight. The
incidence in babies born <37 weeks was 43% higher than those born at 39-40 weeks,
and the incidence decreased by an average of 16% for each additional completed week
of gestation.
After adjusting for year of birth, maternal age and birth order, large premature babies
had the highest incidence of T1DM (Figure 7.1). There was a significant decrease in
incidence by an average of 11% with each increase in birth order.
152
The incidence was significantly higher for births in 1986-1991 and 1992-1997
compared to those in 1980-1985 (Table 7.2). However, no statistically significant trend
in the incidence of T1DM was observed with year of birth for the study cohort overall.
This may be due to incomplete 15-year follow-up of births after 1998, since cases in
births up to 2002 were included and follow-up ceased on 31st December 2003.
Figure 7.1: Relationship between gestational age and birth weight and the
incidence of Type 1 diabetes ( < 3000 g, 3000–3499 g, 3500–3999 g, ≥ 4000g,
----- linear trend) *
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1.60
1.80
<37 37-38 39-40 >40
Gestational Age (weeks)
Inci
denc
e R
ate
Rat
io
There was no significant interaction between age group at diagnosis (<5 years, 5–14
years) and maternal age, birth weight, gestational age, birth order or year of birth,
indicating no difference in the effects of these variables by age of disease onset.
* Figure 7.1 illustrates a linear decrease in incidence of T1DM with increasing gestational age at all birth
weights, implying that there is no interaction between gestational age and birth weight.
153
Table 7.2: Number of cases, total person years at risk, incidence of T1DM (per
100,000 person years), unadjusted and adjusted incidence rate ratios by maternal
age, birth weight, gestational age and birth order
Unadjusted Adjusted *
T1DM cases (n = 840)
Total person years
T1DM incidence (per 100,000)
IRR (95% CI) p† IRR (95% CI) p
Maternal age at delivery (years)
< 20 44 269,220 16.3 1.00 (0.73 – 1.38) NS 0.92 (0.66 – 1.27) NS 20 – 24 169 1,181,992 14.3 0.88 (0.73 – 1.06) NS 0.87 (0.72 – 1.05) NS 25 – 29 307 1,884,950 16.3 1.00 ‡ 1.00 30 – 34 234 1,226,668 19.1 1.17 (0.99 – 1.39) NS 1.18 (0.99 – 1.40) NS >=35 86 425,598 20.2 1.24 (0.98 – 1.58) NS 1.28 (1.00 – 1.64) NS Missing 0 226 Trend across categories
1.11 (1.04 – 1.18) 0.003 1.13 (1.05 – 1.21) 0.001
Birth weight (grams)
< 3000 144 958,126 15.0 0.90 (0.74 – 1.09) NS 0.79 (0.64 – 0.98) 0.04 3000 – 3499 311 1,853,775 16.8 1.00 ‡ 1.00 3500 – 3999 281 1,606,120 17.5 1.04 (0.89 – 1.23) NS 1.09 (0.92 – 1.28) NS >= 4000 104 570,332 18.2 1.09 (0.87 – 1.36) NS 1.19 (0.95 – 1.49) NS Missing 0 302 Trend across categories
1.06 (0.99 – 1.14) NS 1.13 (1.04 – 1.23) 0.003
Gestational age (weeks)
< 37 53 273,348 19.4 1.17 (0.88 – 1.56) NS 1.43 (1.04 – 1.96) 0.03 37 – 38 215 1,115,961 19.3 1.17 (0.99 – 1.37) NS 1.24 (1.04 – 1.47) 0.01 39 – 40 438 2,652,022 16.5 1.00 ‡ 1.00 > 40 134 915,656 14.6 0.89 (0.73 – 1.08) NS 0.86 (0.71 – 1.05) NS Missing 0 31,667 Trend across categories
0.89 (0.82 – 0.97) 0.008 0.84 (0.77 – 0.93) <0.001
Birth order
1st 356 1,976,492 18.0 1.00 ‡ 1.00 2nd 265 1,679,817 15.8 0.88 (0.75 – 1.03) NS 0.81 (0.69 – 0.95) 0.01 3rd 149 771,166 17.1 0.95 (0.78 – 1.15) NS 0.83 (0.68 – 1.01) NS 4th + 65 435,980 14.9 0.83 (0.64 – 1.08) NS 0.68 (0.52 – 0.90) 0.01 Missing 5 25,199 Trend across categories
0.95 (0.89 – 1.02) NS 0.89 (0.82 – 0.96) 0.004
Year of birth § 1980-1985 253 1,746,900 14.5 1.00 ‡ 1.00 1986-1991 326 1,782,314 18.3 1.26 (1.07 – 1.49)
0.005 1.20 (1.02 – 1.42) 0.03
1992-1997 213 1,112,727 19.1 1.32 (1.10 – 1.59) 0.003 1.22 (1.01 – 1.46) 0.04 1998-2002 48 346,713 13.8 0.96 (0.70 – 1.30) NS 0.85 (0.62 – 1.16) NS Missing 0 0 Trend across categories
1.07 (0.99 – 1.15) NS 1.03 (0.95 – 1.11) NS
* Adjusted for maternal age, birth weight, gestational age, birth order and year of birth
† NS = non-significant (p > 0.05)
‡ The most numerous category was used as the baseline group for all variables except year of birth, for which the earliest birth cohort was used as the baseline
§ 1980-1985 birth cohort has complete 15 years of follow-up; follow-up to 15 years of age is increasingly incomplete with successive birth cohorts as case ascertainment is to 31st December 2003
154
No statistically significant interactions were found between birth weight or gestational
age and maternal age or parity. There were inadequate case numbers in all categories to
assess the interactions between maternal age and parity, and between birth weight and
gestational age.
Place of residence at time of birth
The incidence of T1DM was 18.4 per 100,000 person years in births with an urban (648
cases), and 13.3 per 100,000 person years in those with a non-urban (192 cases)
maternal address at the time of birth. The incidence was 38% higher in births with an
urban compared to non-urban maternal address (IRR 1.38 (95%CI: 1.18 - 1.63),
p<0.001), and this difference remained significant after adjusting for the perinatal
variables examined in this study (IRR 1.33 (95%CI: 1.13 – 1.56), p=0.001).
As significant interactions were found between place of residence and year of birth, as
well as birth order, models were performed for each area separately. Associations
between the perinatal variables and incidence of T1DM, similar to those described
above for the whole cohort, were observed for births with an urban maternal address.
Overall, similar trends were also observed for births with a non-urban maternal address,
but these did not reach statistical significance.
Socioeconomic status at time of birth
Due to the limited availability of data on socioeconomic status, the analysis of the
incidence of T1DM by socioeconomic status at birth was further restricted to births
between 1984 and 1998. Socioeconomic Indexes for Area scores were obtained for
90% of the singleton live births to Caucasian mothers with no diabetes that occurred
during this time period (588 cases, 279,825 non-cases). There was an equal proportion
155
of missing scores for cases and non-cases. Possible explanations for missing scores
include having an address that could not be mapped to a census collection district, or
being in a census collection district for which the Australian Bureau of Statistics did not
publish an index value [379].
There was no significant difference in the incidence of T1DM between the five quintiles
of socioeconomic status, based on the Index of Relative Socioeconomic Disadvantage,
and no significant trend across the categories (IRR 1.00 (95%CI: 0.95 – 1.06), p=0.93).
After adjusting for year of birth, maternal age, birth weight, gestational age and birth
order, the effect of socioeconomic status remained unchanged (IRR 0.98 (95%CI: 0.93–
1.04), p=0.60). In addition, after adjusting each of the perinatal variables for
socioeconomic status, their effects on the incidence of T1DM remained unchanged.
7.2.6 Discussion
Using complete population-based data collected over 20 years, this study reports several
significant associations between perinatal factors and the incidence of childhood T1DM
in Western Australia.
There was a five-fold higher incidence of T1DM in babies of mothers with pre-existing
diabetes. While this probably reflects an increased genetic susceptibility in these
children, intra-uterine exposure to hyperglycaemia may also contribute to their
increased risk of T1DM [424]. A small, but non-significant increase in
the incidence of T1DM in babies of mothers with gestational diabetes, who would also
be exposed to hyperglycaemia, was found in this study. This may be due to
156
differences in genetic susceptibility between babies born to mothers with pre-existing
compared to gestational diabetes, or in the timing and degree of intra-uterine exposure
to hyperglycaemia.
The incidence of T1DM was almost four-fold higher in the offspring of Caucasian
compared to non-Caucasian mothers in Western Australia [376]. This may be a
reflection of increased genetic susceptibility to T1DM in Caucasians [425] or protection
in non-Caucasians, but could also be a marker of ethnic differences in environmental
and lifestyle factors including diet, antenatal care, breast feeding and child rearing
practices.
For singleton, Caucasian births to mothers without diabetes in this study, an increased
risk of T1DM was found with older maternal age, higher birth weight, lower gestational
age and lower birth order. These factors have been previously found to be associated
with T1DM in other populations. For example, older maternal age has been associated
with a higher incidence of T1DM in several populations [342, 351, 355]. This could be
due to increasing maternal age being a marker for factors associated with an increased
risk of T1DM such as accumulated toxins or exposure to infections [342], chromosomal
aberrations [426] and complications of pregnancy.
Findings on the association between birth weight and T1DM have been inconsistent
between studies in other populations. Higher birth weight has been associated with an
increased risk of T1DM in several studies [345, 346, 351], although others have found
no association between birth weight and T1DM [341, 349]. In this study, higher birth
weight was associated with an increased incidence of T1DM and the highest risk of
T1DM was observed in large, premature babies. A reduced risk of T1DM has been
157
previously observed in babies born small for gestational age [427]. The association
between birth weight and T1DM may be due to genetic factors that are associated with
both higher birth weight and risk of T1DM [359, 428]. Alternatively, higher birth
weight babies may have an altered metabolic profile that affects their future risk of
T1DM. For example, hyperinsulinaemia has been found in large-for-gestational-age
babies of mothers without diabetes during pregnancy [429]. The increased pancreatic
beta cell activity associated with hyperinsulinaemia may increase the risk of immune
mediated destruction of these cells [345].
Consistent with the findings of some studies [346, 430], a lower incidence of T1DM
was found with increasing gestational age in this study. However, other studies have
found no association between gestational age and the risk of T1DM [345, 351].
Similarly, this study reports a lower incidence of T1DM with increasing birth order
which has also been previously reported [346, 352], although again, others have found
no association between birth order and the risk of T1DM [341, 355]. An increased risk
of T1DM in first born compared to later born children may be due to reduced exposure
of first born children to infections, which are thought to be important for the appropriate
maturation of the developing immune system [213]. In addition, the association may be
due to variation in factors such as feeding practices and child care arrangements which
vary with family size [342].
As it has been suggested that some exposures have a greater effect in early-onset and
others in later-onset T1DM [346, 431], the associations between perinatal factors
examined in this study and T1DM were analysed for differences by age of disease
onset. No significant difference was found in this study in the effects of maternal age,
158
birth weight, gestational age, birth order or year of birth and the incidence of T1DM by
age at disease onset.
Residence in urban areas and higher socioeconomic status at the time of diagnosis have
been shown to be independently associated with an increased risk of T1DM in Western
Australian children [24]. In this study, residence in urban areas at the time of birth was
associated with a significantly increased incidence of T1DM. However, no association
was found with socioeconomic status at the time of birth.
The observed difference in incidence by place of residence at the time of birth is not due
to differing case ascertainment between these areas, as children diagnosed with T1DM
throughout the State are referred to Princess Margaret Hospital [24]. The higher
incidence of T1DM found in births with an urban maternal address at the time of birth is
most likely due to differences in environmental exposures between urban and non-urban
areas of Western Australia. The association between T1DM and place of residence at
both the time of diagnosis and the time of birth suggests that unknown environmental
factors may be acting to increase the risk of T1DM, either in utero or the early post-
natal period, as well as around the time of diagnosis.
The association previously found in Western Australia with socioeconomic status at
time of diagnosis, but not found in this study at the time of birth, is counter-intuitive.
Perhaps the area-based measure of socioeconomic status used in these studies is an
inadequate marker of factors related to T1DM risk that vary by socioeconomic status
during the perinatal period, but is an adequate marker for environmental exposures
involved around the time of diagnosis.
159
Inconsistent findings between studies investigating the relationship between perinatal
factors and T1DM may be due to differences in study design or the use of sample sizes
with insufficient power to detect small effects. This study was based on complete
population-based data routinely recorded at the time of birth for all live births in the
State and so was not subject to either recall or selection bias. However, a limitation of
using routinely collected data is the lack of information on variables which may be
important confounders and could not be accounted for, such as smoking during
pregnancy, paternal factors such as age and ethnicity, maternal weight gain and
individual measures of socioeconomic status.
The findings from this study are based on complete total population-based data over an
extended period and provide further evidence that factors during the perinatal period
have an important role in the aetiology of childhood T1DM. Further studies are
required to identify factors which may be contributing to the increasing incidence of
childhood T1DM in Western Australia and to elucidate possible causal pathways.
7.3 Chapter Summary
This chapter describes the association found between several perinatal risk factors and
the incidence of childhood Type 1 diabetes in Western Australia. The associations
found included a higher incidence of childhood Type 1 diabetes in babies born to older
mothers, to mothers with pre-existing or gestational diabetes, first born compared to
later born babies and those with a higher birth weight. In addition, a higher incidence of
childhood Type 1 diabetes was found in babies born to mothers with an urban place of
residence at the time of birth but no association was found with socioeconomic status of
the area. The findings from this study add to the growing literature suggesting that
160
factors in early life, including those in utero and in the early perinatal period, are
important in the aetiology of childhood Type 1 diabetes.
The following chapter, Chapter 8 (Synopsis) is the final chapter in this thesis and
discusses findings from all of the studies undertaken in this research into the
epidemiology of childhood Type 1 diabetes in Western Australia. In particular how
these findings may relate to increasing our knowledge of childhood Type 1 diabetes and
to developing future studies into the epidemiology of childhood Type 1 diabetes in
Western Australia are discussed.
161
Chapter 8: Synopsis
162
8.1 Preface
This chapter draws together key findings of the various studies presented in the previous
chapters. The implication and relevance of these findings are discussed in relation to
other studies to highlight the contributions that have been made to knowledge in the
field of childhood Type 1 diabetes epidemiology. In addition, the strengths and
limitations of the research studies presented in this thesis are discussed and
recommendations made for the future investigation of childhood Type 1 diabetes
epidemiology in Western Australia.
8.2 Succinct summary of findings
Since 1985, the incidence of childhood Type 1 diabetes has been increasing in Western
Australia by an average of 2.2% a year. Between 1985 and 2010, the mean incidence of
childhood Type 1 diabetes in Western Australia was 18.1 per 100,000 person years
(95% CI: 17.5 – 19.2), placing Western Australia in the “high incidence” category for
childhood Type 1 diabetes when compared to other populations [15]. Of particular
interest, a sinusoidal 5-year cyclical variation in the incidence of childhood Type 1
diabetes was observed, with almost identical peak and trough years observed in Western
Australia as have been reported in North-East England [399].
The incidence increased in both boys and girls, and in all age groups. No difference
was found in the mean incidence or incidence rate trends between boys and girls. The
lowest rate of increase in incidence was observed in children under the age of 5 years,
and a similar rate of increase observed in children aged 5 to 9 and 10 to 14 years.
Other key findings from the research presented in this thesis include the finding that, in
163
Western Australia, residence at the time of diagnosis in urban areas or in areas with a
higher socioeconomic status is independently associated with an increased risk of
childhood Type 1 diabetes. Although the mean incidence of childhood Type 1 diabetes
was higher in children living in urban compared to non-urban areas of Western
Australia, the gap appears to have narrowed in more recent years. Between 1985 and
2010, the incidence of childhood Type 1 diabetes has increased at a greater rate in non-
urban compared to urban areas of Western Australia, particularly in the rural rather than
remote parts of the State.
The final study presented in this thesis used linked data to investigate the role of
perinatal factors on the incidence of childhood Type 1 diabetes in Western Australia.
Significant associations were found between a higher incidence of childhood Type 1
diabetes and several maternal and infant characteristics. A higher incidence of
childhood Type 1 diabetes was associated with pre-existing maternal diabetes,
gestational diabetes, Caucasian ethnicity and older maternal age, as well as with higher
birth weight, lower gestational age, lower birth order and urban place of residence at the
time of birth.
8.3 Comparison with other studies
8.3.1 Temporal trends
The rate of increase in incidence of childhood Type 1 diabetes in Western Australia is
similar to that found in other populations both in Australia [61], and around the world
[17, 18, 55]. Non-linear temporal incidence rate trends of childhood Type 1 diabetes
have recently been reported in some populations [55], and a levelling off in incidence
has been reported in others [66]. A recent Australian report found there had been no
164
marked increase in the number of new cases of childhood Type 1 diabetes reported
nationally between 2005 and 2008 [7]. However, the findings from studies presented in
this thesis do not provide any evidence of a plateau in the incidence of childhood Type
1 diabetes in Western Australia. Instead, there is evidence for peaks and troughs in
incidence occurring approximately every 5 years on a background average increase in
incidence of 2.2% a year.
A 4-year cyclical pattern in the incidence of childhood Type 1 diabetes has been
previously reported in Yorkshire, England [398], and more recently a significant
sinusoidal 6-year cyclical variation in incidence was reported in North-East England
[399]. A finding of great interest from the research presented in this thesis is the
observation of almost identical peak and trough incidence years in Western Australia as
have been reported in North-East England [399]. Similar peak and trough incidence
years in these two populations, which are on opposite sides of the globe, and therefore
experience opposite seasons and very different demographic and climatic conditions, is
an intriguing finding. As described in section 5.3.6 of Chapter 5, the data presented in
this thesis and the study in North East England are the annual incidence rates so no
comparison can be made of differences in seasonal variation in incidence that may
occur due to the populations experiencing opposite seasons and infectious disease
cycles.
Epidemics, or peaks in the incidence of childhood Type 1 diabetes have been reported
in several populations over the past few decades [393, 394, 396] but no environmental
factor has been identified to explain this finding. Factors that have a cyclical nature,
and which might modify the risk of childhood Type 1 diabetes include infections [233,
248, 256] and climatic factors [403]. The climate and cyclical weather patterns may
165
play a role by altering the load of viral infections, the prevalence of environmental
pollutants [197, 198], or causing changes in dietary and exercise habits, including
exposure to sunlight and Vitamin D levels.
Analysis of temporal trends in the incidence of childhood Type 1 diabetes relates to
when children are diagnosed with the disease. The cyclical pattern observed in Western
Australia may reflect the effect of environmental factors that initiate the autoimmune
process, propagate progression of the disease process or precipitate the clinical onset of
the disease. Prospective longitudinal cohort studies, such as the BABYDIAB [432] and
TEDDY [112] studies, are required to answer the question of whether or not there is
cyclical variation in the temporal trend of pre-diabetes or the appearance of diabetes
related autoantibodies. If a cyclical pattern was observed in the temporal trend of pre-
diabetes, this would suggest that environmental factors may act to increase the risk of
childhood Type 1 diabetes by initiating the disease process, rather than precipitating the
clinical presentation of the disease.
8.3.2 Gender
The research findings presented in this thesis show that there was no difference in the
mean incidence or incidence rate trends of childhood Type 1 diabetes between boys and
girls in Western Australia. This is similar to findings from several studies, including an
Australia-wide study [55, 56, 60, 388], but in contrast to findings from other studies that
have observed an excess in girls [57, 58], or boys [268, 404, 433].
The reasons for these inconsistent findings are unclear. An example of a factor whose
influence may vary according to sex is body mass index or obesity, which may have a
greater effect in girls during puberty as they have greater insulin resistance compared to
166
boys during this developmental stage. Differential transmission of familial Type 1
diabetes by fathers and mothers with Type 1 diabetes to their male and female offspring
[59, 150] may also play a role, although this is likely to be consistent between different
populations. Another possible explanation for the lack of gender difference observed in
our study may be due to our study being based on children aged below 15 years. A few
studies which have reported a greater incidence in males compared to females have
observed this difference in older adolescents and young adults [55].
It is possible that environmental factors may have different effects on the risk of Type 1
diabetes according to gender [434] and the genetic make-up of different populations.
Could there be a different effect of risk factors such as viral infections according to sex
in populations with different genetic risk factors? If so, environmental influences in
countries whose populations have different genetic risk factors may result in differences
in the incidence of childhood Type 1 diabetes between boys and girls in these
populations.
8.3.3 Age at diagnosis
In many countries, the incidence of childhood Type 1 diabetes increases with increasing
age and the highest incidence is observed in 10 to 14 year olds [15]. In Western
Australia, the incidence in 0 to 4 year olds was significantly lower than in 5 to 9 and 10
to 14 year olds and there was no difference in the incidence between 5 to 9 and 10 to 14
year olds. Elsewhere in Australia, the greatest rate of increase was observed in children
under the age of 5 years in Victoria [65] but no difference was found in the rate of
increase between different age groups in New South Wales [61].
167
In Western Australia, the lowest rate of increase in incidence of childhood Type 1
diabetes was observed in children under the age of 5 years, which is in contrast to the
finding in many European studies that have observed the greatest rate of increase in
incidence in this age group [18, 55, 68-71, 206, 435]. Studies done in some populations
suggest that rather than an increase in the overall lifetime risk of Type 1 diabetes, the
disease is manifesting earlier and therefore presenting more frequently in the youngest
age groups [72, 73, 76, 404].
A possible explanation for the lowest rate of increase being observed in children under 5
years of age in Western Australia is that there is a difference in environmental factors
acting in utero, or in early childhood, that are modifying the risk of Type 1 diabetes in
our population. The onset of Type 1 diabetes under the age of 5 years is a marker of
high familial, and presumably genetic risk [152]. Perhaps there is a higher relative
contribution of genetic predisposition in the youngest age group and a greater
environmental influence as children get older [152, 436, 437]. If this were so, then the
stability of genetic factors over the past 25 years would account for the low rate of
increase in incidence seen in the youngest age group, and the change in environmental
influences over the same time period may account for the higher rate of increase in
incidence seen in 5 to 9 and 10 to14 year olds. However, it is difficult to postulate why
this may be the case in Western Australia and not in other parts of Australia or other
predominantly Caucasian populations worldwide.
8.3.4 Regional variation
The incidence of childhood Type 1 diabetes in Western Australia was found to be
significantly higher in children living in urban compared to non-urban areas at the time
of diagnosis.
168
The geographical variation in the incidence of childhood Type 1 diabetes both between
[15, 71] and within countries [189, 192] has been observed for many years. Findings on
the regional variation of incidence within countries have been inconsistent, with some
countries reporting a higher incidence in urban areas [189, 407] and others reporting a
higher incidence in rural areas [192]. Some studies, including an Australian study from
New South Wales [372], have found no difference in incidence between different
regions [408]. Prior to the research presented in this thesis, no study into the regional
variation of the incidence of childhood Type 1 diabetes had been undertaken in Western
Australia.
Differences in the findings between studies examining the regional variation in
incidence of childhood Type 1 diabetes may be due to the use of varying definitions of
urban/rural or differences in the characteristics of areas classified as urban/rural in
different countries. For example, children living in urban areas of Western Australia are
likely to be exposed to different environmental and lifestyle exposures compared with
urban areas of European countries and even New South Wales. This is because,
although Western Australia encompasses over a third of Australia’s land mass, it only
accounts for about 10% of the total population. Over 70% of Western Australia’s
population live in and around the capital city of Perth, which despite recent growth is
not a highly urbanised or population-dense city.
Between 1985 and 2010, the incidence rate increased at a greater rate in non-urban
compared to urban areas of Western Australia. In particular, the difference in incidence
between urban and rural parts of the State has narrowed. The converging of incidence
rates in rural and urban areas of Western Australia is particularly interesting, as it may
169
be a reflection of environmental factors important in the aetiology of childhood Type 1
diabetes which have changed due to increasing urbanisation over the past decade in
some rural parts of the State. A similar finding was reported in Austria, where the
incidence of childhood Type 1 diabetes was low in the Western parts of the country in
the early 1990s, but then showed a large increase, and by 2005 the incidence in this area
was greater than that in the Eastern parts of the country [194]. No cause was identified
to explain these changes in the geographic variation of incidence in Austria [194].
Similarly, the environmental exposures that have changed in rural areas of Western
Australia which could account for the increasing incidence observed in these areas are
yet to be identified. Increasing urbanisation may either be resulting in an increase in
factors which predispose children living in non-urban areas of Western Australia to
developing Type 1 diabetes, or a decrease in factors which protect them from the
disease.
In Western Australia, a higher incidence of childhood Type 1 diabetes was associated
with an urban place of residence at both the time of diagnosis and the time of birth. The
association between childhood Type 1 diabetes and place of residence at both the time
of diagnosis and the time of birth suggests that environmental factors which vary
between the urban and non-urban areas of Western Australia, may be acting to increase
the risk of Type 1 diabetes both in utero or the early post-natal period, as well as around
the time of diagnosis. Multiple environmental factors may be involved, with some
acting as initiators of the autoimmune process and others acting as disease propagators
or triggers of disease onset.
170
In some countries, regional variation in incidence between areas has been accounted for
by differences in characteristics such as population density and socioeconomic status.
In the research presented in this thesis, a higher socioeconomic status of the place of
residence at the time of diagnosis was found to be independently associated with an
increased incidence of childhood Type 1 diabetes. After taking the socioeconomic
status of the area into account, an urban place of residence at the time of diagnosis was
still found to be associated with a higher incidence of childhood Type 1 diabetes.
Therefore, in Western Australia, the regional variation in incidence of childhood Type 1
diabetes is not explained by differences in socioeconomic status of the areas.
8.3.5 Perinatal risk factors
Factors in early life have been thought to be important in the aetiology of childhood
Type 1 diabetes for some time. In particular, the rising incidence of childhood Type 1
diabetes observed in the youngest children in some populations suggests that factors in
early life and in utero may be important in the aetiology of the disease [68, 69].
In Western Australia, a higher incidence of childhood Type 1 diabetes was found to be
associated with maternal diabetes, gestational diabetes, Caucasian ethnicity of mother,
older maternal age, higher birth weight, lower gestational age and lower birth order. A
similar data linkage study undertaken in New South Wales found that maternal diabetes,
older maternal age, late pre-term birth and being born by Caesarean section were
associated with an increased risk of Type 1 diabetes in children up to the age of 6 years
[420]. No association was found in this study with birth weight, but being small-for-
gestational age reduced the risk of disease [420]. The increased risk of childhood Type
1 diabetes in the offspring of mothers with pre-existing diabetes may be due to an
increased genetic susceptibility in these children. Intra-uterine exposure to
171
hyperglycaemia may also contribute to their increased risk [424]. In addition, maternal
diabetes increases the risk of preterm birth, high birth weight and complications of
pregnancy [438], and this may in turn increase the risk of childhood Type 1 diabetes in
the offspring.
Recent meta-analyses which included the data used for the research presented in this
thesis have confirmed that there is an increased risk of childhood Type 1 diabetes in
older mothers [344], babies with a higher birth weight [350] and first born compared to
later born babies [354]. However, some studies in individual populations have reported
no association between birth weight [349], birth order [355], or gestational age [345,
351] and childhood Type 1 diabetes. A seasonal effect of month of birth on the risk of
childhood Type 1 diabetes has been reported in some populations [439, 440] but was
not observed in Western Australia.
Inconsistent findings between studies investigating the relationship between perinatal
factors and childhood Type 1 diabetes may be due to differences in study design, or the
use of sample sizes in some studies resulting in insufficient power to detect small
effects. However, true differences between populations in the association between
perinatal factors and childhood Type 1 diabetes cannot be excluded.
The following section describes the strengths and limitations of the research studies
presented in this thesis.
172
8.4 Strengths and limitations
8.4.1 Based on whole population at risk over a l ong study period
The studies undertaken during this research were all based on complete population-
based data over an extended study period. Therefore, a significant strength of the
findings presented in this thesis is that they were obtained by comparing factors
between all cases in Western Australia with the remainder of the total population at risk.
Although childhood Type 1 diabetes is one of the commonest chronic conditions
affecting children, it is still a rare disease. With rare diseases such as childhood Type 1
diabetes, studies using the whole population at risk are more likely to have sufficient
power to adequately investigate the role of potential risk factors.
As well as being based on complete population-based data, the studies presented in this
thesis examined data collected for over 25 years. This enabled the analysis of a greater
number of cases and a thorough and accurate investigation of incidence rate trends over
time.
Potential limitations of assessing the incidence over a long study period include varying
case ascertainment levels, demographic changes including those relating to changes in
migration patterns and changes in disease classification over time. Any of these factors
could influence the accuracy of estimated incidence rates. In addition, environmental
and lifestyle factors related to the risk of Type 1 diabetes, which the studies undertaken
as part of this research aimed to identify, may themselves have changed over the years.
In Western Australia all children diagnosed with diabetes are managed by hospital-
based paediatric endocrinologists at Princess Margaret Hospital. As this is the only
173
tertiary paediatric referral centre in the State, case ascertainment levels of the Western
Australian Children’s Diabetes Database, maintained at this hospital, have been
consistently high over time. Regarding demographic changes, there have been
significant changes in the patterns of immigration to Western Australia over the past
two decades. Available demographic changes measured by the Australian Bureau of
Statistics in the 5-yearly censuses were taken into account during this research, however
no relevant data on migration patterns were available. In particular, as discussed in
Section 8.4.6, there were no detailed population level data on migration patterns for
children in Western Australia.
8.4.2 Accurate classification of Type 1 diabetes
Another strength of the data used in these studies is that cases of Type 1 diabetes were
accurately classified as having the disease. This is becoming increasingly important, as
the number of children being diagnosed with Type 2 and other types of diabetes has
increased significantly in recent years [30, 441]. All children diagnosed with childhood
diabetes in Western Australia are followed-up prospectively by the same clinical team.
Therefore, any changes to their diagnosis following their initial admission are updated
accordingly in the Western Australian Children’s Diabetes Database.
8.4.3 Use of linked data
As described in Chapter 3 (Background), the studies presented in this thesis used data
from several State-wide data collections. The advantage of using routinely collected
data prior to the onset of Type 1 diabetes is that the data are not subject to either recall
or selection bias. However, as the data are routinely collected primarily for clinical or
administrative purposes, a limitation of using routinely collected data is the lack of
information on variables which may be important risk factors or confounders. For
174
example, when investigating the effect of socioeconomic status on the incidence of
childhood Type 1 diabetes, no data were available on individual measures of
socioeconomic status, such as parental education or income; and when investigating the
role of perinatal factors, complete data were not available for potentially important
variables such as smoking during pregnancy, maternal weight gain in pregnancy and
paternal factors such as age, ethnicity and pre-existing Type 1 diabetes.
8.4.4 Socioeconomic status measured at area, not individual level
For the research presented in this thesis, the Index of Relative Disadvantage was used as
the measure of socioeconomic status as it was the only measure available for the whole
population and enabled this important factor to be included in the analyses. The Index
of Relative Socioeconomic Disadvantage score is a summary measure of the
socioeconomic status of the geographical area, based on the characteristics of
individuals living in that area, rather than a measure of socioeconomic status at the
individual level [378]. A more detailed description of applying the Socioeconomic
Indexes for Area as a measure of socioeconomic status is provided in Chapter 4
(Methods).
A limitation of the findings regarding socioeconomic status from this research is that
they relate to the socioeconomic status of the area in which children live and cannot be
interpreted as relating to the child’s socioeconomic status at the individual or family
level. In order to minimise the heterogeneity of the socioeconomic status of individuals
living in an area, the index was assigned to individuals at the census collection district
level. This is the smallest geographical area for which the index scores are available
and it has been shown to provide a more accurate proxy measure of an individual’s
socioeconomic status than if the index is assigned at the larger, postcode level [416].
175
Another limitation of the findings regarding socioeconomic status is that it is difficult to
compare findings with other studies, due to different studies using different measures of
socioeconomic status. Different measures of socioeconomic status probably reflect
different underlying environmental factors which are associated with the risk of
childhood Type 1 diabetes. With this in mind, could different findings reported in the
literature reflect a real difference in the association found in different studies, or be an
artefact of analysis?
Socioeconomic status is a summary measure used to infer various environmental
exposures, which in turn have been hypothesised to being related to the risk of
childhood Type 1 diabetes. The nature of these environmental exposures and how they
may influence disease risk would be greatly influenced by the measure of
socioeconomic status used. For example, the measure used in this thesis is a summary
of factors measured at the area or place level, not the individual level and therefore
would be more reflective of factors such as housing, mean income levels, employment
levels and not individual socioeconomic factors per se. Other studies based on
individual levels of income or education for example would be a more accurate
individual measure of advantage and opportunity or lifestyle. In addition, the level of
socioeconomic advantage or disadvantage being measured would be highly variable
between different populations and countries. Therefore, it is difficult to conclude
whether or not the inconsistent findings reflect a real difference in association between
different studies. The differences are more likely to be a reflection of the different
measures and methods used in such studies.
176
8.4.5 Heterogeneity within non-urban areas of We stern Australia
In order to investigate the effect of place on the incidence of childhood Type 1 diabetes
in this research, the residential postcode at the time of diagnosis and at the time of birth
was used to categorise cases into urban, rural or remote areas, as defined by the Western
Australian Department of Health (See Background chapter, Figure 3.1).
As Western Australia encompasses a land area of over 2.5 million square kilometres,
there is wide variation in the climate and lifestyle characteristics of different regions
within the State. Therefore, areas defined as rural in one part of Western Australia may
have very different characteristics to areas defined as rural elsewhere in the State. For
example, rural areas in the South-west part of Western Australia have a Mediterranean
climate whereas those in more central and northern parts of Western Australia generally
experience hotter and drier conditions. In addition, rural areas in the South-west of the
State are closer to Perth and its surrounding urban areas compared to rural areas
elsewhere in the State. The varying characteristics are likely to influence the lifestyle
and exposures, including infections, which are experienced by residents living in these
different regions.
Therefore, a limitation of the analysis undertaken in this research was the use of area
definitions which resulted in heterogeneity within areas classified to be the same (rural
or remote). This heterogeneity makes it difficult to identify potential risk factors to
explain the observed differences in incidence of childhood Type 1 diabetes between
urban and non-urban areas. The differing proportion of Indigenous Australians in the
population of these areas was the only factor known to differ between the areas that was
taken into account in this analysis, and was shown not explain the difference in
177
incidence of childhood Type 1 diabetes between these areas.
8.4.6 No information available on interstate or overseas migration
In the study investigating the association between perinatal factors and childhood Type
1 diabetes, it was assumed that all babies born in Western Australia remained in the
State until 15 years of age and would therefore be detected as a case if they were
diagnosed with Type 1 diabetes. Loss of follow-up due to deaths in Western Australia
occurring before the age of 15 years was taken into account, but no information was
available on interstate or overseas migration. Total population level data on Interstate
and overseas migration figures are estimated by the Australian Bureau of Statistics
based on census data. The census questions that relate to migration are collected every
5 years and relate to the individual’s usual place of residence 1 year and 5 years prior to
the particular census collection date. Data on migration figures for children aged 0-14
years are limited, and therefore it was not possible to make any adjustments for
migration in this study [442]. However, migration of children aged 0 to 14 years is
likely to be random and unlikely to be different for cases and non-cases prior to disease
onset. Therefore, this should not have biased the results obtained from the study
investigating the role of perinatal factors and childhood Type 1 diabetes in Western
Australia.
8.5 Discussion
The incidence of childhood Type 1 diabetes continues to increase both in Australia and
around the world and the cause of this increase remains unclear. Type 1 diabetes is a
lifelong, chronic disease that places significant physical, psycho-social and financial
burdens on the affected individual and their family. To be able to identify factors which
result in Type 1 diabetes and to further the search for a prevention or cure for the
178
disease would mean alleviating lifelong suffering for patients diagnosed with this
serious condition, as well as reducing the health cost for society at large.
For many years, researchers have agreed that Type 1 diabetes results from complex
interactions between multiple genetic and environmental factors, some of which
increase the susceptibility to disease, and others which confer protection from it.
Over half of the genetic risk of Type 1 diabetes is accounted for by HLA genes and
more than 60 other genetic loci have now been identified [170-172]. However, the rate
of increase in the incidence of childhood Type 1 diabetes over the past few decades
cannot be explained by changes in genetic factors, as these are likely to have remained
relatively unchanged in populations over this period of time.
A key finding presented in this thesis which may help to generate hypotheses regarding
the aetiology of childhood Type 1 diabetes is that although the incidence of childhood
Type 1 diabetes in Western Australia is increasing at a similar rate to that observed in
other populations, the increase is non-linear. The significant 5-year cyclical variation in
the incidence of childhood Type 1 diabetes observed in Western Australia between 1985
and 2010 is strong evidence for the role of environmental factors in the aetiology or
clinical onset of this disease. Even stronger support for the role of environmental
factors is provided by the finding of almost identical peak and trough incidence years in
Western Australia as have been reported in North East England [399]. These findings
suggest that environmental factors that vary with a similar periodicity to the observed
peaks in incidence of childhood Type 1 diabetes, such as viral infections or climatic
conditions, are important in the aetiology of this disease.
Several other key findings presented in this thesis all strongly support the role of
179
environmental factors in childhood Type 1 diabetes. Firstly, the seasonal variation in
the incidence of childhood Type 1 diabetes, with a higher number of cases diagnosed in
the cooler autumn and winter months also supports the role of an infectious agent, or
environmental factor with an annual periodicity. Secondly, the difference in incidence
found between urban and non-urban areas of Western Australia, and the demonstration
of a narrowing in this difference in more recent years due to a higher rate of increase in
incidence in the non-urban compared to urban areas of the State over the past decade
can only be accounted for by changes in environmental and lifestyle factors in these
areas.
In addition to environmental factors acting during childhood, perinatal factors and
factors acting in utero are also thought to be important in the aetiology of childhood
Type 1 diabetes. As presented in this thesis, using a complete population-based cohort
of births in Western Australia between 1980 and 2002, pre-existing maternal diabetes,
gestational diabetes, older maternal age, higher birth weight, lower gestational age and
lower birth order were all identified as factors associated with a higher risk of
developing childhood Type 1 diabetes before the age of 15 years. Could changes in
these factors over time explain the increasing incidence of childhood Type 1 diabetes
observed in Western Australia?
In Western Australia, the proportion of women having babies aged 35 years and older
has increased significantly [443] from 4.7% of all births in 1980 to 20.8% in 2009
[444]. Over the same time, there has been a gradual decrease in the number of women
having babies between the age of 20 and 34 years. As well as the effect of older
maternal age on the risk of childhood Type 1 diabetes in their babies, older mothers are
at higher risk of complications of pregnancy, such as gestational diabetes, which may
180
themselves affect the risk of Type 1 diabetes in their babies. However, in the research
presented in this thesis, the increase in maternal age over time did not account for the
increase in incidence of childhood Type 1 diabetes over time.
As well as an increase in the proportion of babies born to older mothers in Western
Australia over the past two decades, there has also been an increase in the proportion of
babies born to mothers with gestational or pre-existing diabetes. From 1998 to 2009,
the proportion of singleton live births born to mothers with gestational diabetes or pre-
existing diabetes increased from 3.4% to 5.1%, and 0.4% to 0.8% respectively (Source:
Midwives Notification System (Ref:MCH11173) provided by Alan Joyce, December
2011). However, the increase in incidence of childhood Type 1 diabetes over time in
Western Australia was not explained by the increasing number of babies born to
mothers with gestational or pre-existing diabetes in this research.
A factor that has changed dramatically since the 1970s is the survival of babies born to
mothers with Type 1 diabetes. Could their improved survival, ability to achieve their
own reproductive potential and increasing fertility[153] be contributing to the
increasing incidence of childhood Type 1 diabetes by increasing the prevalence of
genetic factors associated with Type 1 diabetes in the population gene pool? There is
evidence to suggest that as well as the increased genetic susceptibility to Type 1
diabetes in babies born to mothers with Type 1 diabetes, non-genetic mechanisms that
alter foetal metabolism in response to intra-uterine exposure to hyperglycaemia, may
also contribute to an increased risk of the disease in these children [424]. Therefore, the
increased survival of babies born to mothers with Type 1 diabetes could be an important
factor in the increasing incidence of the disease. In the study of perinatal risk factors for
childhood Type 1 diabetes presented in this thesis, there was a 5-fold higher incidence
181
of Type 1 diabetes observed in babies born to mothers with pre-existing diabetes.
However, Type 1 diabetes is a relatively rare condition, and although the number of
babies born to mothers with Type 1 diabetes may have increased over the past few
decades and be partially contributing to the increasing incidence of the disease, it is
unlikely to account for the rapid rate of increase reported over this time in many
populations worldwide [17, 18].
Regarding other perinatal factors found to be associated with an increased risk of
childhood Type 1 diabetes in Western Australia, there has been no marked increase over
the past two decades in the number of preterm births [443] or number of babies born
with higher birth weight (Source: Midwives Notification System (Ref:MCH11173)
provided by Alan Joyce, December 2011).
As well as a higher incidence of childhood Type 1 diabetes being found in children
living in urban compared to non-urban areas of Western Australia at the time of
diagnosis, a higher incidence was also found in babies born to mothers living in urban
compared to non-urban areas of Western Australia at the time of birth. This difference
in incidence by place of birth remained after adjusting for perinatal factors such as
maternal age, parity and gestational age which may differ between urban and non-urban
areas due to different lifestyle and occupation choices made by mothers living in these
areas.
Therefore, although significant associations were found in this research between several
perinatal risk factors and the incidence of childhood Type 1 diabetes in Western
Australia, these associations did not help to explain either the increase in incidence over
time or the difference in incidence between urban and non-urban areas of the State.
182
Furthermore, as maternal age and other perinatal risk factors are unlikely to vary in a
cyclical nature, they would not account for the non-linear variation in the incidence of
childhood Type 1 diabetes observed in Western Australia. As yet unidentified
environmental factors remain the most likely explanation for both the non-linear
increase in incidence of childhood Type 1 diabetes in Western Australia and the
changing epidemiology of Type 1 diabetes in urban and non-urban areas of the State.
Perinatal factors that were not investigated in this research which may be relevant
include changes in the prevalence of maternal obesity, maternal smoking and babies
born by Caesarean section, both over time and between different areas of Western
Australia. As maternal obesity is associated with an increased risk of maternal diabetes
and gestational diabetes, it is likely to increase the risk of Type 1 diabetes in the
offspring. Maternal pre-pregnancy weight and weight gain during pregnancy have both
been associated with an increased risk of autoimmunity in their genetically at risk
offspring [445]. However, data on maternal pre-pregnancy weight or weight gain
during pregnancy are not currently obtained as part of the Midwives’ Notification data
collection.
Smoking in pregnancy is known to increase the risk of low birth weight, low gestational
age and complications of pregnancy [444] and may therefore modify the association
between these factors and the risk of childhood Type 1 diabetes developing in later
childhood. Therefore, it is important that the association of perinatal factors and
childhood Type 1 diabetes observed in Western Australia is re-analysed, taking the
effect of smoking in pregnancy on the in utero environment into account. Data on
whether or not mothers smoked during pregnancy have only been collected in Western
Australia since 1998 and no information is available on which trimester of pregnancy
183
smoking occurred or was stopped, or on the strength or number of cigarettes smoked.
Despite these limitations, it is recommended that the variables available on smoking in
pregnancy are analysed once sufficient data are available.
Finally, another perinatal factor which remains to be investigated in Western Australia
is the mode of delivery. A recent meta-analysis reported a 20% increase in the risk of
childhood Type 1 diabetes in children born by Caesarean section which was not
accounted for by known confounders such as maternal age, gestational age or maternal
diabetes [446]. Therefore, could the increasing incidence of childhood Type 1 diabetes
in Western Australia be accounted for by the significant increase in number of babies
born by Caesarean section in Western Australia over the past two decades?
Why have the environmental factors associated with Type 1 diabetes been so elusive?
Despite several decades of research into their investigation, very few environmental
factors have been confirmed to be involved in the aetiology of this disease. Congenital
rubella infection is a known risk factor for Type 1 diabetes, with ~20% of affected
children developing the disease in later life. However congenital rubella is a very rare
infection now due to vaccination of mothers against the disease. Other environmental
factors that have been implicated include enteroviral infections [248], mumps virus
infections , Vitamin D deficiency [270, 447], dietary factors such as cow’s milk protein
and gluten [300], rapid early growth and weight gain [318, 321], obesity and overweight
[315, 322]. However, none of these factors have been confirmed as risk factors that can
be modified to effectively prevent or cure Type 1 diabetes.
The lack of a thorough understanding of the time scale involved in the natural history or
pathogenesis of Type 1 diabetes is one of the reasons for the difficulty in identifying
184
risk factors for the disease. It is still not known how long it takes between the initiation
of the autoimmune process and clinical onset of the disease. Similarly, it is not known
how long it takes between triggers of Type 1 diabetes to act and patients to present with
clinical symptoms of the disease. Therefore, it is not known whether factors associated
with childhood Type 1 diabetes act as initiators of the autoimmune process, propagators
of the disease process or triggers of the clinical onset of disease, or at what time point
each of these stages occur. Therefore, large-scale prospective studies are needed to try
and identify what factors may be associated and when and how such factors are acting
in the disease pathway. An example of one such study is The Environmental
Determinants of Diabetes in the Young (TEDDY) study [112, 448], in which over
400,000 newborn babies have been screened for genetic risk factors for Type 1 diabetes.
Those identified as being at high risk are then being followed up regularly until the age
of 15 years. Tests being performed include blood tests for infections, nutritional
biomarkers and the appearance of islet autoantibodies.
Another reason why specific environmental factors for childhood Type 1 diabetes have
been so difficult to identify is because there are likely to be multiple causal pathways
that result in the disease [449]. Therefore, by investigating whole populations at risk
with the assumption of a single causal pathway, it may not be possible to clearly
identify the various risk factors that are present and acting in different ways in different
subgroups of the population. Furthermore, several mechanisms or factors acting
together in different combinations may be required to cause the disease [450].
This concept is supported by an as yet unpublished study done by researchers
investigating the genetic basis of Type 1 diabetes, who presented a novel finding of 6
distinct subtypes of Type 1 diabetes identified using data on over 3000 affected families
185
in the Type 1 Diabetes Genetics Consortium (T1DGC) database [451]. The subtypes
were found to differ both in terms of their genetic risk factors for Type 1 diabetes as
well as clinical features of the disease, such as age at disease onset and prevalence of
related autoimmune conditions [451]. This finding may provide a new way of
characterising patients with Type 1 diabetes and help in the search for environmental
risk factors which may only be relevant to certain subgroups. In addition, it may help to
consolidate the understanding of how genetic risk factors contribute to the disease
process and may provide evidence for there being multiple causal pathways that result
in this complex disease. Findings from another recent study suggest that it is important
to take into account genetic heterogeneity when analysing risk factors associated with
Type 1 diabetes, as the effect of the risk factor may vary according to what genetic
factors are present [452]. With this in mind, it is important to note that many of the
large-scale studies currently being undertaken in Type 1 diabetes epidemiology research
such as TEDDY, BABYDIAB and TRIGR, are based on cohorts with an increased
genetic predisposition to Type 1 diabetes. Therefore, will their results be directly
transferrable to the general population who may not have an increased genetic
susceptibility to the disease?
An added complexity to the understanding of childhood Type 1 diabetes is the growing
field of epigenetics. Epigenetics is the study of heritable changes in gene expression
that do not involve changes in the DNA sequence. Gene expression, or the turning “on”
and “off” of genes within cells, is regulated by changes to the chromatin structure by
mechanisms including DNA methylation and histone acetylation [453]. Changes in gene
expression result in different phenotypes or individuals being created without any
changes in the DNA sequence itself. If modifications of gene expression occur in
186
parents’ sperm or eggs, they become heritable traits which can be passed on to their
offspring and subsequent generations.
Several environmental factors are known to alter DNA methylation including dietary
factors such as folic acid, methionine, choline and lycopene as well as UV light [453,
454], exposure to prenatal tobacco smoke, alcohol consumption and environmental
pollutants [455]. The study of epigenetics may help with deciphering the aetiology of
childhood Type 1 diabetes by explaining how environmental factors such as viral
infections and dietary factors implicated in childhood Type 1 diabetes may influence the
expression of genes. In turn, this could help with the understanding of gene-
environment interactions found to be significant in the pathogenesis of this disease [449,
453, 455, 456]. For example, a relationship between food and epigenetic mechanisms
during the intrauterine and perinatal periods has been demonstrated in both animal [457]
and human studies [452]. Epigenetic mechanisms may provide the link between genetic
and environmental risk factors implicated in Type 1 diabetes and further studies in this
area are needed [458].
Together with the field of epigenetics, there have been recent discoveries made on the
role in many disease pathways of previously un-identified, small, non-coding RNA
molecules, called microRNAs (miRNAs), which act as regulators of gene expression
[459, 460]. Currently, ~20 miRNAs are known that have a role in pancreatic beta-cell
development and function. In vitro analyses and animal studies have shown that
miRNAs play key gene-regulatory roles during the early stages of pancreas
development and their expression appears to be crucial for proper beta-cell
development, survival and function [461]. Adding a further layer of complexity,
miRNA expression can regulate, or be regulated by, epigenetic modifications, resulting
187
in an epigenetic-miRNA regulatory circuit and possible biological pathways by which
gene-environment interactions influence disease risk [462].
As described above, there have been significant advances made in recent years
regarding the understanding of the molecular basis of Type 1 diabetes. The availability
of new technologies such as next-generation sequencing technology, metabolomics and
proteomics, in conjunction with advances in bio-informatics should continue to add to
the understanding of the underlying pathogenesis and aetiology of Type 1 diabetes.
8.6 Implication of Findings
The research presented in this thesis is the first comprehensive analysis of the
epidemiology of childhood Type 1 diabetes in Western Australia performed for an
extended period of over 25 years. In addition, the findings are from studies using
complete population-based data with accurate diagnosis of cases with Type 1 diabetes.
The key significant findings are the 5-year cyclical pattern in incidence of childhood
Type 1 diabetes observed in Western Australia, together with the variation in incidence
and incidence rate trends between urban and non-urban areas of the State.
The increasing incidence of childhood Type 1 diabetes in Western Australia shown by
this research illustrates the growing number of young children who continue to be
diagnosed with this lifelong, burdensome and serious disease. Individuals with Type 1
diabetes have an increased risk of multiple serious health problems which occur as
complications of the disease, as well as an increased mortality rate [11, 12]. The
mortality rate of children with Type 1 diabetes in Western Australia is 3 times higher
188
than the background population (Personal communication, Dr O’Grady, Princess
Margaret Hospital).
The novel finding of a non-linear, 5-year cyclical variation in incidence of childhood
Type 1 diabetes is a significant finding that will need to be taken into account when
projecting future case numbers, planning health services and estimating future health
costs. The prevalence of Type 1 diabetes in Australia is expected to increase by 10%
between 2008 and 2013 [46] and the estimated health expenditure on diabetes in
Australia is projected to increase by over 400% between 2003 and 2033 to
approximately $1.8 billion a year [463]. Although the majority of this cost is attributed
to the predicted rise in prevalence of Type 2 diabetes, the estimated life time cost of
Type 1 diabetes is significantly higher than Type 2 diabetes - $190,000 for Type 1
compared to $25,000 for Type 2 diabetes [464].
Current projections are likely to be based on linear estimates and the assumption that
the incidence of childhood Type 1 diabetes will continue to increase at the same rate
each year. However, if the incidence varies in a cyclical nature, it is important to take
this into account and not base future projections on either peak or trough incidence
years. Without the ability to accurately predict which years will be peak or trough
years, it will make the accurate allocation of annual health budgets and health service
planning more difficult.
Of great interest is the finding of almost identical peak and trough incidence years in
Western Australia as have recently been reported in North-East England [399]. This
provides strong evidence for the role of environmental factors in the aetiology or
clinical onset of childhood Type 1 diabetes, and the next step is to try and identify what
189
factors could be influencing the risk of childhood Type 1 diabetes in these distinct
populations during the same calendar years.
Another key finding presented in this research is the higher rate of increase in incidence
of childhood Type 1 diabetes observed in non-urban compared to urban areas of
Western Australia. This finding gives further support for the role of environmental
factors, suggesting that changes in lifestyle or environmental factors in non-urban areas
of Western Australia are resulting in a convergence of the incidence of childhood Type
1 diabetes to that found in urban parts of the State. The next step is to try and identify
such factors that have changed disproportionately in non-urban areas of Western
Australia that may account for the incidence in these areas becoming more similar to
that found in urban areas of the State.
8.7 Recommendations for Future Research
The findings presented in this thesis strongly support the role of environmental factors
in the aetiology or clinical onset of childhood Type 1 diabetes. Further studies are
required to identify the factors which may be contributing to the increasing incidence of
childhood Type 1 diabetes in Western Australia and to elucidate possible causal
pathways. The identification of aetiological factors of Type 1 diabetes is an important
area of research as this will ultimately enable the future development of early
intervention and prevention strategies for this serious and lifelong disease.
8.7.1 Cyclical variation in incidence
The next step in this area of research is to try and identify potential environmental
factors that vary in a cyclical nature and that may be associated with childhood Type 1
diabetes. This will need consultation with experts in the field of infectious diseases to
190
establish what monitoring systems are available and what infectious agents could be
relevant that we can study. Advice will need to be obtained on the Department of
Health’s Virus Watch facility, the merits of undertaking a data linkage study using data
from the local pathology specimen database (PathWest) which is available from 2000
onwards, and limitations of using retrospective data such as changes in virus detection
methods over time.
In addition, collaboration with colleagues in North East England where peaks in
incidence were observed in the same calendar years as Western Australia will be
important to try and identify any potential factors associated with childhood Type 1
diabetes occurring with the same periodicity in these two distinct populations.
It would also be informative to collaborate with the National Diabetes Register data
custodians at the Australian Institute of Health and Welfare, to determine whether the
temporal incidence rate trend of childhood Type 1 diabetes has a non-linear, cyclical
pattern in Australia as a whole, or if the pattern is specific to just the Western Australian
population. Could the apparent levelling off in the rate of new cases observed since
2005 [7] be due to a down side of a previous peak in incidence?
8.7.2 Regional variation in incidence
The narrowing of the gap in incidence of childhood Type 1 diabetes between urban and
rural areas of Western Australia is an interesting finding. Further studies are needed to
try and identify what environmental factors are changing in the rural areas of Western
Australia which may explain this.
Initially, it would be of interest to undertake a spatial analysis of the incidence of
191
childhood Type 1 diabetes in Western Australia to identify any potential clusters of the
disease as have been identified by studies in other populations [406, 410, 465]. Is the
increased incidence in rural areas accounted for by cases occurring in particular hot
spots or localities that are more urban in their characteristics? Is there an association
between the incidence of childhood Type 1 diabetes and population density?
Another study which may help to identify the importance of environmental factors
occurring in urban compared to non-urban areas would be to investigate the effect of re-
locating from non-urban to urban areas of Western Australia. In the research studies
presented in this thesis, there was a higher incidence of childhood Type 1 diabetes in
babies born to mothers with an urban compared to non-urban address at the time of
birth, and in children living in urban compared to non-urban areas of Western Australia
at the time of diagnosis. For the cases diagnosed with Type 1 diabetes, 70% of those
born to mothers with an urban address at the time of birth were still resident in urban
areas at the time of diagnosis. In contrast, only 20% of those born to mothers with a
non-urban address at the time of birth were still resident in non-urban areas at the time
of diagnosis. This may be a reflection of the migration of young families to urban areas
for work and schooling opportunities.
It would be interesting to examine the question of whether or not children born in non-
urban areas of Western Australia who develop Type 1 diabetes are more likely to have
moved to urban areas of the State prior to being diagnosed with the disease i.e. are they
more likely to have moved to urban areas than those born in non-urban areas who do
not develop Type 1 diabetes? However, as the time scale for the natural history of Type
1 diabetes is still not fully understood, a difficulty in answering this question is in
192
selecting the time point prior to disease onset that would be most appropriate for
assessing the degree of intrastate migration.
8.7.3 Childhood obesity
The prevalence of obesity and overweight in children has increased significantly over
the past few decades [311]. Rapid growth and obesity/overweight have been associated
with a higher incidence of childhood Type 1 diabetes [314, 318, 322]. Therefore, it
would be important to determine whether or not the increasing prevalence of obesity
and overweight in the population could partly explain the increasing incidence of
childhood Type 1 diabetes in Western Australia. In addition, could there be a difference
in the prevalence of obesity/overweight by area [466]? Has there been a
disproportionate increase in the prevalence of obesity/overweight in non-urban areas of
Western Australia that might explain the higher rate of increase in the incidence of
childhood Type 1 diabetes observed in rural areas of the State?
8.7.4 Smoking in pregnancy
As described earlier, data in the Midwives’ Notification System on maternal smoking in
pregnancy are complete for births from 1998 onwards. The study investigating
perinatal risk factors for childhood Type 1 diabetes that is presented in this thesis was
based on births between 1980 and 2002. Therefore, data on smoking in pregnancy were
only available for births occurring in the last 4 years of the study period. The small
number of cases meant that a detailed or multivariate analysis of the association
between smoking in pregnancy and the risk of childhood Type 1 diabetes was not
feasible at that time.
Smoking in pregnancy is known to have effects on birth weight, gestational age [467]
193
and complications of pregnancy. In addition, the prevalence of smoking in pregnancy
varies by maternal age [468] and socioeconomic status and has been reported by some
studies as reducing the risk of childhood Type 1 diabetes [469-471]. Therefore, it is an
important confounder that needs to be taken into account when analysing the effects of
perinatal factors such as maternal age, birth weight and gestational age on the risk of
developing childhood Type 1 diabetes in later life. It is recommended that the
association between perinatal risk factors and childhood Type 1 diabetes is re-examined
in the future, when data on smoking in pregnancy become available for a larger number
of births who have been followed up to the age of 15 years.
8.7.5 Mode of delivery
Again, as described earlier, there has been a significant increase in the number of babies
born by Caesarean section in Western Australia over the past two decades and being
born by Caesarean section has been associated with an increased risk of childhood Type
1 diabetes [446, 472]. Data on the mode of delivery are available from the Midwives’
Notification System, including whether or not the procedure was undertaken electively
or in an emergency. Therefore, it is important to investigate the role of being born by
Caesarean section on the future risk of childhood Type 1 diabetes and to determine
whether or not this factor accounts for the increasing incidence of childhood Type 1
diabetes observed in Western Australia.
8.7.6 Genetic factors
As well as trying to identify environmental factors that are associated with childhood
Type 1 diabetes, future research studies are needed which take genetic factors into
account. Current advances in technology and biostatistics can be applied to investigate
gene-environment interactions and the role of genetic factors known to be associated
194
with Type 1 diabetes, and this is becoming increasingly relevant [451, 452].
In Western Australia, the HLA-genotype of children diagnosed with Type 1 diabetes is
determined at the time of diagnosis but has not yet been investigated in any research
studies. For example, the data could be used to examine the proportion of cases with
high and low risk HLA types over time to see if there has been a decrease in the
proportion with the high risk types as has been observed elsewhere [181-184]. In
addition, it would be of interest to examine the proportion of cases with high and low
risk HLA types by age at diagnosis and calendar year, to see if there is any difference
found in the youngest age group that might help to explain the lowest rate of increase in
incidence of childhood Type 1 diabetes observed in this age group in Western Australia.
In addition to analysis of HLA-genotypes in children diagnosed with Type 1 diabetes in
Western Australia, it would of interest to investigate the role of the multiple other
candidate genes which have recently been identified for Type 1 diabetes [473].
Collaborative studies with researchers in the field of genetics and epigenetics may lead
to new avenues of investigating the role of such candidate genes in conjunction with the
detailed clinical characteristics that are available in the Western Australia Children’s
Diabetes Database. The use of newer technologies, such as next generation sequencing
technology and the concept of epigenome-wide association studies (EWAS) could also
be applied to further research questions in this field [474, 475].
8.7.7 Clinical and basic research
Results from observational studies may be subject to limitations such as potential
residual confounding, measurement errors and selection bias. Experimental studies in
animal models of Type 1 diabetes and in molecular biology have been used to further
investigate associations observed in human epidemiological studies to try and identify
195
potential biologically plausible pathways relevant to all aspects of the
immunopathogenesis and natural history of the disease [89, 476, 477]. As well as
furthering our understanding of the pathogenesis of Type 1 diabetes, animal and
molecular biology studies are fundamental to the investigation and testing of potential
preventative measures of the disease [478].
Over the past two decades, biological samples, including DNA have been collected on
children diagnosed with Type 1 diabetes in Western Australia at the time of diagnosis.
As yet, no comprehensive studies have been undertaken using these biological samples
to investigate the aetiology or changing incidence of Type 1 diabetes in this population.
It would be worthwhile to explore the application of emerging fields in diabetes
research such as proteomics and metabolomics [479] in planning future studies that
could investigate these samples. For example, a metabolomics approach was used in
the Finnish longitudinal DIPP study to analyse the metabolic profiles observed in blood
samples from the time of birth in children who later went on to develop type 1 diabetes
compared to those that did not [480]. Studies such as these could contribute important
information on the biological mechanisms which precede the initiation of Type 1
diabetes and onset of islet autoimmunity, occur during progression of the disease
process, or which precipitate clinical onset of the disease. Application of these new
technologies, as well as those in the fields of epigenetics, miRNA studies and
microbiome studies, in the area of Type 1 diabetes research are exciting developments
in the study of this complex and as yet, incompletely understood disease.
8.7.8 Continued monitoring of incidence
Finally, continued monitoring of the incidence of childhood Type 1 diabetes in Western
Australia is highly recommended. As total population-based data are available for all
196
children diagnosed with Type 1 diabetes in Western Australia, the data from Western
Australia are accurate and complete. Therefore, these data will provide a valuable
contribution to the local, national and global understanding of this disease.
As described earlier in this chapter, the time scale of the natural history of Type 1
diabetes is still poorly understood. Therefore, large scale prospective population-based
studies are more likely to be successful in identifying potential aetiological factors. In
addition, there may be multiple causal pathways to Type 1 diabetes with different
environmental and genetic factors acting to cause the disease in different subsets of the
population, which means it is of even greater importance to study a large number of
cases so that there is sufficient power to identify these subsets.
Novel study designs using low cost technology can now enable cost-effective large
scale prospective studies, such as the Million Women Study [481], to be undertaken.
The establishment of a “whole of life” database, or a longitudinal database that enables
the follow-up of patients once they transition from paediatric to adult care services will
be a valuable asset for investigating patterns of Type 1 diabetes and its complications in
our population. Collaboration with other key researchers in the field, both in Australia
and worldwide will be essential for maximising the likelihood of identifying risk factors
for childhood Type 1 diabetes. An example of such an initiative in Australia is the
newly commenced Environmental Determinants of Islet Immunity (ENDIA) study
[482]. This prospective, longitudinal study aims to enrol 1,400 children across
Australia to investigate various environmental and genetic risk factors that have been
implicated in the pathogenesis of Type 1 diabetes, such as weight gain, nutrition, the
microbiome and viral infections during pregnancy and early life. The establishment of
large scale prospective population-based studies such as this will be important for the
197
next step in building our understanding of the aetiology and immunopathogenesis of
Type 1 diabetes, and informing future studies which aim to prevent this serious, chronic
and life-threatening disease.
198
199
References
200
1. World Health Organization Department of Noncommunicable Disease
Surveillance, G., Definition, diagnosis and classification of diabetes mellitus
and its complications. Part 1: Diagnosis and Classification of Diabetes
Mellitus. Report of a WHO consultation. Accessed 6 November 2011 from
www.who.int/diabetes/currentpublications/en/ 1999.
2. American Diabetes Association, Diagnosis and Classification of Diabetes
Mellitus. Diabetes Care, 2012. 35(Suppl 1): p. S64-S71.
3. Couper JJ and D. KC, Phases of Diabetes in children and adolescents. Pediatric
Diabetes, 2009. 10 (Suppl 12): p. 13–16.
4. McMahon, S.K., et al., Increase in type 2 diabetes in children and adolescents in
Western Australia. Medical Journal of Australia, 2004. 180(9): p. 459 - 461.
5. American Diabetes Association, Diagnosis and Classification of Diabetes
Mellitus. Diabetes Care, 2010. 33(Suppl 1): p. S62-S69.
6. Aanstoot, H.-J., et al., The global burden of youth diabetes: perspectives and
potential. Pediatric Diabetes, 2007. 8(Suppl s8): p. 4-44.
7. Australian Institute of Health and Welfare (AIHW), Incidence of Type 1 diabetes
in Australian children 2000-2008. AIHW Catalogue number CVD 51, Canberra:
AIHW, 2010.
8. Banting, F.G. and C.H. Best, The internal secretion of the pancreas. The Journal
of Laboratory and Clinical Medicine, 1922. 7: p. 251.
9. Diabetes Control and Complications Trial/Epidemiology of Diabetes
Interventions and Complications (DCCT/EDIC) Research Group, et al., Modern-
day clinical course of type 1 diabetes mellitus after 30 years' duration: the
diabetes control and complications trial/epidemiology of diabetes interventions
201
and complications and Pittsburgh epidemiology of diabetes complications
experience (1983-2005). Archives of Internal Medicine, 2009. 169(14): p. 1307-
16.
10. Laing, S.P., et al., The British Diabetic Association Cohort Study, II: cause-
specific mortality in patients with insulin-treated diabetes mellitus. Diabetic
Medicine, 1999. 16(6): p. 466-471.
11. Skrivarhaug, T., et al., Long-term mortality in a nationwide cohort of childhood-
onset type 1 diabetic patients in Norway. Diabetologia, 2006. 49(2): p. 298-305.
12. Harjutsalo, V., C. Forsblom, and P.-H. Groop, Time trends in mortality in
patients with type 1 diabetes: nationwide population based cohort study. British
Medical Journal, 2011. 343: p. d5364.
13. LaPorte, R., et al., Geographic differences in the risk of insulin-dependent
diabetes mellitus: the importance of registries. Diabetes Care, 1985. 8(Suppl 1):
p. 101-107.
14. WHO DIAMOND Project Group, WHO Multinational Project for Childhood
Diabetes. Diabetes Care, 1990. 13: p. 1063-68.
15. Karvonen, M., et al., Incidence of Childhood Type 1 Diabetes Worldwide.
Diabetes Care, 2000. 23(10): p. 1516-1526.
16. Green, A., C. and Patterson, and on behalf of the EURODIAB TIGER Study
Group, Trends in the incidence of childhood-onset diabetes in Europe 1989-
1998. Diabetologia, 2001. 44 [Suppl. 3]: p. B3 - B8.
17. The DIAMOND Project Group, Incidence and trends of childhood Type 1
diabetes worldwide 1990 - 1999. Diabetic Medicine, 2006. 23: p. 857-866.
202
18. Patterson, C.C., et al., Incidence trends for childhood type 1 diabetes in Europe
during 1989-2003 and predicted new cases 2005-20: a multicentre prospective
registration study. The Lancet, 2009. 373(9680): p. 2027-2033.
19. Kelly, H.A. and G.C. Byrne, Incidence of IDDM in Western Australia in
children aged 0-14 yr from 1985 to 1989. Diabetes Care, 1992. 15: p. 515-517.
20. Holman CDJ, et al., Population-based linkage of health records in Western
Australia: development of a health services research linked database. Australian
and New Zealand Journal of Public Health, 1999. 23(5): p. 453-459.
21. Data Linkage Branch, D.o.H.W.A., Data Collections. Accessed 5 November
2011 from www.datalinkage-wa.org/data-linkage/data-collections.
22. Kelly, H.A., et al., Dramatic increase in incidence of insulin dependent diabetes
mellitus in Western Australia. Medical Journal of Australia, 1994. 161: p. 426-
429.
23. Haynes, A., et al., Continued increase in the incidence of childhood Type 1
diabetes in a population-based Australian sample (1985-2002). Diabetologia,
2004. 47(5): p. 866-870.
24. Haynes, A., et al., Independent effects of socio-economic status and place of
residence on the incidence of Type 1 diabetes in Australian children.
Diabetologia, 2004. 47: p. [Suppl 1]: I - VIII, Abstract 43.
25. Haynes, A., et al., Perinatal risk factors for childhood Type 1 diabetes in
Western Australia - a population-based study (1980-2002). Diabetic Medicine,
2007. 24(5): p. 564-570.
26. Gale, E.A.M., Declassifying diabetes. Diabetologia, 2006. 49: p. 1989-1995.
203
27. Wilkin, T., Changing perspectives in diabetes: their impact on its classification.
Diabetologia, 2007. 50(8): p. 1587-1592.
28. Daneman, D., Type 1 diabetes. Lancet, 2006. 367: p. 847-58.
29. Australian Institute of Health and Welfare (AIHW), Diabetes: Australian facts
2008. AIHW Catalogue number CVD 40, Canberra: AIHW, 2008.
30. Craig, M.E., A. Hattersley, and K.C. Donaghue, Definition, epidemiology and
classification of diabetes in children and adolescents. Pediatric Diabetes, 2009.
10 [Suppl 12]: p. 3-12.
31. Borchers, A.T., R. Uibo, and M.E. Gershwin, The geoepidemiology of type 1
diabetes. Autoimmunity Reviews, 2010. 9(5): p. A355-A365.
32. Australian Institute of Health and Welfare (AIHW), A rising epidemic: obesity
in Australian children and adolescents. AIHW Risk Factors Data Briefing
Number 2, Canberra: AIHW, 2004.
33. Australian Bureau of Statistics, Health of Children in Australia: A Snapshot,
2004-05. Catalogue No. 4829.0.55.001, Canberra: Australian Government
Publishing Service. Accessed 14 December 2011 from
http://abs.gov.au/AUSSTATS/[email protected]/allprimarymainfeatures/9C310A5E79B
9BF7CCA256F390074DF3E?opendocument 2007.
34. Craig, M.E., et al., Type 2 diabetes in Indigenous and non-Indigenous children
and adolescents in New South Wales. Medical Journal of Australia, 2007.
186(10): p. 497-9.
35. Mayer-Davis, E., et al., The Many Faces of Diabetes in American Youth: Type 1
and Type 2 Diabetes in Five Race and Ethnic Populations: The SEARCH for
Diabetes in Youth Study. Diabetes Care, 2009. 32([Suppl 2]): p. S99-S101.
204
36. Rana, M., et al., Increased detection of cystic-fibrosis-related diabetes in
Australia. Archives of Disease in Childhood, 2011. 96(9): p. 823-826.
37. Imagawa, A. and T. Hanafusa, Fulminant type 1 diabetes—an important subtype
in East Asia. Diabetes/Metabolism Research and Reviews, 2011. 27(8): p. 959-
964.
38. Imagawa, A., et al., A Novel Subtype of Type 1 Diabetes Mellitus Characterized
by a Rapid Onset and an Absence of Diabetes-Related Antibodies. New England
Journal of Medicine, 2000. 342(5): p. 301-307.
39. Ambler, G.R., et al., Contemporary Australian outcomes in childhood and
adolescent type 1 diabetes: 10 years post the Diabetes Control and
Complications Trial. Journal of Paediatrics & Child Health, 2006. 42(7-8): p.
403-10.
40. Edge, J.A., M.E. Ford-Adams, and D.B. Dunger, Causes of death in children
with insulin dependent diabetes 1990-96. Archives of Disease in Childhood,
1999. 81(4): p. 318-23.
41. Alhonen, S.M., et al., Extended Family History of Diabetes and Autoimmune
Diseases in Children With and Without Type 1 Diabetes. Diabetes Care, 2011.
34(1): p. 115.
42. Triolo, T.M., et al., Additional autoimmune disease found in 33% of patients at
type 1 diabetes onset. Diabetes Care, 2011. 34(5): p. 1211-3.
43. Milton, B., P. Holland, and M. Whitehead, The social and economic
consequences of childhood-onset Type 1 diabetes mellitus across the lifecourse:
a systematic review. Diabetic Medicine, 2006. 23(8): p. 821-9.
205
44. Colagiuri, S., et al., DiabCo$t Australia Type 1: Assessing the burden of Type 1
Diabetes in Australia. Canberra : Diabetes Australia. Accessed 7 March 2012
from http://www.jdrf.org.au/s/media/diabco_t.pdf, 2009.
45. International Diabetes Federation (IDF), IDF diabetes atlas (4th edition).
Brussels: IDF, 2009.
46. Australian Institute of Health and Welfare (AIHW), Prevalence of Type 1
diabetes in Australian children, 2008. AIHW Catalogue number CVD 54,
Canberra: AIHW. Accessed 14 December 2011 from
http://www.aihw.gov.au/publication-detail/?id=10737419239, 2011.
47. Green, A., et al., Incidence of childhood-onset insulin-dependent diabetes
mellitus: the EURODIAB ACE study. Lancet, 1992. 339: p. 905-09.
48. WHO DIAMOND Project Group on Epidemics, Childhood Diabetes,
Epidemics, and Epidemiology: An approach for controlling diabetes. American
Journal of Epidemiology, 1992. 135: p. 803-16.
49. Diabetes Epidemiology Research International Group, Geographic patterns of
childhood insulin-dependent diabetes mellitus. Diabetes, 1988. 37: p. 1113-19.
50. Karvonen, M., et al., A review of the recent epidemiological data on the
worldwide incidence of type 1 (insulin-dependent) diabetes mellitus.
Diabetologia, 1993. 36: p. 883-92.
51. Rytkonen, M., et al., Bayesian analysis of geographical variation in the
incidence of Type 1 diabetes in Finland. Diabetologia, 2001. 44(Suppl 3): p.
B37-B44.
52. Kondrashova, A., et al., A six-fold gradient in the incidence of type 1 diabetes at
the eastern border of Finland. Annals of Medicine, 2005. 37(1): p. 67-72.
206
53. Cherubini, V., et al., Regional variability in the epidemiology of childhood
diabetes in Italy. Journal of Pediatric Endocrinology & Metabolism, 1997. 10: p.
471-478.
54. Samuelsson, U. and O. Lofman, Geographical mapping of type 1 diabetes in
children and adolescents in south east Sweden. J Epidemiol Community Health,
2004. 58(5): p. 388-392.
55. Harjutsalo, V., L. Sjöberg, and J. Tuomilehto, Time trends in the incidence of
type 1 diabetes in Finnish children: a cohort study. The Lancet, 2008.
371(9626): p. 1777-1782.
56. Jarosz-Chobot, P., et al., Rapid increase in the incidence of type 1 diabetes in
Polish children from 1989 to 2004, and predictions for 2010 to 2025.
Diabetologia, 2011. 54(3): p. 508-515.
57. Rewers, M., et al., Trends in the prevalence and incidence of diabetes: insulin-
dependent diabetes mellitus in childhood. World Health Statistics Quarterly,
1988. 41: p. 179-189.
58. Karvonen, M., et al., Sex difference in the incidence of insulin-dependent
diabetes mellitus: An analysis of the recent epidemiological data. Diabetes-
Metabolism Reviews, 1997. 13: p. 275-291.
59. Gale, E.A.M. and K. Gillespie, Diabetes and gender. Diabetologia, 2001. 44: p.
3-15.
60. Australian Institute of Health and Welfare (AIHW), Incidence of Type 1 diabetes
in Australians under 40 years: a snapshot of National Diabetes Register data for
2004. AIHW Catalogue No. AUS 76, Canberra: AIHW, 2006.
207
61. Taplin, C.E., et al., The rising incidence of childhood type 1 diabetes in New
South Wales, 1990 - 2002. Medical Journal of Australia, 2005. 183: p. 243-246.
62. Catanzariti, L., et al., Australia's national trends in the incidence of Type 1
diabetes in 0-14-year-olds, 2000-2006. Diabetic Medicine, 2009. 26(6): p. 596-
601.
63. Diabetes Epidemiology Research International Group, Secular Trends in
Incidence of Childhood IDDM in 10 Countries. Diabetes, 1990. 39: p. 858-864.
64. Onkamo, P., et al., Worldwide increase in incidence of Type 1 diabetes - the
analysis of the data on published incidence trends. Diabetologia, 1999. 42: p.
1395-1403.
65. Chong, J.W., et al., Marked increase in type 1 diabetes mellitus incidence in
children aged 0–14 yr in Victoria, Australia, from 1999 to 2002. Pediatric
Diabetes, 2007. 8(2): p. 67-73.
66. Berhan, Y., et al., Thirty Years of Prospective Nationwide Incidence of
Childhood Type 1 Diabetes: The Accelerating Increase by Time Tends to Level
Off in Sweden. Diabetes, 2011. 60(2): p. 577.
67. Schoenle, E., et al., Epidemiology of Type 1 diabetes mellitus in Switzerland:
steep rise in incidence in under 5 year old children in the past decade.
Diabetologia, 2001. 44: p. 286-289.
68. Schober, E., B. Rami, and T. Waldhoer, Steep increase of incidence of childhood
diabetes since 1999 in Austria. Time trend analysis 1979–2005. A nationwide
study. European Journal of Pediatrics, 2008. 167(3): p. 293-297.
208
69. Gardner, S.G., et al., Rising incidence of insulin dependent diabetes in children
aged under 5 years in the Oxford region: time trend analysis. British Medical
Journal, 1997. 315: p. 713-717.
70. Karvonen, M., et al., The Onset of Type 1 Diabetes in Finnish Children Has
Become Younger. Diabetes Care, 1999. 22(7): p. 1066-1070.
71. EURODIAB ACE Study Group, Variation and trends in incidence of childhood
diabetes in Europe. Lancet, 2000. 355: p. 873-876.
72. Weets, I., et al., The incidence of type 1 diabetes in the age group 0-39 years has
not increased in Antwerp (Belgium) between 1989 and 2000: evidence for
earlier disease manifestation. Diabetes Care, 2002. 25(5): p. 840-6.
73. Pundziute-Lycka, A., et al., The incidence of Type 1 diabetes has not increased
but shifted to a younger age at diagnosis in the 0-34 years age group in Sweden
1983 to 1998. Diabetologia, 2002. 45: p. 783-791.
74. Zhao, H.X., et al., The average age at onset is decreasing in children with Type
1 diabetes in Devon and Cornwall, England. Diabetic Medicine, 2000. 17(12):
p. 877-877.
75. Feltbower, R.G., et al., Type 1 diabetes in Yorkshire, UK: time trends in 0–14
and 15–29-year-olds, age at onset and age-period-cohort modelling. Diabetic
Medicine, 2003. 20(6): p. 437-441.
76. Dahlquist, G.G., et al., Incidence of Type 1 Diabetes in Sweden Among
Individuals Aged 0–34 Years, 1983–2007. Diabetes Care, 2011. 34(8): p. 1754-
1759.
77. Gale, E.A.M., Spring harvest? Reflections on the rise of type 1 diabetes.
Diabetologia, 2005. 48: p. 2445-2450.
209
78. Lammi, N., et al., Marked temporal increase in the incidence of type 1 and type
2 diabetes among young adults in Finland. Diabetologia, 2008. 51(5): p. 897-9.
79. Bruno, G., et al., The incidence of type 1 diabetes is increasing in both children
and young adults in Northern Italy: 1984–2004 temporal trends. Diabetologia,
2009. 52(12): p. 2531-2535.
80. Glatthaar, C., et al., Diabetes in Western Australian children: descriptive
epidemiology. Medical Journal of Australia, 1988. 148: p. 117-123.
81. Soltesz, G., et al., Worldwide childhood type 1 diabetes incidence – what can we
learn from epidemiology? Pediatric Diabetes, 2007. 8: p. 6-14.
82. LaPorte, R., J. Dorman, and K. Cruickshanks, Methods for epidemiologic
investigation of insulin-dependent diabetes mellitus. In: Larner J, Pohl SL,
Clarke WL, eds. Methods in Diabetes Research, Volume II: Clinical Methods.
New York. John Wiley and Sons, Inc, 1986: p. 547-556.
83. Dahlquist, G., The aetiology of type 1 diabetes: an epidemiological perspective.
Acta Paediatrica, 1998. 87([Suppl s425]): p. 5-10.
84. Green, A., What can the epidemiological approach to type 1 diabetes contribute
to the unravelling of autoimmune pathological mechanisms? Springer Seminars
in Immunopathology, 2002. 24: p. 323-330.
85. Maahs, D.M., et al., Epidemiology of Type 1 Diabetes. Endocrinology &
Metabolism Clinics of North America, 2010. 39(3): p. 481-497.
86. Ziegler, A.G., E. Bonifacio, and t.B.-B.S. Group, Age-related islet autoantibody
incidence in offspring of patients with type 1 diabetes. Diabetologia, 2012.
55(7): p. 1937-1943.
210
87. Parikka, V., et al., Early seroconversion and rapidly increasing autoantibody
concentrations predict prepubertal manifestation of type 1 diabetes in children
at genetic risk. Diabetologia, 2012. 55(7): p. 1926-1936.
88. Steck, A.K., et al., Age of islet autoantibody appearance and mean levels of
insulin, but not GAD or IA-2 autoantibodies, predict age of diagnosis of type 1
diabetes: diabetes autoimmunity study in the young. Diabetes Care, 2011. 34(6):
p. 1397-9.
89. Rossini, A., et al., Immunopathogenesis of diabetes mellitus. Diabetes Reviews,
1993. 1: p. 43-75.
90. Atkinson, M.A., Thirty Years of Investigating the Autoimmune Basis for Type 1
Diabetes: Why Can't We Prevent or Reverse This Disease? Diabetes, 2005.
54(5): p. 1253-1263.
91. Skyler, J. and C. Ricordi, Stopping Type 1 Diabetes: Attempts to Prevent or
Cure Type 1 Diabetes in Man. Diabetes, 2011. 60(1): p. 1.
92. Bettini, M. and D.A.A. Vignali, T cell-driven initiation and propagation of
autoimmune diabetes. Current Opinion in Immunology, 2011. 23(6): p. 754-760.
93. Chen, W., A. Xie, and L. Chan, Mechanistic basis of immunotherapies for type 1
diabetes mellitus. Translational Research, 2013. 161(4): p. 217-229.
94. Galgani, M., et al., Meta-immunological profiling of children with type 1
diabetes identifies new biomarkers to monitor disease progression. Diabetes,
2013.
95. Alkanani, A.K., et al., Dysregulated Toll-Like Receptor–Induced Interleukin-1β
and Interleukin-6 Responses in Subjects at Risk for the Development of Type 1
Diabetes. Diabetes, 2012. 61(10): p. 2525-2533.
211
96. Roep, B.O., β-Cells, Autoimmunity, and the Innate Immune System: “un Ménage
á Trois”? Diabetes, 2013. 62(6): p. 1821-1822.
97. Gan, M.J., A. Albanese-O’Neill, and M.J. Haller, Type 1 Diabetes: Current
Concepts in Epidemiology, Pathophysiology, Clinical Care, and Research.
Current Problems in Pediatric and Adolescent Health Care, 2012. 42(10): p.
269-291.
98. Gepts, W., Pathologic anatomy of the pancreas in juvenile diabetes mellitus.
Diabetes, 1965. 14: p. 619-633.
99. Bottazzo, G., A. Florin-Christensen, and D. Doniach, Islet-cell antibodies in
diabetes mellitus with autoimmune polyendocrine deficiencies. Lancet, 1974. 2:
p. 1279-1283.
100. Nerup, J., P. Platz, and O. Anderssen, HLA antigens and diabetes mellitus.
Lancet, 1974. 2: p. 864-866.
101. Mallone, R. and B.O. Roep, Biomarkers for immune intervention trials in type 1
diabetes. Clinical Immunology, 2013.
102. Eisenbarth, G.S., Type 1 diabetes: a chronic autoimmune disease. New England
Journal of Medicine, 1986. 314: p. 1360-1368.
103. Gorsuch, A., et al., Evidence for a long prediabetic period in type I (insulin-
dependent) diabetes mellitus. Lancet, 1981. 2: p. 1363-1365.
104. Sosenko, J.M., et al., The Metabolic Progression to Type 1 Diabetes as
Indicated by Serial Oral Glucose Tolerance Testing in the Diabetes Prevention
Trial–Type 1. Diabetes, 2012. 61(6): p. 1331-1337.
105. Fourlanos, S., et al., Latent autoimmune diabetes in adults (LADA) should be
less latent. Diabetologia, 2005. 48: p. 2206-2212.
212
106. Gale, E.A.M., Latent autoimmune diabetes in adults: a guide for the perplexed.
Diabetologia, 2005. 48: p. 2195-2199.
107. Groop, L., et al., Latent autoimmune diabetes in adults (LADA) - more than a
name. Diabetologia, 2006. 49: p. 1996-1998.
108. Kogevinas, M., A.-M.N. Andersen, and J. Olsen, Collaboration is needed to co-
ordinate European birth cohort studies. International Journal of Epidemiology,
2004. 33(6): p. 1172-1173.
109. Rewers, M., et al., Newborn screening for HLA markers associated with IDDM:
Diabetes Autoimmunity Study in the Young (DAISY). Diabetologia, 1996. 39(7):
p. 807-12.
110. Bonifacio, E., et al., IDDM1 and Multiple Family History of Type 1 Diabetes
Combine to Identify Neonates at High Risk for Type 1 Diabetes. Diabetes Care,
2004. 27(11): p. 2695-2700.
111. Kimpimaki, T., et al., The first signs of beta-cell autoimmunity appear in infancy
in genetically susceptible children from the general population: the Finnish
Type 1 Diabetes Prediction and Prevention Study. Journal of Clinical
Endocrinology & Metabolism, 2001. 86(10): p. 4782-8.
112. The Teddy Study Group, The Environmental Determinants of Diabetes in the
Young (TEDDY) study: study design. Pediatric Diabetes, 2007. 8(5): p. 286-298.
113. Wenzlau, J.M., et al., The cation efflux transporter ZnT8 (Slc30A8) is a major
autoantigen in human type 1 diabetes. Proceedings of the National Academy of
Sciences of the United States of America, 2007. 104: p. 17040–17045.
213
114. Andersson, C., et al., The three ZNT8 autoantibody variants together improve
the diagnostic sensitivity of childhood and adolescent type 1 diabetes.
Autoimmunity, 2011. 44(5): p. 394-405.
115. Yu, L., et al., Zinc Transporter-8 Autoantibodies Improve Prediction of Type 1
Diabetes in Relatives Positive for the Standard Biochemical Autoantibodies.
Diabetes Care, 2012.
116. Lindberg, B., et al., Islet autoantibodies in cord blood from children who
developed Type 1 (insulin-dependent) diabetes mellitus before 15 years of age.
Diabetologia, 1999. 42: p. 181-187.
117. Ziegler, A.G., et al., Autoantibody appearance and risk for development of
childhood diabetes in offspring of parents with type 1 diabetes: the 2-year
analysis of the German BABYDIAB Study. Diabetes, 1999. 48(3): p. 460-8.
118. Hummel, S. and A.G. Ziegler, Early determinants of type 1 diabetes: experience
from the BABYDIAB and BABYDIET studies. American Journal of Clinical
Nutrition, 2011. 94(6 Suppl): p. 1821S-1823S.
119. Yu, L., et al., Antiislet autoantibodies usually develop sequentially rather than
simultaneously. Journal of Clinical Endocrinology & Metabolism, 1996. 81(12):
p. 4264-7.
120. Kupila, A., et al., Genetic risk determines the emergence of diabetes-associated
autoantibodies in young children. Diabetes, 2002. 51(3): p. 646-51.
121. Long, A.E., et al., Rising Incidence of Type 1 Diabetes Is Associated With
Altered Immunophenotype at Diagnosis. Diabetes, 2012. 61(3): p. 683-686.
122. Bonifacio, E., et al., Quantification of islet-cell antibodies and prediction of
insulin-dependent diabetes. Lancet, 1990. 335(8682): p. 147-9.
214
123. Verge, C.F., et al., Anti-glutamate decarboxylase and other antibodies at the
onset of childhood IDDM: a population-based study. Diabetologia, 1994.
37(11): p. 1113-20.
124. Bonifacio, E. and P.J. Bingley, Islet autoantibodies and their use in predicting
insulin-dependent diabetes. Acta Diabetologica, 1997. 34(3): p. 185-93.
125. Maclaren, N., et al., Multiple autoantibodies as predictors of Type 1 diabetes in
a general population. Diabetologia, 2003. 46: p. 873-4 (Observation).
126. Orban, T., et al., Pancreatic islet autoantibodies as predictors of type 1 diabetes
in the Diabetes Prevention Trial-Type 1. Diabetes Care, 2009. 32(12): p. 2269-
74.
127. Hummel, M., et al., Brief communication: early appearance of islet
autoantibodies predicts childhood type 1 diabetes in offspring of diabetic
parents. Annals of Internal Medicine, 2004. 140(11): p. 882-6.
128. Barker, J.M., et al., Prediction of autoantibody positivity and progression to type
1 diabetes: Diabetes Autoimmunity Study in the Young (DAISY). Journal of
Clinical Endocrinology & Metabolism, 2004. 89(8): p. 3896-902.
129. Achenbach, P., et al., Stratification of type 1 diabetes risk on the basis of islet
autoantibody characteristics.[Erratum appears in Diabetes 2004; 53(4):1175-
6]. Diabetes, 2004. 53(2): p. 384-92.
130. Kulmala, P., et al., Prediction of insulin-dependent diabetes mellitus in siblings
of children with diabetes. A population-based study. The Childhood Diabetes in
Finland Study Group. Journal of Clinical Investigation, 1998. 101(2): p. 327-36.
215
131. Verge, C.F., et al., Prediction of type I diabetes in first-degree relatives using a
combination of insulin, GAD, and ICA512bdc/IA-2 autoantibodies. Diabetes,
1996. 45(7): p. 926-33.
132. Achenbach, P., E. Bonifacio, and A.-G. Ziegler, Predicting type 1 diabetes.
Current Diabetes Reports, 2005. 5(2): p. 98-103.
133. Achenbach, P., et al., Combined testing of antibody titer and affinity improves
insulin autoantibody measurement: Diabetes Antibody Standardization
Program. Clinical Immunology, 2007. 122(1): p. 85-90.
134. Mrena, S., et al., Models for Predicting Type 1 Diabetes in Siblings of Affected
Children. Diabetes Care, 2006. 29(3): p. 662-667.
135. Ziegler, A.-G. and G.T. Nepom, Prediction and Pathogenesis in Type 1
Diabetes. Immunity, 2010. 32(4): p. 468-478.
136. Atkinson, M.A. and G.S. Eisenbarth, Type 1 diabetes: new perspectives on
disease pathogenesis and treatment. Lancet, 2001. 358(9277): p. 221-229.
137. Devendra, D., E. Liu, and G.S. Eisenbarth, Type 1 diabetes: recent
developments. British Medical Journal, 2004. 328(7442): p. 750-754.
138. Kyvik, K.O., A. Green, and H. Beck-Nielsen, Concordance rates of insulin
dependent diabetes mellitus: a population based study of young Danish twins.
British Medical Journal, 1995. 311(7010): p. 913-7.
139. Redondo, M.J., et al., Concordance for Islet Autoimmunity among Monozygotic
Twins. New England Journal of Medicine, 2008. 359(26): p. 2849-2850.
140. Redondo, M.J., et al., Genetic determination of islet cell autoimmunity in
monozygotic twin, dizygotic twin, and non-twin siblings of patients with type
1 diabetes: prospective twin study. BMJ, 1999. 318(7185): p. 698-702.
216
141. Campbell-Stokes, P.L., B.J. Taylor, and on behalf of the New Zealand
Children's Diabetes Working Group, Prospective incidence study of diabetes
mellitus in New Zealand children aged 0 to 14 years. Diabetologia, 2005. 48: p.
643-648.
142. Larsson, K., et al., Genetic and perinatal factors as risk for childhood type 1
diabetes. Diabetes/Metabolism Research Reviews, 2004. 20(6): p. 429-37.
143. Redondo, M.J. and G.S. Eisenbarth, Genetic control of autoimmunity in Type I
diabetes and associated disorders. Diabetologia, 2002. 45(5): p. 605-22.
144. Bingley, P.J. and E.A. Gale, Genetics of diabetes. Lessons from family studies.
Baillieres Clinical Endocrinology & Metabolism, 1991. 5(2): p. 261-83.
145. Tillil, H. and J. Kobberling, Age-corrected empirical genetic risk estimates for
first-degree relatives of IDDM patients. Diabetes, 1987. 36(1): p. 93-9.
146. Warram, J.H., et al., Differences in risk of insulin-dependent diabetes in
offspring of diabetic mothers and diabetic fathers. New England Journal of
Medicine, 1984. 311(3): p. 149-52.
147. Pociot, F., et al., A nationwide population-based study of the familial
aggregation of type 1 (insulin-dependent) diabetes mellitus in Denmark. Danish
Study Group of Diabetes in Childhood. Diabetologia, 1993. 36(9): p. 870-5.
148. The EURODIAB ACE Study Group and and the EURODIAB ACE Substudy 2
Study Group, Familial risk of Type 1 diabetes in European Children.
Diabetologia, 1998. 41: p. 1151-1156.
149. Hamalainen, A.-M. and M. Knip, Autoimmunity and familial risk of type 1
diabetes. Current Diabetes Reports, 2002. 2(4): p. 347-53.
217
150. Harjutsalo, V., A. Reunanen, and J. Tuomilehto, Differential Transmission of
Type 1 Diabetes from Diabetic Fathers and Mothers to Their Offspring.
Diabetes, 2006. 55(5): p. 1517-1524.
151. Hemminki, K., et al., Familial association between type 1 diabetes and other
autoimmune and related diseases. Diabetologia, 2009. 52(9): p. 1820-8.
152. Gillespie, K.M., E.A.M. Gale, and P. Bingley, High Familial Risk and Genetic
Susceptibility in Early Onset Childhood Diabetes. Diabetes, 2002. 51: p. 210-
214.
153. Sjöberg, L., et al., Fertility in people with childhood-onset type 1 diabetes.
Diabetologia, 2013. 56(1): p. 78-81.
154. Stene, L.C. and E.A.M. Gale, The prenatal environment and type 1 diabetes.
Epub before print April 2013. Diabetologia, 2013.
155. Bonifacio, E., et al., Maternal type 1 diabetes reduces the risk of islet
autoantibodies: relationships with birthweight and maternal HbA(1c).
Diabetologia, 2008. 51(7): p. 1245-52.
156. Erlich, H., et al., HLA DR-DQ haplotypes and genotypes and type 1 diabetes
risk: analysis of the type 1 diabetes genetics consortium families. Diabetes,
2008. 57(4): p. 1084-92.
157. Pociot, F., et al., Genetics of type 1 diabetes: what's next? Diabetes, 2010. 59(7):
p. 1561-71.
158. Singal, D. and M. Blajchman, Histocompatibility (HL-A) antigens,
lymphocytotoxic antibodies and tissue antibodies in patients with diabetes
mellitus. Diabetes, 1973. 22: p. 429-432.
218
159. Gianani, R., et al., Limited loss of tolerance to islet autoantigens in ICA+ first
degree relatives of patients with type I diabetes expressing the HLA dqb1*0602
allele. Journal of Autoimmunity, 1996. 9(3): p. 423-5.
160. Pugliese, A., et al., HLA-DQB1*0602 is associated with dominant protection
from diabetes even among islet cell antibody-positive first-degree relatives of
patients with IDDM. Diabetes, 1995. 44(6): p. 608-13.
161. Schenker, M., et al., Early expression and high prevalence of islet
autoantibodies for DR3/4 heterozygous and DR4/4 homozygous offspring of
parents with Type I diabetes: the German BABYDIAB study. Diabetologia, 1999.
42(6): p. 671-7.
162. Knip, M., et al., Environmental triggers and determinants of type 1 diabetes.
Diabetes, 2005. 54([Suppl 2]): p. S125-36.
163. Lambert, A.P., et al., Absolute Risk of Childhood-Onset Type 1 Diabetes Defined
by Human Leukocyte Antigen Class II Genotype: A Population-Based Study in
the United Kingdom. Journal of Clinical Endocrinology and Metabolism, 2004.
89(8): p. 4037-4043.
164. Lipton, R.B., et al., HLA-DQ haplotypes differ by ethnicity in patients with
childhood-onset diabetes. Pediatric Diabetes, 2011. 12(4 Pt 2): p. 388-95.
165. Todd, J.A., From genome to aetiology in a multifactorial disease, type 1
diabetes. Bioessays, 1999. 21(2): p. 164-74.
166. Pociot, F., et al., Association of tumor necrosis factor (TNF) and class II major
histocompatibility complex alleles with the secretion of TNF-alpha and TNF-
beta by human mononuclear cells: a possible link to insulin-dependent diabetes
mellitus. European Journal of Immunology, 1993. 23(1): p. 224-31.
219
167. Bell, G.I., S. Horita, and J.H. Karam, A polymorphic locus near the human
insulin gene is associated with insulin-dependent diabetes mellitus. Diabetes,
1984. 33(2): p. 176-83.
168. Bottini, N., et al., Role of PTPN22 in type 1 diabetes and other autoimmune
diseases. Seminars in Immunology, 2006. 18(4): p. 207-13.
169. Field, L.L., Genetic linkage and association studies of Type I diabetes:
challenges and rewards. Diabetologia, 2002. 45(1): p. 21-35.
170. Barrett, J.C., et al., Genome-wide association study and meta-analysis find that
over 40 loci affect risk of type 1 diabetes. Nature Genetics, 2009. 41(6): p. 703-
7.
171. Bradfield, J.P., et al., A genome-wide meta-analysis of six type 1 diabetes
cohorts identifies multiple associated loci. PLoS Genetics, 2011. 7(9): p.
e1002293.
172. Morahan, G., Insights into type 1 diabetes provided by genetic analyses. Curr
Opin Endocrinol Diabetes Obes, 2012. 4: p. 263-70.
173. Todd, J.A., Etiology of Type 1 Diabetes. Immunity, 2010. 32(4): p. 457-467.
174. Noble, J.A. and H.A. Erlich, Genetics of Type 1 Diabetes. Cold Spring Harbor
Perspectives in Medicine, 2012. 2(1).
175. Pugliese, A., The Insulin Gene In Type 1 Diabetes. IUBMB Life, 2005. 57(7): p.
463-468.
176. Bakay, M., R. Pandey, and H. Hakonarson, Genes Involved in Type 1 Diabetes.
Type 1 Diabetes. 2013.
220
177. Bergholdt, R., et al., Identification of Novel Type 1 Diabetes Candidate Genes
by Integrating Genome-Wide Association Data, Protein-Protein Interactions,
and Human Pancreatic Islet Gene Expression. Diabetes, 2012. 61(4): p. 954-
962.
178. Steck, A.K., et al., Do non-HLA genes influence development of persistent islet
autoimmunity and type 1 diabetes in children with high-risk HLA-DR,DQ
genotypes? Diabetes, 2009. 58(4): p. 1028-33.
179. Steck, A.K., et al., Effects of Non-HLA Gene Polymorphisms on Development of
Islet Autoimmunity and Type 1 Diabetes in a Population With High-Risk HLA-
DR,DQ Genotypes. Diabetes, 2012. 61(3): p. 753-758.
180. Lempainen, J., et al., Effect of the PTPN22 and INS Risk Genotypes on the
Progression to Clinical Type 1 Diabetes After the Initiation of β-Cell
Autoimmunity. Diabetes, 2012. 61(4): p. 963-966.
181. Hermann, R., et al., Temporal changes in the frequencies of HLA genotypes in
patients with Type 1 diabetes - indication of an increased environmental
pressure? Diabetologia, 2003. 46(3): p. 420-425.
182. Fourlanos, S., et al., The Rising Incidence of Type 1 Diabetes Is Accounted for
by Cases With Lower-Risk Human Leukocyte Antigen Genotypes. Diabetes Care,
2008. 31(8): p. 1546.
183. Gillespie, K., et al., The rising incidence of childhood type 1 diabetes and
reduced contribution of high-risk HLA haplotypes. Lancet, 2004. 364: p. 1699-
700 (Letter).
221
184. Steck, A.K., et al., Stepwise or linear decrease in penetrance of type 1 diabetes
with lower-risk HLA genotypes over the past 40 years. Diabetes, 2011. 60(3): p.
1045-9.
185. Vehik, K. and D. Dabelea, The changing epidemiology of type 1 diabetes: why is
it going through the roof? Diabetes/Metabolism Research and Reviews, 2010.
27(1): p. 3-13.
186. Feltbower, R.G., et al., Trends in the incidence of childhood diabetes in south
Asians and other children in Bradford, UK. Diabetic Medicine, 2002. 19(2): p.
162-6.
187. Yang, Z., et al., Childhood diabetes in China. Enormous variation by place and
ethnic group. Diabetes Care, 1998. 21(4): p. 525-529.
188. Ronningen, K.S., N. Keiding, and A. Green, Correlations between the incidence
of childhood-onset Type I diabetes in Europe and HLA genotypes. Diabetologia,
2001. 44([Suppl 3]): p. B51-9.
189. Cherubini, V., et al., Large incidence variation of Type 1 diabetes in central-
southern Italy 1990-1995: lower risk in rural areas. Diabetologia, 1999. 42: p.
789-792.
190. Pundziute-Lycka, A., et al., Incidence of type 1 diabetes in Lithuanians aged 0-
39 years varies by the urban-rural setting, and the time change differs for men
and women during 1991-2000. Diabetes Care, 2003. 26: p. 671-676.
191. Staines, A., et al., Small area variation in the incidence of childhood insulin-
dependent diabetes mellitus in Yorkshire, UK: links with overcrowding and
population density. International Journal of Epidemiology, 1997. 26(6): p. 1307-
1313.
222
192. Rytkonen, M., et al., The incidence of type 1 diabetes among children in Finland
- rural-urban difference. Health & Place, 2003. 9: p. 315-325.
193. Patterson, C., et al., Epidemiology of childhood IDDM in Northern Ireland
1989-1994: Low incidence in areas with highest population density and most
household crowding. Diabetologia, 1996. 39: p. 1063-1069.
194. Waldhoer, T., et al., Changing geographical distribution of diabetes mellitus
type 1 incidence in Austrian children 1989-2005. European Journal of
Epidemiology, 2008. 23: p. 213-218.
195. Cardwell, C.R., D.J. Carson, and C.C. Patterson, Higher incidence of childhood-
onset type 1 diabetes mellitus in remote areas: a UK regional small-area
analysis. Diabetologia, 2006. 49: p. 2074-2077.
196. Kukko, M., et al., Geographical Variation in Risk HLA-DQB1 Genotypes for
Type 1 Diabetes and Signs of beta-Cell Autoimmunity in a High-Incidence
Country. Diabetes Care, 2004. 27(3): p. 676-681.
197. Hathout, E.H., et al., Air pollution and type 1 diabetes in children. Pediatric
Diabetes, 2006. 7(2): p. 81-7.
198. Parslow, R.C., et al., Incidence of childhood diabetes mellitus in Yorkshire,
northern England, is associated with nitrate in drinking water: an ecological
analysis. Diabetologia, 1997. 40(5): p. 550-6.
199. Tarn, A.C., et al., Diabetes and social class. Lancet, 1983. 2: p. 631-32 (Letter).
200. LaPorte, R., et al., The Pittsburgh Insulin Dependent Diabetes Mellitus Registry:
the relationship of insulin dependent diabetes mellitus incidence to social class.
American Journal of Epidemiology, 1981. 114(3): p. 379-384.
223
201. Debono, J., C. Johnson, and P. Betts, Juvenile diabetes and social class. Lancet,
1983. 1: p. 1113-14.
202. Torres-Aviles, F., et al., Clustering of cases of type 1 diabetes in high
socioeconomic communes in Santiago de Chile: spatio-temporal and
geographical analysis. Acta Diabetologica, 2010. 47(3): p. 251-7.
203. Sipetic, S., et al., The Belgrade childhood diabetes study: prenatal and social
associations for type 1 diabetes. Paediatric and Perinatal Epidemiology, 2004.
18: p. 33-39.
204. Patterson, C., et al., Is childhood-onset type 1 diabetes a wealth-related disease?
An ecological analysis of European incidence rates. Diabetologia, 2001.
44([Suppl 3]): p. B9-16.
205. Gyurus, E., et al., Dynamic changes in the trends in incidence of type 1 diabetes
in children in Hungary (1978-98). Pediatric Diabetes, 2002. 3: p. 194-199.
206. Long, A.E. and P.J. Bingley, The epidemiology of childhood diabetes.
Paediatrics and Child Health, 2009. 19(7): p. 304-308.
207. Strachan DP, Hay fever, hygeine and household size. British Medical Journal,
1989. 299: p. 1259-1260.
208. Bresciani, M.a., et al., The hygiene hypothesis: does it function worldwide?
Current Opinion in Allergy & Clinical Immunology, 2005. 5(2): p. 147-151.
209. Gale, E.A.M., A missing link in the hygiene hypothesis? Diabetologia, 2002. 45:
p. 588-594.
210. Guarner, F., et al., Mechanisms of Disease: the hygiene hypothesis revisited.
Nature Clinical Practice. Gastroenterology & Hepatology, 2006. 3(5): p. 275-
284.
224
211. Okada, H., et al., The 'hygiene hypothesis' for autoimmune and allergic diseases:
an update. Clinical & Experimental Immunology, 2010. 160(1): p. 1-9.
212. Wiess, S.T., Eat dirt - the hygeine hypothesis and allergic diseases. New
England Journal of Medicine, 2002. 347: p. 930-2.
213. Bach, J.-F., The Effect of Infections on Susceptibility to Autoimmune and
Allergic Diseases. New England Journal of Medicine, 2002. 347(12): p. 911-
920.
214. Kolb, H. and R. Elliott, Increasing incidence of IDDM a consequence of
improved hygiene? Diabetologia, 1994. 37: p. 729 (Letter).
215. Cardwell, C., et al., Atopy, home environment and the risk of childhood-onset
type 1 diabetes: a population-based case-control study. Pediatric Diabetes,
2008. 9: p. 191-196.
216. The EURODIAB Substudy 2 Study group, Infections and vaccinations as risk
factors for childhood type I (insulin-dependent) diabetes mellitus: a multicentre
case-control investigation. EURODIAB Substudy 2 Study Group. Diabetologia,
2000. 43(1): p. 47-53.
217. Wraith, D.C., M. Goldman, and P.H. Lambert, Vaccination and autoimmune
disease: what is the evidence? Lancet, 2003. 362(9396): p. 1659-66.
218. Shoenfeld, Y., Infections, vaccines and autoimmunity. Lupus, 2009. 18(13): p.
1127-8.
219. Graves, P.M., et al., Lack of association between early childhood immunizations
and beta-cell autoimmunity. Diabetes Care, 1999. 22(10): p. 1694-7.
225
220. Hummel, M., et al., No major association of breast-feeding, vaccinations, and
childhood viral diseases with early islet autoimmunity in the German
BABYDIAB Study. Diabetes Care, 2000. 23: p. 969-974.
221. DeStefano, F., et al., Childhood vaccinations, vaccination timing, and risk of
type 1 diabetes mellitus. Pediatrics, 2001. 108(6): p. E112.
222. Hviid, A., et al., Childhood vaccination and type 1 diabetes. New England
Journal of Medicine, 2004. 350(14): p. 1398-404.
223. Moltchanova, E.V., et al., Seasonal variation of diagnosis of Type 1 diabetes
mellitus in children worldwide. Diabetic Medicine, 2009. 26(7): p. 673-8.
224. Karvonen, M., et al., Seasonality in the Clinical Onset of Insulin-dependent
Diabetes Mellitus in Finnish Children. American Journal of Epidemiology,
1996. 143: p. 167-76.
225. Elliott, J.C., et al., Population density determines the direction of the association
between ambient ultraviolet radiation and type 1 diabetes incidence. Pediatric
Diabetes, 2010. 11(6): p. 394-402.
226. Craighead, J.E., Does insulin dependent diabetes mellitus have a viral etiology?
Human Pathology, 1979. 10(3): p. 267-78.
227. Gamble, D., et al., Viral antibodies in diabetes mellitus. BMJ, 1969. 3: p. 627.
228. Gunby, P., Viral link sought in juvenile diabetes. JAMA: The Journal of the
American Medical Association, 1978. 240(12): p. 1219-25.
229. Huff, J.C., J.C. Hierholzer, and W.A. Farris, An "outbreak" of juvenile diabetes
mellitus: consideration of a viral etiology. American Journal of Epidemiology,
1974. 100(4): p. 277-87.
226
230. Menser, M.A., J.M. Forrest, and R.D. Bransby, Rubella infection and diabetes
mellitus. Lancet, 1978. 1(8055): p. 57-60.
231. Gale, E.A.M., Congenital rubella: citation virus or viral cause of type 1
diabetes? Diabetologia, 2008. 51(9): p. 1559-66.
232. Ward, K., W. Galloway, and I. Auchterlonie, Congenital cytomegalovirus
infection and diabetes. Lancet, 1979. 1: p. 497.
233. Sultz, H.A., et al., Is mumps virus an etiologic factor in juvenile diabetes
mellitus. Journal of Pediatrics, 1975. 86(4): p. 654-6.
234. Honeyman, M.C., et al., Association between rotavirus infection and pancreatic
islet autoimmunity in children at risk of developing type 1 diabetes. Diabetes,
2000. 49(8): p. 1319-24.
235. Couper, J.J., Environmental triggers of type 1 diabetes. Journal of Paediatrics &
Child Health, 2001. 37(3): p. 218-20.
236. Gamble, D.R., K.W. Taylor, and H. Cumming, Coxsackie viruses and diabetes
mellitus. British Medical Journal, 1973. 4(5887): p. 260-2.
237. Roivainen, M., et al., Several different enterovirus serotypes can be associated
with prediabetic autoimmune episodes and onset of overt IDDM. Childhood
Diabetes in Finland (DiMe) Study Group. Journal of Medical Virology, 1998.
56(1): p. 74-8.
238. Wagenknecht, L., J. Reoseman, and W. Herman, Increased incidence of insulin-
dependent diabetes mellitus following an epidemic of coxsackievirus B5.
American Journal of Epidemiology, 1991. 133: p. 1024-1031.
227
239. Lönnrot, M., et al., Enterovirus infection as a risk factor for beta-cell
autoimmunity in a prospectively observed birth cohort: the Finnish Diabetes
Prediction and Prevention Study. Diabetes, 2000. 49(8): p. 1314-1318.
240. Lonnrot, M., et al., Enterovirus RNA in serum is a risk factor for beta-cell
autoimmunity and clinical type 1 diabetes: a prospective study. Childhood
Diabetes in Finland (DiMe) Study Group. Journal of Medical Virology, 2000.
61(2): p. 214-20.
241. Oikarinen, M., et al., Type 1 Diabetes Is Associated With Enterovirus Infection
in Gut Mucosa. Diabetes, 2012. 61(3): p. 687-691.
242. Oikarinen, S., et al., Enterovirus RNA in Blood Is Linked to the Development of
Type 1 Diabetes. Diabetes, 2011. 60(1): p. 276.
243. Salminen, K., et al., Enterovirus infections are associated with the induction of
beta-cell autoimmunity in a prospective birth cohort study. Journal of Medical
Virology, 2003. 69(1): p. 91-8.
244. Simonen-Tikka, M.L., et al., Human enterovirus infections in children at
increased risk for type 1 diabetes: the BABYDIET study. Diabetologia, 2011.
54(12): p. 2995-3002.
245. Stene, L.C., et al., Enterovirus Infection and Progression From Islet
Autoimmunity to Type 1 Diabetes. Diabetes, 2010. 59(12): p. 3174-3180.
246. Tauriainen, S., et al., Enteroviruses in the pathogenesis of type 1 diabetes.
Seminars In Immunopathology, 2011. 33(1): p. 45-55.
247. Craig, M.E., et al., Reduced frequency of HLA DRB1*03-DQB1*02 in children
with type 1 diabetes associated with enterovirus RNA. Journal of Infectious
Diseases, 2003. 187(10): p. 1562-70.
228
248. Yeung, W.-C.G., W.D. Rawlinson, and M.E. Craig, Enterovirus infection and
type 1 diabetes mellitus: systematic review and meta-analysis of observational
molecular studies. British Medical Journal, 2011. 342.
249. Graves, P.M., et al., Prospective study of enteroviral infections and development
of beta-cell autoimmunity. Diabetes Autoimmunity Study in the Young (DAISY).
Diabetes Research & Clinical Practice, 2003. 59(1): p. 51-61.
250. Tapia, G., et al., Human Enterovirus RNA in Monthly Fecal Samples and Islet
Autoimmunity in Norwegian Children With High Genetic Risk for Type 1
Diabetes: The MIDIA study. Diabetes Care, 2011. 34(1): p. 151.
251. Harkonen, T., et al., Enterovirus infection may induce humoral immune response
reacting with islet cell autoantigens in humans. Journal of Medical Virology,
2003. 69(3): p. 426-40.
252. Honeyman, M.C., et al., Evidence for molecular mimicry between human T cell
epitopes in rotavirus and pancreatic islet autoantigens. Journal of Immunology,
2010. 184(4): p. 2204-10.
253. Elshebani, A., et al., Effects on isolated human pancreatic islet cells after
infection with strains of enterovirus isolated at clinical presentation of type 1
diabetes. Virus Research, 2007. 124(1-2): p. 193-203.
254. Dotta, F., et al., Coxsackie B4 virus infection of beta cells and natural killer cell
insulitis in recent-onset type 1 diabetic patients. Proceedings of the National
Academy of Sciences of the United States of America, 2007. 104(12): p. 5115-
20.
255. Filippi, C.M. and M.G. von Herrath, Viral Trigger for Type 1 Diabetes.
Diabetes, 2008. 57(11): p. 2863-2871.
229
256. Hyoty, H. and K. Taylor, The role of viruses in human diabetes. Diabetologia,
2002. 45: p. 11353-1361.
257. Viskari, H., et al., Relationship between the incidence of type 1 diabetes and
enterovirus infections in different European populations: results from the
EPIVIR project. Journal of Medical Virology, 2004. 72(4): p. 610-7.
258. Sadeharju, K., et al., The HLA-DR phenotype modulates the humoral immune
response to enterovirus antigens. Diabetologia, 2003. 46(8): p. 1100-5.
259. Round, J.L. and S.K. Mazmanian, The gut microbiota shapes intestinal immune
responses during health and disease. Nat Rev Immunol, 2009. 9: p. 313-23.
260. Vaarala, O., Is the origin of type 1 diabetes in the gut[quest]. Immunol Cell
Biol, 2012. 90(3): p. 271-276.
261. Giongo, A., et al., Toward defining the autoimmune microbiome for type 1
diabetes. The Isme Journal, 2011. 5(1): p. 82-91.
262. Atkinson, M. and A.V. Chervonsky, Does the gut microbiota have a role in type
1 diabetes? Early evidence from humans and animal models of the disease.
Diabetologia 2012. 55(11): p. 2868-2877
263. Mathis, D. and C. Benoist, Microbiota and Autoimmune Disease: The Hosted
Self. Cell Host & Microbe, 2011. 10(4): p. 297-301.
264. Ljungberg, M., et al., Probiotics for the prevention of beta cell autoimmunity in
children at genetic risk of type 1 diabetes--the PRODIA study. Annals of the
New York Academy of Sciences, 2006. 1079: p. 360-4.
265. Boerner, B.P. and N.E. Sarvetnick, Type 1 diabetes: role of intestinal
microbiome in humans and mice. Annals of the New York Academy of
Sciences, 2011. 1243: p. 103-18.
230
266. Patrick, C., et al., Promotion of Autoimmune Diabetes by Cereal Diet in the
Presence or Absence of Microbes Associated With Gut Immune Activation,
Regulatory Imbalance, and Altered Cathelicidin Antimicrobial Peptide.
Diabetes, 2013. 62(6): p. 2036-2047.
267. Jones, R., P. Ford, and R. Hamman Seasonality comparisons among groups
using incidence data. Biometrics, 1988. 44(1131-44).
268. Rosenbauer, J., et al., Temporal, seasonal and geographical incidence patterns
of Type1 diabetes mellitus in children under 5 years of age in Germany.
Diabetologia, 1999. 42: p. 1055-1059.
269. Norris, J.M., Can the sunshine vitamin shed light on type 1 diabetes? Lancet,
2001. 358: p. 1476-8 (Commentary).
270. Hyppönen, E., et al., Intake of vitamin D and risk of type 1 diabetes: a birth-
cohort study. Lancet, 2001. 358(9292): p. 1500-1503.
271. Bener, A., et al., High prevalence of vitamin D deficiency in type 1 diabetes
mellitus and healthy children. Acta Diabetologica, 2009. 46(3): p. 183-9.
272. Janner, M., et al., High prevalence of vitamin D deficiency in children and
adolescents with type 1 diabetes. Swiss Medical Weekly, 2010. 140: p. w13091.
273. Pozzilli, P., et al., Low levels of 25-hydroxyvitamin D3 and 1,25-
dihydroxyvitamin D3 in patients with newly diagnosed type 1 diabetes. Hormone
& Metabolic Research, 2005. 37(11): p. 680-3.
274. Svoren, B.M., et al., Significant vitamin D deficiency in youth with type 1
diabetes mellitus. Journal of Pediatrics, 2009. 154(1): p. 132-4.
231
275. The EURODIAB Substudy 2 Study group, Vitamin D supplement in early
childhood and risk for type 1 (insulin-dependent) diabetes mellitus.
Diabetologia, 1999. 42: p. 51–54.
276. Stene, L.C., G. Joner, and the Norwegian Childhood Diabetes Study Group, Use
of cod liver oil during the first year of life is associated with lower risk of
childhood-onset type 1 diabetes: a large, population-based, case-control study.
American Journal of Clinical Nutrition, 2003. 78(6): p. 1128-1134.
277. Simpson, M., et al., No association of vitamin D intake or 25-hydroxyvitamin D
levels in childhood with risk of islet autoimmunity and type 1 diabetes: the
Diabetes Autoimmunity Study in the Young (DAISY). Diabetologia, 2011.
54(11): p. 2779-2788.
278. Marjamäki, L., et al., Maternal intake of vitamin D during pregnancy and risk of
advanced beta cell autoimmunity and type 1 diabetes in offspring. Diabetologia,
2010. 53(8): p. 1599-1607.
279. Sørensen, I.M., et al., Maternal Serum Levels of 25-Hydroxy-Vitamin D During
Pregnancy and Risk of Type 1 Diabetes in the Offspring. Diabetes, 2012. 61(1):
p. 175-178.
280. Hyppönen, E., Vitamin D and increasing incidence of type 1 diabetes—evidence
for an association? Diabetes, Obesity and Metabolism, 2010. 12(9): p. 737-743.
281. Mohr, S.B., et al., Is There a Role of Vitamin D Deficiency in Type 1 Diabetes of
Children? American Journal of Preventive Medicine, 2010. 39(2): p. 189-190.
282. Mathieu, C., et al., Vitamin D and diabetes. Diabetologia, 2005. 48(7): p. 1247-
1257.
232
283. Guo, S.-W., et al., Meta-analysis of vitamin D receptor polymorphisms and type
1 diabetes: a HuGE review of genetic association studies. American Journal of
Epidemiology, 2006. 164(8): p. 711-24.
284. Blanton, D., et al., Reduced serum vitamin D-binding protein levels are
associated with type 1 diabetes. Diabetes, 2011. 60(10): p. 2566-70.
285. Bailey, R., et al., Association of the vitamin D metabolism gene CYP27B1 with
type 1 diabetes. Diabetes, 2007. 56(10): p. 2616-21.
286. Cooper, J.D., et al., Inherited variation in vitamin D genes is associated with
predisposition to autoimmune disease type 1 diabetes. Diabetes, 2011. 60(5): p.
1624-31.
287. Mathieu, C., Vitamin D and diabetes: the devil is in the D-tails. Diabetologia,
2010. 53(8): p. 1545-1548.
288. Virtanen, S.M. and M. Knip, Nutritional risk predictors of beta-cell
autoimmunity and type 1 diabetes at a young age. American Journal of Clinical
Nutrition, 2003. 78(6): p. 1053-1067.
289. Rosenbauer, J., et al., Early nutrition and risk of Type 1 diabetes mellitus--a
nationwide case-control study in preschool children. Experimental & Clinical
Endocrinology & Diabetes, 2007. 115(8): p. 502-8.
290. Knip, M., S.M. Virtanen, and H.K. Akerblom, Infant feeding and the risk of type
1 diabetes. American Journal of Clinical Nutrition, 2010. 91(5): p. 1506S-
1513S.
291. Borch-Johnsen, K., et al., Relation between breast-feeding and incidence rates
of insulin-dependent diabetes mellitus. A hypothesis. Lancet, 1984. 2(8411): p.
1083-6.
233
292. Mayer, E.J., et al., Reduced risk of IDDM among breast-fed children. The
Colorado IDDM Registry. Diabetes, 1988. 37(12): p. 1625-32.
293. Sadauskait÷-Kuehne, V., et al., Longer breastfeeding is an independent
protective factor against development of type 1 diabetes mellitus in childhood.
Diabetes/Metabolism Research and Reviews, 2004. 20(2): p. 150-157.
294. Norris, J.M., et al., Lack of association between early exposure to cow's milk
protein and beta-cell autoimmunity. Diabetes Autoimmunity Study in the Young
(DAISY). JAMA: The Journal of the American Medical Association, 1996.
276(8): p. 609-14.
295. Couper, J.J., et al., Lack of association between duration of breast-feeding or
introduction of cow's milk and development of islet autoimmunity. Diabetes,
1999. 48(11): p. 2145-9.
296. Ziegler, A.-G., et al., Early infant feeding and risk of developing type 1 diabetes-
associated autoantibodies. JAMA: The Journal of the American Medical
Association, 2003. 290(13): p. 1721-8.
297. Norris, J.M., et al., Timing of initial cereal exposure in infancy and risk of islet
autoimmunity. JAMA: The Journal of the American Medical Association, 2003.
290(13): p. 1713-20.
298. Vaarala, O., et al., Cow's milk formula feeding induces primary immunization to
insulin in infants at genetic risk for type 1 diabetes. Diabetes, 1999. 48(7): p.
1389-94.
299. The TRIGR Study Group, The Trial to Reduce IDDM in the Genetically at Risk
(TRIGR) study: recruitment, intervention and follow-up. Diabetologia, 2011.
54(3): p. 627-633.
234
300. TRIGR Study Group, Study design of the Trial to Reduce IDDM in the
Genetically at Risk (TRIGR). Pediatric Diabetes, 2007. 8(3): p. 117-37.
301. Knip, M., et al., Dietary intervention in infancy and later signs of beta-cell
autoimmunity. New England Journal of Medicine, 2010. 363(20): p. 1900-8.
302. Schmid, S., et al., BABYDIET, a feasibility study to prevent the appearance of
islet autoantibodies in relatives of patients with Type 1 diabetes by delaying
exposure to gluten. Diabetologia, 2004. 47(6): p. 1130-1.
303. Hummel, S., et al., Primary dietary intervention study to reduce the risk of islet
autoimmunity in children at increased risk for type 1 diabetes: the BABYDIET
study. Diabetes Care, 2011. 34(6): p. 1301-5.
304. Chase, H.P., et al., Nutritional Intervention to Prevent (NIP) Type 1 Diabetes - A
Pilot Trial. Infant, Child, & Adolescent Nutrition, 2009. 1(2): p. 98-107.
305. Coughlan, M.T., et al., Advanced Glycation End Products Are Direct
Modulators of β-Cell Function. Diabetes, 2011. 60(10): p. 2523-2532.
306. Forbes, J.M., et al., Receptor for advanced glycation end-products (RAGE)
provides a link between genetic susceptibility and environmental factors in type
1 diabetes. Diabetologia, 2011. 54(5): p. 1032-1042.
307. Atkinson, M. and E.A.M. Gale, Infant Diets and Type 1 Diabetes: too early, too
late or just too complicated? JAMA: The Journal of the American Medical
Association, 2003. 290(13): p. 1771-1772.
308. Norris, J.M., Infant and childhood diet and type 1 diabetes risk: recent advances
and prospects. Current Diabetes Reports, 2010. 10(5): p. 345-9.
235
309. Verbeeten, K.C., et al., Association between childhood obesity and subsequent
Type 1 diabetes: a systematic review and meta-analysis. Diabetic Medicine,
2011. 28(1): p. 10-8.
310. Magarey, A., L. Daniels, and J. Boulton, Prevalence of overweight and obesity
in Australian children and adolescents: reassessment of 1985 and 1995 data
against new standard international definitions. Medical Journal of Australia,
2001. 174: p. 561-4.
311. Norton, K., et al., Descriptive epidemiology of childhood overweight and obesity
in Australia: 1901–2003. International Journal of Pediatric Obesity, 2006. 1(4):
p. 232-238.
312. Evans, J.M., et al., Socio-economic status, obesity and prevalence of Type 1 and
Type 2 diabetes mellitus. Diabetic Medicine, 2000. 17(6): p. 478-80.
313. Liu, L.L., et al., Prevalence of overweight and obesity in youth with diabetes in
USA: the SEARCH for Diabetes in Youth Study. Pediatric Diabetes, 2010. 11(1):
p. 4-11.
314. Libman, I., et al., Changing prevalence of overweight children and adolescents
at onset of insulin-treated diabetes. Diabetes Care, 2003. 26(10): p. 2871-2875.
315. Dahlquist, G., Can we slow the rising incidence of childhood-onset autoimmune
diabetes? The overload hypothesis. Diabetologia, 2006. 49(1): p. 20-24.
316. Knip, M., et al., Does the secular increase in body mass in children contribute to
the increasing incidence of type 1 diabetes? Pediatric Diabetes, 2008. 9(3 Pt 2):
p. 46-9.
236
317. Waldhör, T. and E. Schober, Regional distribution of risk for childhood diabetes
in Austria and possible association with body mass index. European Journal of
Pediatrics, 2003. 162(6): p. 380-384.
318. EURODIAB Substudy 2 Study Group, Rapid early growth is associated with
increased risk of childhood type 1 diabetes in various European populations.
Diabetes Care, 2002. 25(10): p. 1755.
319. Lamb, M.M., et al., Height growth velocity, islet autoimmunity and type 1
diabetes development: the Diabetes Autoimmunity Study in the Young.
Diabetologia, 2009. 52(10): p. 2064-71.
320. Pundziute-Lycka, A., et al., Diet, Growth, and the Risk for Type 1 Diabetes in
Childhood: A matched case-referent study Diabetes Care, 2004. 27(12): p. 2784-
2789.
321. Hypponen, E., et al., Infant feeding, early weight gain, and risk of type 1
diabetes. Childhood Diabetes in Finland (DiMe) Study Group. Diabetes Care,
1999. 22(12): p. 1961-1965.
322. Hypponen, E., et al., Obesity, increased linear growth and risk of type 1 diabetes
in children. Diabetes Care, 2000. 23: p. 1755-1760.
323. Blom, L., L.A. Persson, and G. Dahlquist, A high linear growth is associated
with an increased risk of childhood diabetes mellitus. Diabetologia, 1992. 35(6):
p. 528-33.
324. Couper, J.J., et al., Weight Gain in Early Life Predicts Risk of Islet
Autoimmunity in Children With a First-Degree Relative With Type 1 Diabetes.
Diabetes Care, 2009. 32(1): p. 94-99.
237
325. Harder, T., et al., Birth Weight, Early Weight Gain, and Subsequent Risk of Type
1 Diabetes: Systematic Review and Meta-Analysis. American Journal of
Epidemiology, 2009. 169(12): p. 1428-1436.
326. Dahlquist, G.G., et al., Dietary factors and the risk of developing insulin
dependent diabetes in childhood. British Medical Journal, 1990. 300(6735): p.
1302-6.
327. Wilkin, T., The accelerator hypothesis: weight gain as the missing link between
Type 1 and Type 2 diabetes. Diabetologia, 2001. 44: p. 914-922.
328. Betts, P., et al., Increasing body weight predicts the earlier onset of insulin-
dependant diabetes in childhood: testing the 'accelerator hypothesis' (2).
Diabetic Medicine, 2005. 22(2): p. 144-151.
329. Clarke, S.L., et al., Increased Adiposity at Diagnosis in Younger Children With
Type 1 Diabetes Does Not Persist. Diabetes Care, 2006. 29(7): p. 1651-1653.
330. Kibirige, M., et al., Testing the Accelerator Hypothesis. The relationship
between body mass and age at diagnosis of type 1 diabetes. Diabetes Care,
2003. 26(10): p. 2865-2870.
331. Knerr, I., et al., The ‘accelerator hypothesis’: relationship between weight,
height, body mass index and age at diagnosis in a large cohort of 9,248 German
and Austrian children with type 1 diabetes mellitus. Diabetologia, 2005. 48(12):
p. 2501-2504.
332. Fourlanos, S., et al., Insulin resistance is a risk factor for progression to Type 1
diabetes. Diabetologia, 2004. 47(10): p. 1661-1667.
238
333. O'Connell, M.A., S. Donath, and F.J. Cameron, Major increase in Type 1
diabetes: no support for the Accelerator Hypothesis. Diabetic Medicine, 2007.
24(8): p. 920-3.
334. Porter, J.R. and T.G. Barrett, Braking the accelerator hypothesis? Diabetologia,
2004. 47(2): p. 352-3.
335. Gale, E., To boldly go—or to go too boldly? The accelerator hypothesis
revisited. Diabetologia, 2007. 50(8): p. 1571-1575.
336. Fourlanos, S., L.C. Harrison, and P.G. Colman, The accelerator hypothesis and
increasing incidence of type 1 diabetes. Current Opinion in Endocrinology,
Diabetes & Obesity, 2008. 15(4): p. 321-5.
337. Wilkin, T., Comment on: Gale EAM (2007) To boldly go—or to go too boldly?
The accelerator hypothesis revisited. Diabetologia 50:1571–1575—a reply to
the editor. Diabetologia, 2007. 50(12): p. 2604-2606.
338. Wilkin, T.J., The accelerator hypothesis: a review of the evidence for insulin
resistance as the basis for type I as well as type II diabetes. International Journal
of Obesity, 2009. 33(7): p. 716-726.
339. McKinney, P.A., et al., Antenatal risk factors for childhood diabetes mellitus; a
case-control study of medical record data in Yorkshire, UK. Diabetologia, 1997.
40: p. 933-939.
340. Patterson, C.C., et al., A case-control investigation of perinatal risk factors for
childhood IDDM in Northern Ireland and Scotland. Diabetes Care, 1994. 17: p.
376-381.
239
341. Bache, I., et al., Previous maternal abortion, longer gestation and younger
maternal age decrease the risk of Type 1 diabetes in male offspring. Diabetes
Care, 1999. 22: p. 1063-1065.
342. Stene, L., et al., Maternal and paternal age at delivery, birth order, and risk of
childhood onset type 1 diabetes: population based cohort study. British Medical
Journal, 2001. 323: p. 369-71.
343. Gale, E.A.M., Maternal age and diabetes in childhood. British Medical Journal,
2010. 340.
344. Cardwell, C.R., et al., Maternal age at birth and childhood type 1 diabetes: a
pooled analysis of 30 observational studies. Diabetes, 2010. 59(2): p. 486-94.
345. Stene, L., et al., Birth weight and childhood onset type 1 diabetes: population
based cohort study. British Medical Journal, 2001. 322: p. 889-92.
346. Cardwell, C.R., D.J. Carson, and C.C. Patterson, Parental age at delivery, birth
order, birth weight and gestational age are associated with the risk of childhood
Type 1 diabetes: a UK regional retrospective cohort study. Diabet Med, 2005.
22(2): p. 200-206.
347. Dahlquist, G.G., et al., Birthweight and risk of type 1 diabetes in children and
young adults: a population-based register study. Diabetologia, 2005. 48(6): p.
1114-1117.
348. Stenhouse, E., H.X. Zhao, and B.A. Millward, Birthweight is not a risk factor
for childhood onset Type 1 diabetes mellitus. Diabetic Medicine, 2000. 17(7): p.
552-552.
349. Kyvik, K.O., et al., No association between birth weight and Type 1 diabetes
mellitus - a twin-control study. Diabetic Medicine, 2000. 17(2): p. 158-162.
240
350. Cardwell, C.R., et al., Birthweight and the risk of childhood-onset type 1
diabetes: a meta-analysis of observational studies using individual patient data.
Diabetologia, 2010. 53(4): p. 641-51.
351. Dahlquist, G., et al., Perinatal risk factors for childhood type 1 diabetes in
Europe: the EURODIAB Substudy 2 Study Group. Diabetes Care, 1999. 22: p.
1698-1702.
352. Wadsworth, E.J.K., et al., A case-control study of environmental factors
associated with diabetes in the under 5s. Diabetic Medicine, 1997. 14: p. 390-
396.
353. Jones, M., et al., Pre-natal and early life risk factors for childhood onset
diabetes mellitus: a record linkage study. International Journal of Epidemiology,
1998. 27(3): p. 444-449.
354. Cardwell, C.R., et al., Birth order and childhood type 1 diabetes risk: a pooled
analysis of 31 observational studies. International Journal of Epidemiology,
2011. 40(2): p. 363-374.
355. McKinney, P.A., et al., Perinatal and neonatal determinants of childhood type 1
diabetes: a case-control study in Yorkshire, UK. Diabetes Care, 1999. 22: p.
928-932.
356. Cardwell, C.R., et al., Interbirth Interval Is Associated With Childhood Type 1
Diabetes Risk. Diabetes, 2012. 61(3): p. 702-707.
357. Dunger, D.B., et al., Association of the INS VNTR with size at birth. Nat Genet,
1998. 19(1): p. 98-100.
358. McCarthy, M., Weighing in on diabetes risk. Nature Genetics, 1998. 19: p. 209-
210.
241
359. Larsson, H.E., et al., Diabetes-associated HLA genotypes affect birthweight in
the general population. Diabetologia, 2005. 48(8): p. 1484-1491.
360. Geoscience Australia, Area of States and Territories. Accessed 5 November
2011 from
http://web.archive.org/web/20080730134442/http://www.ga.gov.au/education/fa
cts/dimensions/areadime.htm, 2005.
361. Bureau of Meteorology and Commonwealth of Australia, Climate of Western
Australia. Accessed 5 November 2011 from
http://www.bom.gov.au/lam/climate/levelthree/ausclim/ausclimwa.htm, 2011.
362. Australian Bureau of Statistics, Population by Age and Sex, Regions of
Australia, 2010. Catalogue No. 3235.0, Canberra: Australian Government
Publishing Service, 2011.
363. Australian Bureau of Statistics, Western Australia at a glance. Accessed 5
November 2011 from
http://www.ausstats.abs.gov.au/ausstats/subscriber.nsf/0/EFF65DC1D23D314F
CA25741D000DC292/$File/13065_2008_reissue.pdf, 2008.
364. Australian Bureau of Statistics, Experimental Estimates and Projections,
Aboriginal and Torres Strait Islander Australians, 1991 to 2021. Catalogue No.
3238.0, Canberra: Australian Government Publishing Service, 2009.
365. Epidemiology & GIS Branch and W.A. Department of Health, Rates Calculator
(Verson 9.5.1). Accessed from
http://intranet.health.wa.gov.au/epidemiology/resources/index.cfm, 2009.
366. Diabetes Epidemiology Research International Group, Diabetes registries in
Asia. Bulletin of the World Health Organization,, 1987. 65(6): p. 897-903.
242
367. LaPorte, R., et al., Counting Diabetes in the Next Millenium: Application of
capture-recapture technology. Diabetes Care, 1993. 16(2): p. 528-534.
368. Newcombe HB, Handbook of Record Linkage. Oxford University Press, New
York, 1988.
369. Kelman, C.W., A.J. Bass, and C.D.J. Holman, Research use of linked health
data – a best practice protocol. Australian and New Zealand Journal of Public
Health, 2002. 26(3): p. 251-255.
370. Libman, I.M. and D.J. Becker, Coexistence of type 1 and type 2 diabetes
mellitus: “double” diabetes? Pediatric Diabetes, 2003. 4(2): p. 110-113.
371. Owen, K. and M. McCarthy, Type 1 and type 2 diabetes—chalk and cheese?
Diabetologia, 2009. 52(10): p. 1983-1986.
372. Verge, C., M. Silink, and N. Howard, The Incidence of Childhood IDDM in New
South Wales, Australia. Diabetes Care, 1994. 17(7): p. 693-696.
373. Australian Bureau of Statistics, Estimated Resident Population by Age and Sex
in Statistical Local Areas, Western Australia. Catalogue No. 3203.5, Canberra:
Australian Government Publishing Service, 1985 - 2001.
374. Australian Bureau of Statistics, Population by Age and Sex, Western Australia.
Catalogue No. 3235.5, Canberra: Australian Government Publishing Service,
2002.
375. Epidemiology Branch, et al., Rates Calculator (version 9.1.3).
376. Gee, V. and T. Green, Perinatal Statistics in Western Australia, 2003. Twenty-
first Annual Report of the Western Australian Midwives' Notification System.
Department of Health, Perth, Western Australia. 2004.
243
377. Le M and Tran BN, Perinatal Statistics in Western Australia, 2008: Twenty-
sixth Annual Report of the Western Australian Midwives’ Notification System.
Department of Health, Western Australia. Accessed 6 November 2011 from
http://www.health.wa.gov.au/publications/documents/perinatal/PerinatalStatistic
sWA2008.pdf, 2008.
378. Australian Bureau of Statistics, Information Paper. 1996 Census of Population
and Housing. Socio-economic Indexes for Areas. Catalogue No. 3203.5,
Canberra: Australian Government Publishing Service, 1996.
379. Australian Bureau of Statistics, Information Paper: Census of Population and
Housing. Socio-economic Indexes for Areas. Australia 2001. Catalogue No.
2039.0, Canberra: Australian Government Publishing Service, 2001.
380. Stata Corporation, STATA statistical software (version 9.0).
381. Haynes, A. and E.A. Davis, Why is Type 1 diabetes becoming more common in
children? Diabetes Management Journal, 2006. 17: p. 7.
382. Gale, E., The rise of childhood type 1 diabetes in the 20th century. Diabetes,
2002. 51: p. 3353-3361.
383. Tuomilehto, J., et al., Record-high incidence of type 1 (insulin-dependent)
diabetes mellitus in Finnish children. Diabetologia, 1999. 42: p. 655-660.
384. Epidemiology Branch, et al., Population characteristics of residents of the State.
2002.
385. Stolwijk, A.M., H. Straatman, and G.A. Zielhuis, Studying seasonality using
sine and cosine functions in regression analysis. Journal of Epidemiology &
Community Health, 1999. 53: p. 235-238.
386. Stata Corporation, STATA statistical software (version 8.0).
244
387. Craig, M., et al., The Rising Incidence of Childhood Type 1 Diabetes in New
South Wales, Australia. Journal of Pediatric Endocrinology & Metabolism,
2000. 13: p. 363-372.
388. Toth, E., et al., High incidence of IDDM over 6 years in Edmonton, Alberta,
Canada. Diabetes Care, 1997. 20(3): p. 311-313.
389. Sebastiani, L., et al., A 5-year (1989-1993) prospective study of the incidence of
IDDM in Rome and the Lazio region in the age-group 0-14 years. Diabetes
Care, 1996. 19(1): p. 70-73.
390. Rangasami, J.J., et al., Rising incidence of type 1 diabetes in Scottish children,
1984-93. Archives of Diseases in Childhood, 1997. 77(3): p. 210-213.
391. Akaike, H., Information theory and an extension of the maximum likelihood
principle. In: Petrov BN, Csaki F (Eds.), Proceedings of the 2nd International
Symposium of Information Theory, Akademiai Kiado, Budapest, 1973: p. 267-
281.
392. Bozdogan, H., Akaike's Information Criterion and Recent Developments in
Information Complexity. Journal of Mathematical Psychology, 2000. 44: p. 62-
91 Accessed on 7 January 2012 from
http://www.yaroslavvb.com/papers/bozdogan-akaike.pdf.
393. Rewers, M., et al., Apparent epidemic of insulin-dependent diabetes mellitus in
Midwestern Poland. Diabetes, 1987. 36(1): p. 106-13.
394. Nystrom, L., et al., The Swedish childhood diabetes study. An analysis of the
temporal variation in diabetes incidence 1978-1987. International Journal of
Epidemiology, 1990. 19(1): p. 141-6.
245
395. Podar, T., et al., Risk of childhood type 1 diabetes for Russians in Estonia and
Siberia. International Journal of Epidemiology, 1993. 22(2): p. 262-7.
396. Bruno, G., et al., Increasing trend of Type 1 diabetes in children and young
adults in the province of Turin (Italy). analysis of age, period and birth cohort
effects from 1984 to 1996. Diabetologia, 2001. 44: p. 22-25.
397. Yang, Z., et al., Epidemics of type 1 diabetes in China. Pediatric Diabetes, 2005.
6(3): p. 122-128.
398. Staines, A., et al., The epidemiology of diabetes mellitus in the United Kingdom:
The Yorkshire regional childhood diabetes register. Diabetologia, 1993. 36(12):
p. 1282-1287.
399. McNally, R.J.Q., et al., Cyclical Variation in Type 1 Childhood Diabetes.
Epidemiology, 2010. 21(6): p. 914.
400. Lipman, T.H., Y. Chang, and K.M. Murphy, The Epidemiology of Type 1
Diabetes in Children in Philadelphia 1990–1994: Evidence of an epidemic.
Diabetes Care, 2002. 25(11): p. 1969-1975.
401. Diaz-Horta, O., et al., Echovirus Epidemics, Autoimmunity, and Type 1
Diabetes. Available from www.intechopen.com/books/type-1-diabetes-
pathogenesis-genetics-immunology/echovirus-epidemics-autoimmunity-and-
type-1-diabetes. Accessed May 2013. Type 1 Diabetes - Pathogenesis, Genetics
and Immunotherapy. 2011: InTech.
402. Muirhead, C., et al., How do childhood diagnoses of type 1 diabetes cluster in
time? PLoS One, 2013. 8(4).
246
403. Nystrom, L., et al., Risk of developing insulin-dependent diabetes mellitus
(IDDM) before 35 years of age: indications of climatological determinants for
age at onset. International Journal of Epidemiology, 1992. 21(2): p. 352-358.
404. Charkaulk, M.-L., P. Czernichow, and C. Levy-Marchal, Incidence Data of
Childhood-Onset Type I Diabetes in France During 1988-1997: The Case for a
Shift Toward Younger Age at Onset. Pediatric Research, 2002. 52(6): p. 859-
862.
405. Rewers, M. and P. Zimmet, The rising tide of childhood type 1 diabetes - what is
the elusive environmental trigger? Lancet, 2004. 364: p. 1645-7 (Comment).
406. Law, G., et al., Clustering of childhood IDDM. Links with age and place of
residence. Diabetes Care, 1997. 20(5): p. 753-756.
407. Joner, G., L.C. Stene, and O. Sovik, Nationwide, Prospective Registration of
Type 1 Diabetes in Children Aged <15 Years in Norway 1989-1998: No
increase but significant regional variation in incidence. Diabetes Care, 2004.
27(7): p. 1618-1622.
408. Casu, A., et al., Type 1 Diabetes Among Sardinian Children Is Increasing: The
Sardinian diabetes register for children aged 0-14 years (1989-1999). Diabetes
Care, 2004. 27(7): p. 1623-1629.
409. Hammam, R., et al., Colorado IDDM Registry Incidence and Validation of
IDDM in Children Aged 0-17 Yr. Diabetes Care, 1990. 13(5): p. 499-506.
410. Patterson, C. and N.R. Waugh, Urban/rural and deprivational differences in
incidence and clustering of childhood diabetes in Scotland. International Journal
of Epidemiology, 1992. 21: p. 108-117.
247
411. Siemiatycki, J., et al., Incidence of IDDM in Montreal by ethnic group and by
social class and comparisons with ethnic groups living elsewhere. Diabetes,
1988. 37: p. 1096-102.
412. Crow YJ, Alberti KGMM, and Parkin JM, Insulin dependent diabetes in
childhood and material deprivation in northern England, 1977-86. BMJ, 1991.
303: p. 158-60.
413. Soltesz, G., Diabetes in the young: a paediatric and epidemiological
perspective. Diabetologia, 2003. 46: p. 447-454.
414. Australian Bureau of Statistics, Census of Population and Housing: Usual
Residents Profile. Catalogue No. 2004.0, Canberra: Australian Government
Publishing Service, 2003.
415. Baumer, J.H., L.P. Hunt, and J.P. Shield, Social disadvantage, family
composition, and diabetes mellitus: prevalence and outcome. Archives of
Disease in Childhood, 1998. 79(5): p. 427-30.
416. Hyndman, J., et al., Misclassification of social disadvantage based on
geographical areas: comparison of postcode and collector's district analyses.
International Journal of Epidemiology, 1995. 24(1): p. 165-176.
417. Australian Institute of Health and Welfare (AIHW), Australia's health 2004.
Canberra: AIHW. Accessed 1 October 2004 from
http://www.aihw.gov.au/publications/index.cfm/title/10014, 2004.
418. Sepp, E., et al., Intestinal microflora of Estonian and Swedish infants. Acta
Paediatrica, 1997. 86(9): p. 956-61.
419. Australian Bureau of Statistics, Regional Population Growth, Australia, 2009-10
(Catalog No. 3128.0). Canberra: Australian Government Publishing Service.
248
Accessed 20 Nov 2011 from
http://www.abs.gov.au/ausstats/[email protected]/Products/3218.0~2009-
10~Main+Features~Western+Australia?OpenDocument, 2011.
420. Algert, C.S., et al., Perinatal risk factors for early onset of Type 1 diabetes in a
2000-2005 birth cohort. Diabetic Medicine, 2009. 26(12): p. 1193-1197.
421. Stanley, F.J., et al., A population database for maternal and child health
research in Western Australia using record linkage. Paediatric and Perinatal
Epidemiology, 1994. 8: p. 433-447.
422. Rosenberg, T.J., et al., Maternal obesity and diabetes as risk factors for adverse
pregnancy outcomes: differences among 4 racial/ethnic groups. Am J Public
Health, 2005. 95: p. 1545 - 1551.
423. Hummel, M., M. Schenker, and A.-G. Ziegler, Influence of perinatal factors on
the appearance of islet autoantibodies in offspring of parents with type 1
diabetes. Pediatric Diabetes, 2001. 2(1): p. 40-42.
424. Weiss, P., et al., Long-term follow-up of infants of mothers with type 1 diabetes:
evidence for hereditary and non-hereditary transmission of diabetes and
precursors. Diabetes Care, 2000. 23: p. 905-911.
425. Dorman, J., et al., Molecular type 1 diabetes epidemiology: International
studies. Diabetes Research and Clinical Practice, 1996. 34([Suppl]): p. S107-
116.
426. Jongbloet, P.H., et al., Seasonality of birth in patients with childhood diabetes in
the Netherlands. Diabetes Care, 1998. 21(1): p. 190.
249
427. Dahlquist, G., S.S. Bennich, and B. Kallen, Intrauterine growth pattern and risk
of childhood onset insulin dependent (type I) diabetes: population based case-
control study. BMJ, 1996. 313(7066): p. 1174-1177.
428. Weedon, M.N., et al., Genetic Regulation of Birth Weight and Fasting Glucose
by a Common Polymorphism in the Islet Cell Promoter of the Glucokinase
Gene. Diabetes, 2005. 54(2): p. 576-581.
429. Evagelidou, E.N., et al., Lipid Profile, Glucose Homeostasis, Blood Pressure,
and Obesity-Anthropometric Markers in Macrosomic Offspring of Nondiabetic
Mothers. Diabetes Care, 2006. 29(6): p. 1197-1201.
430. Dahlquist, G. and B. Kallen, Maternal-child blood-group incompatibility and
other perinatal events increase the risk for early-onset type-1 (insulin-
dependent) diabetes-mellitus. Diabetologia, 1992. 35(7): p. 671-675.
431. Leslie, R.D.G. and M.D. Castelli, Age-Dependent Influences on the Origins of
Autoimmune Diabetes: Evidence and Implications. Diabetes, 2004. 53(12): p.
3033-3040.
432. Ziegler, A.G., et al., On the appearance of islet associated autoimmunity in
offspring of diabetic mothers: a prospective study from birth. Diabetologia,
1993. 36(5): p. 402-8.
433. Songini, M., et al., The Sardinian IDDM Study: 1. Epidemiology and
geographical distribution of IDDM in Sardinia during 1989 to 1994.
Diabetologia, 1998. 41: p. 221-227.
434. Svensson, J., et al., Early childhood risk factors associated with type 1 diabetes-
-is gender important? European Journal of Epidemiology, 2005. 20(5): p. 429-
34.
250
435. Vehik, K., et al., Increasing incidence of type 1 diabetes in 0- to 17-year-old
Colorado youth. Diabetes Care, 2007. 30(3): p. 503-9.
436. Hathout, E.H., et al., Clinical, Autoimmune, and HLA Characteristics of
Children Diagnosed With Type 1 Diabetes Before 5 Years of Age. Pediatrics,
2003. 111(4): p. 860-863.
437. Murao, S., et al., Differences in the Contribution of HLA-DR and -DQ
Haplotypes to Susceptibility to Adult- and Childhood-Onset Type 1 Diabetes in
Japanese Patients. Diabetes, 2004. 53(10): p. 2684-2690.
438. Scholl, T.O., et al., Maternal Glucose Concentration Influences Fetal Growth,
Gestation, and Pregnancy Complications. Am J Epidemiol, 2001. 154(6): p.
514-520.
439. McKinney, P.A. and on behalf of the EURODIAB Seasonality of Birth Group,
Seasonality of birth in patients with childhood Type 1 diabetes in 19 European
regions. Diabetologia, 2001. 44 [Suppl 3]: p. B67-B74.
440. Laron, Z., et al., Month of birth and subsequent development of type 1 diabetes.
J Pediatr Endocrinol Metab, 1999. 12: p. 397-402.
441. Lipton, R.B., et al., Onset features and subsequent clinical evolution of
childhood diabetes over several years. Pediatric Diabetes, 2011. 12(4pt1): p.
326-334.
442. Jones, M.E. and A.J. Swerdlow, Bias caused by migration in case-control
studies of prenatal risk factors for childhood and adult diseases. American
Journal of Epidemiology, 1996. 143: p. 823-831.
443. Hammond, G., et al., Describing trends in gestational age, maternal age and
preterm birth rates in Western Australia from 1984 to 1999 (Abstract #49).
251
Australiasian Epidemiologist. Proceedings from the AEA Annual Scientific
Meeting 2011, 2011. 18(2): p. 56.
444. Joyce A and Tran BN, Western Australia’s Mothers and Babies, 2009. Twenty-
seventh Annual Report of the Western Australian Midwives‘ Notification System.
Department of Health, Western Australia, 2009.
445. Rasmussen, T., et al., Maternal BMI before pregnancy, maternal weight gain
during pregnancy, and risk of persistent positivity for multiple diabetes-
associated autoantibodies in children with the high-risk HLA genotype: the
MIDIA study. Diabetes Care, 2009. 32(10): p. 1904-6.
446. Cardwell, C.R., et al., Caesarean section is associated with an increased risk of
childhood-onset type 1 diabetes mellitus: a meta-analysis of observational
studies. Diabetologia, 2008. 51(5): p. 726-35.
447. Zipitis, C. and A. Akobeng, Vitamin D supplementation in early childhood and
risk of type 1 diabetes: a systematic review and meta-analysis. Archives of
Disease in Childhood, 2008. 93: p. 512-7.
448. Hagopian, W.A., et al., The Environmental Determinants of Diabetes in the
Young (TEDDY): genetic criteria and international diabetes risk screening of
421 000 infants. Pediatric Diabetes, 2011. 12(8): p. 733-743.
449. MacFarlane, A., A. Strom, and F. Scott, Epigenetics: deciphering how
environmental factors may modify autoimmune type 1 diabetes. Mammalian
Genome, 2009. 20(9): p. 624-632.
450. Ludvigsson, J., Why diabetes incidence increases--a unifying theory. Annals of
the New York Academy of Sciences, 2006. 1079: p. 374-82.
252
451. Morahan, G., et al., Genetic analyses define six distinct subtypes of Type 1
diabetes (Abstract #8). Proceedings from Australian Diabetes Society Annual
Scientific Meeting 2011, 2011.
452. Lempainen, J., et al., Interplay between PTPN22 C1858T polymorphism and
cow's milk formula exposure in type 1 diabetes. Journal of Autoimmunity, 2011.
33(2): p. 155-64.
453. Hewagama, A. and B. Richardson, The genetics and epigenetics of autoimmune
diseases. Journal of Autoimmunity, 2009. 33(1): p. 3-11.
454. Poirier, L.A., The Effects of Diet, Genetics and Chemicals on Toxicity and
Aberrant DNA Methylation: an Introduction. The Journal of Nutrition, 2002.
132(8): p. 2336S-2339S.
455. Meda, F., et al., The epigenetics of autoimmunity. Cellular & Molecular
Immunology, 2011. 8(3): p. 226-36.
456. Litherland, S.A., Immunopathogenic interaction of environmental triggers and
genetic susceptibility in diabetes: is epigenetics the missing link? Diabetes,
2008. 57(12): p. 3184-6.
457. Waterland, R.A. and R.L. Jirtle, Transposable elements: targets for early
nutritional effects on epigenetic gene regulation. Molecular Cell Biology, 2003.
23: p. 5293-5300.
458. Heijmans, B.T., et al., The epigenome: archive of the prenatal environment.
Epigenetics: Official Journal of the DNA Methylation Society, 2009. 4(8): p.
526-31.
459. Bartel, D.P., MicroRNAs: Genomics, Biogenesis, Mechanism, and Function.
Cell, 2004. 116(2): p. 281-297.
253
460. Guay, C. and R. Regazzi, Circulating microRNAs as novel biomarkers for
diabetes mellitus. Nat Rev Endocrinol, 2013. advance online publication.
461. Guay, C., et al., Emerging roles of non-coding RNAs in pancreatic -cell function
and dysfunction. Diabetes, Obesity & Metabolism, 2012. 14 Suppl 3: p. 12-21.
462. Sato, F., et al., MicroRNAs and epigenetics. FEBS Journal, 2011. 278(10): p.
1598-609.
463. Goss, J., Projection of Australian health care expenditure by disease, 2003 to
2033. AIHW Catalogue No. HWE 43, Canberra: AIHW, 2008.
464. Mathers, C. and R. Penm, Health system costs of cardiovascular diseases and
diabetes in Australia 1993–94 AIHW Catalogue No. HWE 11, Canberra:
AIHW, 1999.
465. McNally, R., et al., Space-time clustering analyses of type 1 diabetes in children
from north-east England: support for an infectious aetiology? Environmental
Health, 2009. 8 (Suppl 1): S14.
466. Booth, M., et al., The epidemiology of overweight and obesity among Australian
children and adolescents, 1995-97. Australian and New Zealand Journal of
Public Health, 2001. 2: p. 162-9.
467. Aagaard-Tillery, K.M., et al., In utero tobacco exposure is associated with
modified effects of maternal factors on fetal growth. American Journal of
Obstetrics and Gynecology, 2008. 198(1): p. 66.e1-e6.
468. Blair, E., Why do Aboriginal newborns weigh less? Determinants of birthweight
for gestation. Journal of Paediatric and Child Health, 1996. 32: p. 498-503.
254
469. Robertson, L. and K. Harrild, Maternal and neonatal risk factors for childhood
type 1 diabetes: a matched case-control study. BMC Public Health, 2010. 10(1):
p. 281.
470. Marshall, A.L., et al., Type 1 diabetes mellitus in childhood: a matched case
control study in Lancashire and Cumbria, UK. Diabetic Medicine, 2004. 21(9):
p. 1035-1040.
471. Toschke, A.M., et al., Paternal smoking is associated with a decreased
prevalence of type 1 diabetes mellitus among offspring in two national British
birth cohort studies (NCDS and BCS70). Journal Of Perinatal Medicine, 2007.
35(1): p. 43-47.
472. Bonifacio, E., et al., Cesarean Section and Interferon-Induced Helicase Gene
Polymorphisms Combine to Increase Childhood Type 1 Diabetes Risk. Diabetes,
2011. 60(12): p. 3300.
473. Morahan, G., et al., Tests for genetic interactions in type 1 diabetes: linkage and
stratification analyses of 4,422 affected sib-pairs. Diabetes, 2011. 60(3): p.
1030-40.
474. Patel, C.J., J. Bhattacharya, and A.J. Butte, An Environment-Wide Association
Study (EWAS) on Type 2 Diabetes Mellitus. PLoS ONE, 2010. 5(5): p. e10746.
475. Rakyan, V.K., et al., Identification of Type 1 Diabetes–Associated DNA
Methylation Variable Positions That Precede Disease Diagnosis. PLoS Genet,
2011. 7(9): p. e1002300.
476. Vaarala, O., The gut as a regulator of early inflammation in type 1 diabetes.
Current Opinion in Endocrinology, Diabetes & Obesity, 2011. 18(4): p. 241-7.
255
477. Nieminen, J.K., et al., Altered Phenotype of Peripheral Blood Dendritic Cells in
Pediatric Type 1 Diabetes. Diabetes Care, 2012. 35(11): p. 2303-2310.
478. O’Brien, T., Gene Therapy for Type 1 Diabetes Moves a Step Closer to Reality.
Diabetes, 2013. 62(5): p. 1396-1397.
479. Bain, J.R., Targeted Metabolomics Finds Its Mark in Diabetes Research.
Diabetes, 2013. 62(2): p. 349-351.
480. Oresic, M., et al., Dysregulation of lipid and amino acid metabolism precedes
islet autoimmunity in children who later progress to type 1 diabetes. Journal of
Experimental Medicine, 2008. 205(13): p. 2975-84.
481. The Million Women Study Collaborative Group, The Million Women Study:
design and characteristics of the study population. Breast Cancer Research,
1999. 1: p. 73-80.
482. Australian Type 1 Diabetes Clinical Research Network, Environmental
Determinants of Islet Autoimmunity (ENDIA). http://www.t1dcrn.org.au/clinical-
research/environmental-determinants-of-islet-autoimmunity-endia Accessed 10
July 2013.
256
257
Appendices
258
259
Appendix A: Consent for data collection
1. Information sheet for parents/guardians of children aged 0 to 9 years
2. Information sheet for children aged 10 to 14 years
3. Consent form
260
Parent Information Sheet Version 3, 17th September 2009
The Epidemiology of Diabetes in Children and Adolescents in Australia Principal Investigator: Dr Elizabeth Davis, Register Coordinator: Maree Grant, Dept. Endocrinology & Diabetes, Princess Margaret Hospital for Children, WA Tel: 08 9340 7024
INFORMATION SHEET FOR PARENTS We would like you to consider registering your child in the Australasian Paediatric Endocrine Group (APEG) and National Diabetes Registers.
What is the purpose of the Registers? Type 1, (insulin dependent) diabetes, is one of the most common chronic illnesses of childhood. Since the 1990’s, the rate of newly diagnosed type 1 diabetes in children aged < 15 years has increased in Australia. Other forms of diabetes in young people, including type 2 diabetes, cystic fibrosis related diabetes and transient diabetes (eg due to steroid treatment) also appear to be increasing. The purpose of this register is to accurately determine the annual rates of all types of diabetes in young people throughout Australia.
What is involved? To help us understand trends in diabetes in Australian youth, we would like to enrol your child in the APEG diabetes register. The register only applies to young people who live in Australia, whose diabetes was diagnosed in Australia or while on holiday overseas and who were less than 19 years of age at diagnosis. If you consent, demographic information about your child will be recorded in the APEG register. If your child is aged < 15 years, this information will be transferred to the National Diabetes Register, which is managed by the Australian Institute of Health and Welfare (AIHW). If your child is aged 15-19 years, the information will be kept in the APEG register only. The information in both registers is confidential and Commonwealth legislation (the Privacy Act) protects both registers (Commonwealth Privacy Act 1988 and AIHW Act 1987).
These registers, as well as providing information on the rates of diabetes throughout Australia, will also facilitate important research into the causes and management of diabetes. The information in the National Diabetes Register will only be made available to approved medical researchers whose research studies are considered by the AIHW and hospital ethics committees to be scientifically sound, of importance to the community and likely to contribute to the control of diabetes or improvement in diabetes care. Any information collected during research will be confidential and data will be stored in a limited access, password-protected database. Consent forms and questionnaires will be stored in locked filing cabinets in the APEG diabetes register office. All participants are assigned a study number and identifying details, including names and addresses, are removed prior to statistical analysis and publication of data.
Other information As part of being in the registers you may give permission to be contacted to participate in future research studies. There is a separate section on the consent form (Part 2) for this research. However, you are not under any obligation to consent to take part in any research and even if you do agree you may withdraw your consent at any time. Participation in the register is voluntary and if you decide not to take part or decide to withdraw at any time now or in the future, this will not otherwise affect your child’s care at the Hospital. We also ask that your treating doctor complete Part 3 of the consent form. This is a requirement for transfer of data to the National Diabetes Register. If you decide to participate in the study, please return the completed consent form to the Dept. Endocrinology & Diabetes.
Note: To allow us to accurately determine the number of new cases of all forms of diabetes each year, you need to fill in both this registration form and the form provided by the National Diabetes Supply Scheme (NDSS) for the National Diabetes Register.
This research has been approved by the Princess Margaret Hospital for Children Ethics Committee. Should you have concerns about your rights as a participant in this research, or you have a complaint about the manner in which the research is conducted, it may be given to Dr Liz Davis or, if an independent person is preferred, to Dr Mark Salmon, Executive Director Medical Services, Princess Margaret Hospital for Children, on 9340 8221.
Child Information Sheet Version 1, 17th September 2009
The Epidemiology of Diabetes in Children and Adolescents in Australia Principal Investigator: Dr Elizabeth Davis, Register Coordinator: Maree Grant, Dept. Endocrinology & Diabetes, Princess Margaret Hospital for Children, WA Tel: 08 9340 7024 CHILD INFORMATION SHEET (10 – 14 years) We would like to include your name on the Australasian Paediatric Endocrine Group and National Diabetes Registers. Why do we have these registers?
Type 1 diabetes is common in children and is becoming more common in all states of Australia. What is the study about?
To help us understand how common diabetes is in Australian children, we would like to include your name in the Australian Paediatric Endocrine Group (APEG) diabetes register. The register only applies to children who live in Australia. If you agree to your name being in the register, your name and some information about you will be transferred to the National Diabetes Register, which is managed by the Australian Institute of Health and Welfare in Canberra. These registers might help us to find the cause of diabetes and how many children get diabetes each year. Your name will not appear in any reports. Do I have to take part in research?
If your name is on the registers you may be asked if we can contact you in the future for research studies. You do not have to agree to this now or in the future. Even if you take part at the beginning and change your mind later on and don’t want to be part of any study that is okay as well. All you need to do is to tell the researcher that you don’t want to take part anymore.
This project has been approved by the Princess Margaret Hospital for Children Ethics Committee. If you have any questions about this study, please contact the Diabetes Database Officer on 9340 7540, Mrs Linda Hop (Diabetes Clinical Nurse Consultant) on 9340 7179 or Dr Tim Jones (Head of Dept) 9340 8090.
If you would like your child to be registered in the APEG and National Diabetes Registerbut do notwish to be contacted about any research, TWEN ONLY FILL IN PART 1.
If you agree to be contacted aboutfuture research, PLEASE TICK THE *YES" BOXAND COMPLETE PART 2.
Your Doctor Should Complete PART 3.
PART 1
I... ... ... ....the parentlguardian of.having read the information sheet, @nsent to the details below regarding my child's diabetes being
registered in the Australasian Paediatric Endocrine Group's Register yes fl No fland for the data to be transferred to the National Diabetes Register. yes E No ESigned. Dated.
CHILD OR ADOLESCENT DETAILS
Family Name, First/Second Names.
Sex: Malefl r.*ule E Type I diabetes E Otherdiabetes .........tr(Type ofdiabetes)
Date Of Birth I I Country Of Birth.
Current Address:
Postcode.
Postcode of Home Address at Diagnosis..... Date of diabetes diagnosis _/_/_Date of first insulin injection
Indigenous Status: Aboriginal yes fl No El Torres Strait Islander yes E No El
PART 2
I also give my permission to be eontacted about further research studies. yes El No El
Name of Carer. Telephone No...
Signed. Dated
PART 3 (To Be Completed By Your Treating Doctor)
Name: GP
Endocrinologist
Practice Address: Specialist Physician
Other Medical Practitioner
Telephone No... Provider Number.
Signature. Date
Do you wish to be notified about any research involvine contact with this natient? yes E No E
trtrE]tr
Consent Form Version 2, July 2008
265
Appendix B: Midwives’ Notification System
1. Data collection form
2. List of perinatal data variables stored in the Midwives’ Notification System
266
HEALTH DEPARTMENT’S COPY
Health Act (Notification by Midwife) Regulations Form 2 NOTIFICATION OF CASE ATTENDED MR15
Surname Unit
Record No.
Forenames Birth date (Mother)
Address of usual residence
Number and street State Post code
Town or suburb Height
(whole cm)
Maiden name Telephone
Establishment
Ward
Marital status
1=never married 2=widowed 3=divorced
4=separated 5=married (incl. defacto)
6=unknown
Ethnic status
1=Caucasian 2=Aboriginal/TSI
Other
PREGNANCY DETAILS
PREVIOUS PREGNANCIES:
Total number (excluding this pregnancy):
Previous pregnancy outcomes:
– liveborn, now living
– liveborn, now dead
– stillborn
Previous caesarean section 1=yes 2=no
Caesarean last delivery 1=yes 2=no
Previous multiple births 1=yes 2=no
THIS PREGNANCY:Antenatal: Estimated gestation weeks at first antenatal visit (excludes contact to test for pregnancy. None, use ‘98’; undetemined, use ‘99’; in 1st incomplete week, use ‘00’)
Date of LMP: 2 0
This date certain 1=yes 2=no
Expected due date: 2 0
based on 1=clinical signs/dates
2=ultrasound <20 wks
Smoking: Number of tobacco cigarettes usually smoked each day during first 20 weeks of pregnancy (none, use ‘000’; occasional or smoked <1, use ‘998’; undetermined, use ‘999’)
Number of tobacco cigarettes usually smoked each day after 20 weeks of pregnancy. (none, use ‘000’; occasional or smoked <1, use ‘998’; undetermined, use ‘999’)
Complications of pregnancy:1 threatened abortion (<20wks)2 threatened preterm labour (<37 wks)3 urinary tract infection4 pre-eclampsia5 Antepartum haemorrhage (APH) – placenta praevia6 APH – placental abruption7 APH – other8 pre-labour rupture of membranes9 gestational diabetes10 other (specify)
Medical conditions:1 essential hypertension2 pre-existing diabetes mellitus3 asthma4 genital herpes8 other (specify)
Procedures/treatments:1 fertility treatments (include drugs)2 cervical suture3 CVS/placental biopsy4 amniocentesis5 ultrasound6 CTG antepartum7 CTG intrapartum
Intended place of birth at onset of labour:
1=hospital 2=birth centre attached to hospital
3=birth centre free standing 4=home 8=other
MIDWIFE
Name
Signature
Date 2 0
Reg. No.
LABOUR DETAILS
Onset of labour: 1=spontaneous 2=induced 3=no labour
Augmentation (labour has begun):
1 none2 oxytocin3 prostaglandins4 artifical rupture of membranes8 other
Induction (before labour began):
1 none2 oxytocin3 prostaglandins4 artificial rupture of membranes8 other
Analgesia (during labour):
1 none2 nitrous oxide3 intra-muscular narcotics4 epidural/caudal5 spinal7 combined spinal/epidural8 other
Duration of labour: hr min
1st stage (hour & min):
2nd stage (hour & min):
DELIVERY DETAILS
Anaesthesia (during delivery):
1 none2 local anaesthesia to perineum3 pudendal4 epidural/caudal5 spinal6 general7 combined spinal/epidural8 other
Complications of labour and delivery(includes the reason for operative delivery):
1 precipitate delivery 2 fetal distress3 prolapsed cord4 cord tight around neck5 cephalopelvic disproportion 6 PPH(≥500mls)7 retained placenta - manual removal8 persistent occipito posterior9 shoulder dystocia10 failure to progress ≤3cm11 failure to progress > 3cm12 previous caesarean section13 other (specify)
Perineal status:
1=intact 2=1st degree tear/vaginal tear
3=2nd degree tear 4=3rd degree tear
5=episiotomy 6=episiotomy plus tear
7= 4th degree tear 8=other
FORWARD FORM TO
Maternal & Child Health Unit
Department of Health, Western Australia
Reply Paid 70042
(Delivery to Locked Bag 52)
Perth BC WA 6849
NB: Guidelines for completion of this form are available
from the above address or the following email address [email protected] or website: www.health.wa.gov.au/publications/subject_index/p/Perinatal_infant_maternal.cfm
BABY DETAILS
(Please use a separate form for each baby)
Adoption: 1=yes 2=no
Born before arrival: 1=yes 2=no
Birth date: 2 0
Birth time (24hr clock):
Plurality (number of babies this birth):
Birth order (specify this baby, eg, 1=1st baby born, 2=2nd baby born, etc):
Presentation: 1=vertex 2=breech 3=face 4=brow 8=other
Method of birth:1 spontaneous 2 vacuum successful3 vacuum unsuccessful4 forceps successful5 forceps unsuccessful6 breech (vaginal)7 elective caesarean8 emergency caesarean
Accoucheur(s):1 obstetrician2 other medical officer3 midwife4 student5 self/no attendant8 other
Gender:
1=male 2=female 3=indeterminate
Status of baby at birth:
1=liveborn 2=stillborn (unspecified)
3= antepartum stillborn 4=intrapartum stillborn
Infant weight (whole gram):
Length (whole cm):
Head circumference (whole cm):
Time to establish unassisted regular
breathing (whole min):
Resuscitation: (record one only – the most invasive or
highest number)
1 none 2 suction only 3 oxygen therapy only 4 bag and mask (IPPR)5 endotrachaeal intubation 6 ext. cardiac massage and ventilation 8 other
Apgar score: 1 minute
5 minutes
Estimated gestation (whole weeks):
Birth defects (specify):
Birth trauma (specify):
BABY SEPARATION DETAILS
Separation date: 2 0
Mode of separation:
1=transferred 8=died 9=discharged home
Transferred to: (specify establishment code)
Special care: (excludes Level 1; whole days only)
Coder ID:
General guidelines for completion of this form
When completing this form, please use a ballpoint pen and place the form on a firm surface 1.
to ensure legibility of all three copies.
Answer ALL questions.2.
If a particular item of information is not available, then record as “unknown”.3.
When text is required, please PRINT (preferably with the use of block letters).4.
Abbreviations should be limited to those in common use, to avoid miscoding of information.5.
Addressograph labels may be used but please ensure that one is placed on each of the 6.
three copies of the form.
Wherever possible, insert home or contact telephone number to facilitate continuity of care 7.
by Child Health Nurses. If unavailable, indicate with a dash or write “none”.
Where there are more boxes provided than required, please “right adjust” your response, 8.
e.g. a birth weight of 975 grams inserted as 0975.
For all dates, eight boxes are provided, e.g. 6 March 1965 inserted as 06 03 1965.9.
Some items allow more than one response. These are identified by multiple boxes, 10.
e.g. Complications of labour and delivery.
Complications not listed in tick boxes should be recorded as text under the appropriate
headings.
If further information is required for completion of this form, please refer to the “Guidelines for
Completion of the Notication of Case Attended Health Act (Notication by Midwife) Regulations
Form No.2” available from the website below or from the following:
The Manager
Maternal and Child Health Unit
Department of Health, Western Australia
1st Floor, C Block
189 Royal Street
East Perth WA 6004
Telephone: (08) 9222 2417
Email: [email protected]
Web: www.health.wa.gov.au/publications/subject_index/p/Perinatal_infant_maternal.cfm
HP11521 FEB’11 © Department of Health 2011
Midwives Notification System Data Variables Items in bold red are sensitive and require justification. Unless otherwise stated, all variables are collected from 1980 onwards.
Variable Name Variable Code Variable Description Values Comments Mother’s Details
Date of birth MDOB Date of birth of mother DDMMYYYY MMYYYY YYYY
Maternal age MATAGE Mother’s age in years at time of delivery
Number
State STATE State/Territory of residence
Abbreviations e.g. WA, SA, VIC
Height HT Number in centimetres Number
Marital status MS Marital Status of mother
1 = never married 2 = widowed 3 = divorced 4 = separated 5 = married (inc. defacto) 6 = unknown
Ethnic origin ETHNIC Ethnic origin of mother
1 = Caucasian 2 = Aboriginal/TSI 3 – Asian 4 = Indian 5 = African/Negroid 6 = Polynesian 7 = Maori 8 = Other
Pregnancy Details
Previous pregnancies MPREG Total number of previous pregnancies Number
Previous pregnancy outcomes
DEAD ALIVE SB
Outcomes of previous pregnancies
DEAD 0-99 ALIVE. 0-99 SB 0-99
Number of previous outcomes included, e.g. 2 SB = two previous children were stillborn Blank or zero if there are no previous children
Previous caesarean section
PREVCAES Previous caesarean section
1 = Yes 2 = No
Collected September 1997 onwards
Caesarean last delivery CAESLD 1 = Yes 2 = No
Previous multiple birth PREVMBTH Previous multiple birth 1 = Yes 2 = No
Collected September 1997 onwards
Date of last menstrual period LMP Date of last menstrual
period DDMMYYYY or unknown
Not a mandatory field.
Is the LMP date certain? CERT Is the LMP date certain?
1 = Yes 2 = No
Expected due date DATEEXP Expected due date MMYYYY only Not a mandatory field.
Basis of expected due date
DATEXP Due date based on
1 = Clinical signs/dates 2 = Ultrasound <20 weeks 3 = 20-40 weeks 8 = Unknown
Collected September 1997 onwards
Smoking during pregnancy
SMOKE Smoking during pregnancy
1 = Yes 2 = No
Collected September 1997 onwards
Complications of pregnancy
COMPRG Complications of pregnancy
1 = Threatened abortions (<20 weeks) 2 = Threatened preterm labour (<37 weeks) 3 = Urinary tract infection 4 = Pre-eclampsia 5 = APH – placenta praevia 6 = APH – placental abruption 7 = APH – other 8 = Pre-labour rupture of membrances 9 = Gestational diabetes 10 = Other – ICD-10AM codes
A case may have more than one value. Not a mandatory field.
Medical conditions MEDCOND Medical conditions
1 = Essential hypertension 2 = Pre-existing diabetes mellitus 3 = Asthma 4 = Genital herpes 8 = Other – ICD-10AM codes
A case may have more than one value. Not a mandatory field.
Procedures/treatments PROCTRM Procedures/Treatments
1 = Fertility treatments 2 = Cervical suture 3 = CVS/placental biopsy 4 = Amniocentesis 5 = Ultrasounds 6 = CTG antepartum 7 = CTG intrapartum
A case may have more than one value. Not a mandatory field. Collected 1993 onwards
Intended place of birth at onset of labour
INTBTH Intended place of birth at onset of labour
1 = Hospital 2 = Birth centre attached to hospital 3 = Birth centre free standing 4 = Home 8 = Other
Collected September 1997 onwards
Labour Details
Onset of labour ONSET 1 = Spontaneous 2 = Induced 3 = No-labour (elective caesarean)
Augmentation AUGLAB
1 = None 2 = Induced 3 = Prostaglandins 4 = Artificial rupture of membranes 8 = Other
A case may have more than one value. Collected 1990 onwards
Induction INDUCTN
1 = None 2 = Oxytocin 3 = Prostaglandins 4 = Artificial rupture of membranes 8 = Other
A case may have more than one value. Collected September 1997 onwards
Analgesia (during labour) ANALG Anaelgesia during labour
1 = None A case may have more than one value. Collected 1986 onwards
Delivery Details
Duration of labour 1st stage
HRS1ST Duration of 1st stage of labour
Hours and minutes
Collected September 1997 onwards. First stage of labour commences when contractions are of sufficient frequency, intensity and duration bring about dilation of the cervix.
Duration of labour 2nd stage HRS2ST
Duration of 2nd stage of labour
Hours and minutes
Collected September 1997 onwards. Second stage of labour commences when dilation of the cervix is complete and ends with delivery of the child.
Anaesthesia (during delivery)
ANAES
1 = None 2 = Local anaesthesia to perineum 3 = Pudendal 4 = Epidural/Caudal 5 = Spinal 6 = General 7 = Combined Epi/Spinal 8 = Other
A case may have more than one value. Collected 1986 onwards.
Complications of labour and delivery
COMPLAB
1 = Precipitate delivery 2 = Fetal distress 3 = Prolapsed cord 4 = Cord tight around neck 5 = Cephalopelvic disproportion 6 = PPH (≥ 500mLs)
A case may have more than one value. Not a mandatory field.
7 = Retained placenta – manual removal 8 = Persisten Occipito Posterior 9 = Shoulder dystocia 10 = Failure to progress ≤ 3cms 11 = Failure to progress > 3cms 12 = Previous caesarean section 13 = Other – ICD-10AM codes
Perineal status PERINEAL
1 = Intact 2 = 1st degree tear 3 = 2nd degree tear 4 = 3rd degree teat 5 = Episiotomy 6 = Episiotomy plus tear 8 = Other 9 = Not states
Baby Details
Born before arrival BBA 1 = Yes 2 = No
Collected September 1997 onwards.
Date of birth BDOB DDMMYYYY MMYYYY YYYY
Plurality PLURAL Number of babies in this birth
Number 1 = Singleton 2 = Twins 3 = Triplets etc
Baby number BABYNO
Number e.g. 1 = singleton or 1st baby in a multiple group 2 = 2nd baby in multiple group
Presentation PRSNT
1 = Vertex 2 = Breech 3 = Face 4 = Brow 8 = Other 9 = Not stated
Method of birth TYPEBTH Delivery method- hierarchical
1 = Spontaneous vaginal 2 = Vacuum successful 3 = Vacuum unsuccessful 4 = Forceps successful 5 = Forceps unsuccessful 6 = Breech (vaginal)
A case may have more than one value. Each value is entered in two digits, eg. 1 is 01. Also, it's hierarchical so the last two digits represent the last delivery mode.
7 = Elective caesarean 8 = Emergency caesarean
Accoucheur(s) ACCOUCH
1 = Obstetrician 2 = Other medical practioner 3 = Midwife 4 = Student 5 = Self/no attendant 8 = Other
A case may have more than one value. Collected September 1997 onwards.
Gender GENDER 1 = Male 2 = Female 3 = Undetermined
Status of baby at birth BSTATUS
1 = Liveborn 2 = Stillborn 3 = Antepartum Stillborn 4 = Intrapartum Stillborn
Infant weight INFANTWT Number in grams Length of baby (cms) BLGTH Number in centimetres Head circumference HEADC Number in centimetres Collected 1990 onwards Time to establish unassisted regular breathing
TSR Number rounded to nearest minute
Resuscitation RESUSC
1 = None 2 = Suction only 3 = Oxygen therapy 4 = Bag and mask 5 = Endotrachaeal intubation 6 = Ext. cardiac massage and ventilation 7 = Drugs
A case may have more than one value.
Agpar score at 1 minute AGPAR1 Number Collected 1990 onwards Agpar score at 5 minutes AGPAR5 Number
Estimated gestation ESTGEST Completed weeks Number Collected 1986 onwards Calculated GA available before 1986 on request.
Baby separation date BSEPDATE Date baby left hospital DDMMYYYY
Baby length of stay BLOS Baby’s length of stay in hospital
Number of days Calculated field from hospital record. Blank if baby was born at home.
Mode of separation BMOS Baby mode of separation
1 = Transferred 8 = Died 9 = Discharged home
Special Care SPEC Days of special care Number Not a mandatory field. Geocoding
Postcode PCODE Postcode of usual residence
Number e.g. 6000
Collector’s District CD Collectors District Number LGA LGA Local Government Area Number Collected 1996 onwards SLA SLA Statistical Local Area Number Collected 1996 onwards Radius RAD Radius Number Collected 1993 - 2003
Notes: Where there are multiple values on the NOCA form, the extracted data will appear as numbers and spaces. E.g. a record with complications of labour could have ‘urinary tract infection’, ‘pre-eclampsia’ and ‘APH – placental abruption’. In the data extract, this would look like: ‘_ _ _ _0304_ _06_ _ _ _ _ _ _ _’
275
Appendix C: Perinatal data validation
1. Variables not recorded for the whole study period
2. Variables with improbable values
3. Coding changes made in 1997
4. ICD Codes for Diabetes and related conditions
276
277
1. Variables not recorded for the whole study period
VARIABLE CODE ∗∗∗∗ YEARS AVAILABLE ACTION TAKEN
DEMOGRAPHIC
INDIGST
1993 � Used ETHNIC instead
COB
1993 �
FATH_AGE
1980 – 1991. Few values 1992 – 1999
PREGNANCY DETAILS
PREVCAES 1998 � New field 1997
PREVMBTH 1998 � New field 1997
CAESLD
1998 � New field 1997
LMP Not well recorded, better to use EDD
BASELMP 1998 � New field 1997
SMOKE 1997 � New field 1997 Complete 1998 onwards only
Restricted analysis of SMOKE to 1998 – 2002 births
PROCTRMT 1993 � (Coding change 1997)
INTENDED New field 1997
BABY DETAILS ADOPTION Up to 1984
BBA
1998 � New field 1997
∗ Please see listing of Midwives’ Notification System Data Variables for an explanation of what the
variable codes are (this is the document preceding this one in this appendix, Appendix C)
278
ESTGEST
1985 � New field 1986
Used EVEGA
HEADC 1990 � New field 1990
SPEC ~4,000 values 1980 – 2002 Few years with ~ 20,000
VITK 1993 � 1997 New field 1993
BTRAUMA Not well documented
APGAR1 / 5 1990 �, not v accurate APGAR1 new field 1990
TSR
BLGTH Inaccuracy likely
279
2. Variables with improbable values
VARIABLE CODE VALUE ACTION TAKEN
MAGE 0 Set to missing
INFANTWT <500 >6600 9999
Set to missing
MPREGS >=20 Set to missing Calculated PARITY = ALIVE + DEAD + SB (validated values) If validated MPREGS < PARITY & MPREGS = 0, Set MPREGS = missing (Can have MPREGS < PARITY if previous pregnancy was multiple pregnancy, but if mpregs = 0 then must be invalid)
ALIVE >=99 Set to missing
DEAD >=99 Set to missing
SB >=99 Set to missing
HT <140, > 200 e.g. 51, 556 (all controls) For calculation of POBW <147, > 183
Set to missing Set to missing
BLGTH <30, > 60, 99 Set to missing
DEATHDTE BROOTNO_ENC AG1RQ7F0MABXR BDOB = MAR1985 DEATHDTE = 08.03.84 Another record with deathdte = 27.09.2203
Changed to 08.03.85 Changed to 27.09.2003
PCODE 258 postcodes not WA
280
pcodes – could be valid 283 WA pcodes which Mark Peel could find no record of (? Historical) 6200 6216 6245 6310 6520 6553
281
1. Coding changes made in 1997
COMPLICATIONS OF PREGNANCY COMPRG 20 char string Described as binary fixed position string (Reflects which checkboxes checked on the Notification of Case Attended form under complications of pregnancy) Not a binary fixed position string E.g. If have gestational diabetes, should be “09” – may or may not be. If is “09” and no other complication, will probably be at position 1-2, not position 17-18. COMP1 – 14 2 char string Described as containing ICD codes for corresponding checkboxes ie. If ticked checkbox 9 (gestational diabetes), comp9 should have ICD code for gestational diabetes This does not necessarily follow. E.g. ICD code for gestational diabetes could be in any of the comp1 – 14 fields, not just comp9. NB: ICD codes for gestational diabetes also found in mcond8, mcond9. ** Therefore, need to look for ICD codes of interest for complications in all the comp1-14 AND the mcond1-12 fields. Don’t just go on the checkbox value (eg.09) PRE-EXISITING MEDICAL CONDITIONS MEDCOND 16 char string Described as binary fixed position string (Reflects which checkboxes checked on the Notification of Case Attended form under pre-existing medical conditions) Not a binary fixed position string eg. If have diabetes, should be “02” – may or may not be. If is “02” and no other condition, will probably be at position 1-2, not position 3-4. MCOND1-12 2 char string Described as containing ICD codes for corresponding checkboxes ie. If ticked checkbox 2 (pre-existing diabetes), mcond2 should have ICD code for diabetes This does not necessarily follow. E.g. ICD code for diabetes could be in any of the mcond1 – 12 fields, not just mcond2. NB: ICD codes for pre-exisitng diabetes also found in comp10, comp11, comp12 fields.
282
** Therefore, need to look for ICD codes of interest for pre-existing medical conditions in all the mcond1 – 12 AND the comp1-14 fields. Don’t just go on the checkbox value (eg.02) 4. ICD Codes for Diabetes and related conditions Codes in the box are the codes used in the Midwives’ Notification System DIABETES MELLITUS (MCOND2) ICD-9-CM 250.XX Diabetes mellitus [0-9, 0-3] Excludes: gestational diabetes (648.8) hyperglycemia NOS (790.6) neonatal diabetes mellitus (775.1) nonclinical diabetes (790.29) The following fifth-digit subclassification is for use with category 250: 0 type II or unspecified type, not stated as uncontrolled 1 type I, not stated as uncontrolled 2 type II or unspecified type, uncontrolled 3 type I, uncontrolled 250.0 Diabetes mellitus without mention of complication 250.1 Diabetes with ketoacidosis 250.2 Diabetes with hyperosmolarity 250.3 Diabetes with other coma 250.4 Diabetes with renal manifestations 250.5 Diabetes with ophthalmic manifestations 250.6 Diabetes with neurological manifestations 250.7 Diabetes with peripheral circulatory disorders 250.8 Diabetes with other specified manifestations 250.9 Diabetes with unspecified complication 648.0 Diabetes mellitus [0-4] Conditions classifiable to 250 648.00 Diabetes mellitus, unspecified 648.01 Diabetes mellitus, delivered w/o mention of antepartum condition 648.02 Diabetes mellitus, delivered with mention of a post partum condition 648.03 Diabetes mellitus, antepartum condition or complication
283
648.04 Diabetes mellitus, postpartum condition or complication ICD-10-AM E10.XX IDDM E11.XX NIDDM (E11.90 = NIDDM without complications) E13.XX Other specified diabetes mellitus E14.XX Unspecified diabetes mellitus O24.0 Pre-existing diabetes mellitus, IDDM O24.1 Pre-existing diabetes mellitus, NIDDM O24.2 Pre-existing malnutrition-related diabetes mellitus O24.3 Pre-existing diabetes mellitus, unspecified O24.9 Diabetes mellitus in pregnancy, unspecified GESTATIONAL DIABETES (COMP9) ICD-9-CM 648.8X Abnormal glucose tolerance - Gestational diabetes [0-4] 648.80 Abnormal glucose tolerance, unspecified
648.81 Abnormal glucose tolerance, delivered w/o mention of antepartum condition 648.82 Abnormal glucose tolerance, delivered with mention of a postpartum condition
648.83 Abnormal glucose tolerance, antepartum condition or complication 648.84 Abnormal glucose tolerance, postpartum condition or complication ICD-10-AM O24.4 Diabetes mellitus arising in pregnancy - GESTATIONAL
284
285
Appendix D: Outputs arising from Chapter 5
1. Conference and seminar presentations by Dr Aveni Haynes
2. PDF of published journal article Haynes A, Bower C, Bulsara MK, Jones TW and Davis EA. Continued increase in the incidence of childhood Type 1 diabetes in a population-based Australian sample (1985-2002). Diabetologia, 2004. 47(5): p. 866-870
3. Co-author declarations for published work
4. PDF of invited commentary Haynes A and Davis EA. Why is Type 1 diabetes becoming more common in children? Diabetes Management Journal, 2006. 17: p. 7
5. Co-author declaration for invited commentary 6. In Press (Accepted for publication May 2012) - Brief report
Haynes A, Bulsara MK, Bower C, Jones TW and Davis EA. Cyclical variation in the incidence of childhood Type 1 diabetes in Western Australia (1985 – 2010). Diabetes Care
7. Co-author declarations for brief report
286
287
Conference & Seminar Presentations by Dr Aveni Haynes
Oral presentations National August 2011 Temporal and spatial variation in the incidence of childhood
Type 1 diabetes in Western Australia (1985 - 2010) Australian Diabetes Society Annual Scientific Meeting, Perth
September 2003 Continued increase in incidence of childhood-onset Type 1
diabetes in Western Australia (1985 - 2002) Australian Diabetes Society Annual Scientific Meeting, Melbourne
September 2003 Incidence of Type 1 diabetes differs between metropolitan,
rural and remote areas of Western Australia Australasian Paediatric Endocrine Group Annual Scientific Meeting, Melbourne
Local November 2004 Childhood Type 1 diabetes in Western Australia (1985 – 2002)
Diabetes Physicians Group (Western Australia), Perth October 2003 Continued increase in incidence of childhood-onset Type 1
diabetes in Western Australia (1985 - 2002) Research and Advances Conference, Princess Margaret Hospital, Perth
Poster presentations National September 2011 Temporal and spatial variation in the incidence of childhood
Type 1 diabetes in Western Australia (1985 - 2010) Australasian Epidemiological Association Annual Scientific Meeting, Perth
September 2003 Increasing incidence of Type 1 diabetes in Western Australia
(1985 – 2002) Australasian Epidemiological Association Annual Scientific Meeting, Perth
288
Abstract
Aims/hypothesis. Our aim was to determine the inci-dence of Type 1 diabetes in children who were 0 to 14years of age in Western Australia from 1985 to 2002,and to analyse the trends in incidence rate over thesame period.Methods. Primary case ascertainment was from aprospective population-based diabetes register thatwas established in 1987, and secondary case ascertain-ment was from the Western Australia Hospital Mor-bidity Data System. Denominator data were obtainedfrom the Australian Bureau of Statistics. Poisson re-gression was used to analyse the incidence rates bycalendar year, sex and age at diagnosis.Results. There was a total of 1144 cases (560 boys,584 girls). Using the capture–recapture method, caseascertainment was estimated to be 99.8% complete.
The mean age standardised incidence from 1985 to2002 was 16.5 per 100000 person years (95% CI14.7–18.2), ranging from 11.3 per 100000 in 1985 to23.2 per 100000 in 2002. The incidence increased onaverage by 3.1% (95% CI 1.9%–4.2%) a year over theperiod (p<0.001). No significant difference was foundbetween boys and girls. A significant increase in inci-dence was found in all age groups, with no dispropor-tionate increase found in the 0 to 4-year-olds.Conclusions/interpretation. The incidence of child-hood-onset Type 1 diabetes in Western Australia hasincreased significantly over the past 18 years andshows no signs of abating. In contrast to other studies,a higher rate of increase was not found in the youngestchildren.
Keywords Australia · Childhood · Incidence · Type 1diabetes
Received: 29 December 2003 / Accepted: 21 February 2004Published online: 17 April 2004© Springer-Verlag 2004
E. A. Davis (✉)Department of Endocrinology and Diabetes, Princess Margaret Hospital, Subiaco, GPO Box D184, Perth,Western Australia 6008E-mail: [email protected].: +61-8-93408090, Fax: +61-8-93408605
Diabetologia (2004) 47:866–870DOI 10.1007/s00125-004-1385-8
Continued increase in the incidence of childhood Type 1 diabetesin a population-based Australian sample (1985–2002)A. Haynes1, 2 · C. Bower2 · M. K. Bulsara3 · T. W. Jones1, 2 · E. A. Davis1, 2
1 Department of Endocrinology and Diabetes, Princess Margaret Hospital, Perth, Western Australia 60082 Centre for Child Health Research, The University of Western Australia, Telethon Institute of Child Health Research, Subiaco,
Western Australia3 School of Population Health, The University of Western Australia, Nedlands, Western Australia
More recently, several studies in Europe have foundthat the highest rate of increase is occurring in chil-dren younger than 5 years of age [5, 6, 7].
Worldwide, the incidence of childhood-onset Type 1diabetes varies by more than 350-fold [8], ranging from0.1 per 100000 per year in China to 45 per 100000 peryear in Finland [9]. Epidemiological studies of Type 1diabetes are important in identifying possible aetiologi-cal factors that might explain the large geographicalvariation and changing epidemiology of the disease.
Western Australia presents a unique opportunity to characterise every child diagnosed with Type 1 dia-betes under the age of 15 years, enabling the epidemi-ology of childhood-onset Type 1 diabetes in this stateto be described accurately.
Western Australia is the largest state in Australia witha land area of over 2.5 million square kilometres and an
Introduction
The changing epidemiology of childhood-onsetType 1 diabetes has been reported in many countriesand remains unexplained. A significant increase in theincidence of Type 1 diabetes over the past fewdecades has been reported worldwide [1, 2, 3, 4].
A. Haynes et al.: Continued increase in the incidence of childhood Type 1 diabetes 867
estimated population of 1.9 million. Aboriginal peopleaccount for 3.5% of the state’s total population [10] andthe remainder are mainly people of European descent.
Princess Margaret Hospital for Children is the onlytertiary paediatric hospital in the state and the only re-ferral centre for children diagnosed with diabetes. InAustralia, children with diabetes are usually treated byhospital-based paediatricians and endocrinologists.This enables case ascertainment levels of over 99% tobe achieved consistently and the complete population-based data set can be used to describe the characteris-tics of newly diagnosed patients accurately.
The epidemiology of Type 1 diabetes in 0 to 14-year-olds in Western Australia has been describedfrom 1985 to 1992, when it was found to have in-creased significantly in both sexes and across all agegroups [11]. Therefore, the aims of this study were todetermine the incidence of Type 1 diabetes in WesternAustralia in children who were 0 to 14 years of agefrom 1985 to 2002, and to analyse the trends in inci-dence rate over the same period.
Subjects and methods
The primary source of cases was the Western Australia Chil-dren’s Diabetes database, which was established at PrincessMargaret Hospital in 1987 [12]. This prospective population-based register satisfies the criteria of the Diabetes Epidemiolo-gy International Group [13], enabling an international compar-ison of incidence rates. The database also contains retrospec-tive data, obtained using multiple sources, on children diag-nosed with Type 1 diabetes from 1985 to 1987. All patientsseen at Princess Margaret Hospital with newly diagnosedType 1 diabetes are registered on the database. As PrincessMargaret Hospital is the only referral centre for children diag-nosed with diabetes in Western Australia, the case ascertain-ment level of this database is over 99% [11, 12].
Data collected include name, date of birth, date of diagno-sis, age at diagnosis, sex, address at diagnosis, results from autoantibody assays and family history. Patients are seen every3 months at the diabetes clinic, and the database is updatedregularly with results from investigations. Therefore, therecords of any patients found to have been diagnosed incor-rectly with Type 1 diabetes, or to have diabetes secondary toother causes, are changed to include this information.
Secondary case ascertainment was from the Western Aus-tralian Hospital Morbidity Data System, which contains name-identified information on discharges from all public andprivate sector acute-care hospitals in the state. A list of all patients whose diagnosis at discharge included any referenceto Type 1 diabetes, and whose age at admission was less than15 years, was obtained and then cross-checked with cases inthe diabetes database. The capture-recapture method was usedto estimate completeness of ascertainment [14].
All patients diagnosed with Type 1 diabetes by a physicianand who started insulin therapy before the age of 15 years andwere resident in Western Australia at the time of diagnosiswere included in the study. Patients with diabetes secondary toother causes, such as cystic fibrosis, were excluded. Type 1 diabetes was diagnosed according to the definition of theWorld Health Organization [15] and the date of diagnosis wasdefined as the date when insulin therapy was initiated.
Population estimates based on census data were obtainedfrom the Australian Bureau of Statistics.
Ethics approval for this study was obtained from thePrincess Margaret Hospital Ethics Committee and the WesternAustralia Department of Health’s Confidentiality of Health In-formation Committee. Informed consent was obtained from allpatients prior to their data being stored on the diabetesdatabase at Princess Margaret Hospital.
Statistical analysis. Incidence rates were calculated using casesfrom both sources as the numerator and annual population esti-mates from the Australian Bureau of Statistics as the denomi-nator [16, 17]. Age-standardised rates were calculated usingthe direct method with the developed world population, whichhas equal numbers in each subgroup defined by age (0–4, 5–9and 10–14 years) and sex, as the standard [18]. Confidence intervals were estimated assuming a Poisson distribution ofcases.
Poisson regression models were used to analyse the inci-dence rates by calendar year, sex and age group at diagnosis(0–4, 5–9 and 10–14 years), and to estimate the temporaltrends. Models were fitted to test for differences in the inci-dence rate trends between the sexes and between the three agegroups.
To analyse the incidence for seasonal variation, sine andcosine functions in a logistic regression analysis were used[19]. This method also allows for covariate adjustment.
Annual incidence rates were calculated from 1985 to 2002.The rates calculated from 1985 to 1992 were found to be al-most identical to those previously published [11]. However, forconsistency, we present the most recently obtained data to ana-lyse the incidence rate trends over the study period.
A p value of less than 0.05 was considered statistically sig-nificant. Data were analysed by using the STATA statisticalsoftware package (version 8.0, Stata Corporation, College Sta-tion, Tex., USA).
Results
Ascertainment. From 1985 to 2002, there was a totalof 1144 cases (560 boys, 584 girls). Case ascertain-ment for 1985 to 1992 has been estimated as 99.6%[11]. From 1992 to 2002, of the 801 cases in total, 786were ascertained from both the register at PrincessMargaret Hospital and the Hospital Morbidity DataSystem, ten cases were ascertained only from the Hos-pital Morbidity Data System and 15 cases were ascer-tained only from the register at Princess MargaretHospital. Using the capture–recapture method, thecase ascertainment from 1992 to 2002 was estimatedto be 99.98%. Therefore, the case ascertainment forthe whole study period was estimated to be 99.8%complete.
Overall incidence. The mean age standardised inci-dence rate from 1985 to 2002 was 16.5 per 100000person years (95% CI: 14.7–18.2). The incidenceranged from 11.3 per 100000 person years in 1985, to23.2 per 100000 person years in 2002 (Fig. 1). The in-cidence has increased significantly (p<0.001) by anaverage of 3.1% (95% CI: 1.9%–4.2%) a year since1985.
868 A. Haynes et al.:
Sex. From 1985 to 2002, the mean age-standardisedincidence was 15.6 per 100000 person years (95% CI:13.7–17.5) in boys and 17.3 per 100000 person years(95% CI: 15.3–19.4) in girls. The incidence has in-creased significantly in both sexes, with an averageannual increase of 3.9% (95% CI: 2.2%–5.6%,p<0.001) in boys and 2.3% (95% CI: 0.6%–3.9%,p=0.005) in girls. There was no significant differencebetween the age-standardised incidence (incidencerate ratio = 1.10, 95% CI: 0.98–1.24, p=0.10) or inci-dence rate trends in boys compared with girls(p=0.18).
Age group. The mean incidence in 0 to 4-year-olds,from 1985 to 2002, was 11.0 per 100000 person years
(95% CI: 9.2–12.8). This was significantly lower thanthe mean incidence in 5 to 9 and 10 to 14-year-olds(Table 1). There was no significant difference betweenthe overall incidence of 18.8 per 100000 person years(95% CI: 16.3–21.3) in 5 to 9-year-olds and 19.6 per100000 person years in (95% CI: 17.6–21.6) in 10 to14-year-olds. The incidence has increased significant-ly in all age groups since 1985 (Fig. 2). The incidenceincreased by an average of 3.3% a year (95% CI:0.9%–5.9%) in 0 to 4-year-olds, 3.7% a year (95% CI:1.8%–5.7%) in 5 to 9-year-olds and 2.2% a year (95%CI: 0.4%–4.0%) in 10 to 14-year-olds. There was nosignificant difference in the incidence rate trends be-tween the age groups, and a disproportionate rate ofincrease in the youngest age group was not found.
Seasonal variation. There was significant seasonalvariation in the incidence of Type 1 diabetes. A sea-sonal pattern with one maximum and one minimumlevel per year was highly significant (p<0.001). Ahigher number of cases was diagnosed from April toSeptember, the autumn and winter months in Australia(Fig. 3). A non-significant pattern was found when in-vestigating two minimum and maximum levels peryear (p=0.398). There was no significant effect of sex(p=0.098); however, there was a highly significant ef-fect of age (p<0.001). When analysing the individualage groups, a significant seasonal pattern was onlyfound in the 0 to 4-year-olds (p=0.046).
Discussion
The incidence of childhood-onset Type 1 diabetes inWestern Australia is increasing at a rate comparable to
Fig. 1. The trend in the incidence of Type 1 diabetes in 0 to 14-year-olds in Western Australia from 1985 to 2002. Solidline and filled symbols: annual age-standardised incidence (error bars show 95% CI); dotted line: estimated trend
Table 1. Mean annual age- and sex-specific incidence rates per 100000 population (95% CI) in Western Australia from 1985 to2002
Sex and age group n Population at risk Mean annual incidence (95% CI)
Boys0–4 years 125 1155560 10.7 (8.5–13.0)5–9 years 196 1181568 16.4 (13.3–19.5)10–14 years 239 1195858 19.8 (17.3–22.4)0–14 years 560 3532986 15.6 (13.7–17.5)
Girls0–4 years 124 1094048 11.3 (9.2–13.4)5–9 years 241 1117536 21.4 (18.0–24.7)10–14 years 219 1129789 19.3 (16.5–22.1)0–14 years 584 3341373 17.3 (15.3–19.4)
All0–4 years 249 2249608 11.0 (9.2–12.8)5–9 years 437 2299104 18.8 (16.3–21.3)10–14 years 458 2325647 19.6 (17.6–21.6)0–14 years 1144 6874359 16.5 (14.7–18.2)
Data are mean annual incidence rates per 100000 per year (95% CIs) unless otherwise stated. Age-standardised rates are shown forthe 0 to 14-year age group
Continued increase in the incidence of childhood Type 1 diabetes 869
that reported in another state of Australia, New SouthWales [20], and in many European countries (3.2% ayear) [2, 3]. This increase in incidence cannot be ex-plained by increased ascertainment levels as the com-pleteness of the population-based diabetes database inWestern Australia has been estimated to be over 99%since its establishment in 1987, and case ascertain-ment methods have remained unchanged since thattime.
The overall incidence in Western Australia is similarto that in New South Wales [20] and it is in the high-incidence category compared to other countries [8].
There has been a significant increase in the inci-dence in both boys and girls. No significant differencewas found in the overall incidence or incidence ratetrends in boys compared with girls. There are incon-sistent findings on the difference in incidence ofType 1 diabetes by sex. In general, a male excess hasbeen found in countries with a high incidence and afemale excess in countries with a low incidence ofType 1 diabetes [18, 21, 22]. However, a higher inci-dence in girls was recently reported in New SouthWales [20] and similar to our findings, some studieshave found no difference in incidence between boysand girls [23, 24], although both these studies had as-certainment levels of 85%.
Many studies report an increase in incidence withincreasing age, with the highest incidence occurring inthe 10 to 14-year-old age group [8]. In this study, theincidence of Type 1 diabetes in the youngest agegroup was found to be significantly lower than that in5 to 9 and 10 to 14- year-olds. However, there was nosignificant difference in the incidence between 5 to 9and 10 to 14-year-olds.
In contrast to reports from Europe [2], Oxford, UK[5] and Finland [6], in which higher rates of increasein the youngest age group have been documented, adisproportionate rate of increase in 0 to 4-year-oldswas not found in Western Australia. There was no sig-nificant difference in the incidence rate trends be-tween the age groups, which is similar to the findingsin other countries, including Scotland [25, 26]. Al-though a trend towards a higher rate of increase in theyoungest age group was found in New South Wales,no statistically significant difference was found in thetrend between the age groups.
Significant seasonal variation in incidence wasfound with a higher number of cases being diagnosedin the autumn and winter months. This is consistentwith the findings from many studies that have found alower incidence of Type 1 diabetes occurring in thewarmer months [3]. Studies examining seasonal varia-tion by sex and age group have not had consistentfindings. Some studies have found significant season-al variation occurring only in boys [27] and varyingwith age, with some studies showing a seasonal effectin all age groups [3] and others, only in 10 to 14-year-olds [28]. Seasonal variation in incidence may reflect
Fig. 2. Trends in the age-specific annual incidence of Type 1diabetes in 0 to 14-year-olds in Western Australia from 1985 to2002. Observed rates in 0 to 4-year-olds (circles), 5 to 9-year-olds (squares), 10 to 14-year-olds (triangles) and estimatedtrends (dotted lines)
Fig. 3. Seasonal variation in incidence of Type 1 diabetes in 0to 14-year-olds in Western Australia
870 A. Haynes et al.: Continued increase in the incidence of childhood Type 1 diabetes
the role of environmental factors such as increased ex-posure to infections, lower physical activity and die-tary intake in the aetiology of Type 1 diabetes.
The rapid increase in incidence of Type 1 diabetesaround the world is likely to be the result of increasedexposure to environmental risk factors in geneticallysusceptible individuals. The non-differential increasein incidence in both sexes and all age groups, found inWestern Australia, suggests that the increased expo-sure to risk factors affects all of these groups.
The 18 years of complete population-based data onincident cases of Type 1 diabetes in Western Australiamean that our results are an accurate description of theepidemiology of Type 1 diabetes in this state. Contin-ued monitoring of the incidence of diabetes in Western Australia should increase our understandingof the global patterns of this disease and help generatehypotheses concerning potential aetiological factors.
Acknowledgements. We wish to thank Dr Jim Codde (Epide-miology Branch, Department of Health, Western Australia) forhis invaluable advice and Dr Mark S. Pearce (School of Clini-cal Medical Sciences, University of Newcastle, England) forhis advice on seasonality.
References
1. Onkamo P, Vaananen S, Karvonen M, Tuomilehto J (1999)Worldwide increase in incidence of Type 1 diabetes—theanalysis of the data on published incidence trends. Dia-betologia 42:1395–1403
2. EURODIAB ACE Study Group (2000) Variation andtrends in incidence of childhood diabetes in Europe. Lancet355:873–876
3. Green A, Patterson C, on behalf of the EURODIABTIGER Study Group (2001) Trends in the incidence ofchildhood-onset diabetes in Europe 1989–1998. Diabetolo-gia 44 [Suppl 3]:B3–B8
4. Gale EA (2002) The rise of childhood type 1 diabetes inthe 20th century. Diabetes 51:3353–3361
5. Gardner SG, Bingley PJ, Sawtell PA, Weeks S, Gale EA,the Bart’s-Oxford Study Group (1997) Rising incidence ofinsulin dependent diabetes in children aged under 5 years inthe Oxford region: time trend analysis. BMJ 315:713–717
6. Karvonen M, Pitkaniemi J, Tuomilehto J, The FinnishChildhood Type 1 Diabetes Registry Group (1999) The onset of Type 1 diabetes in Finnish children has becomeyounger. Diabetes Care 22:1066–1070
7. Schoenle E, Lang-Muritano M, Gschwend S et al. (2001)Epidemiology of type 1 diabetes mellitus in Switzerland:steep rise in incidence in under 5 year old children in thepast decade. Diabetologia 44:286–289
8. Karvonen M, Viik-Kajander M, Moltchanova E et al.(2000) Incidence of childhood type 1 diabetes worldwide.Diabetes Care 23:1516–1526
9. Tuomilehto J, Karvonen M, Pitkaniemi J et al. (1999)Record-high incidence of type 1 (insulin-dependent) diabe-tes mellitus in Finnish children. Diabetologia 42:655–660
10. Epidemiology Branch, HIC, Department of Health, GovWA (2002) Population characteristics of residents of theState of Western Australia
11. Kelly HA, Russell MT, Jones TW, Byrne GC (1994) Dra-matic increase in incidence of insulin dependent diabetesmellitus in Western Australia. Med J Aust 161:426–429
12. Kelly HA, Byrne GC (1992) Incidence of IDDM in West-ern Australia in children aged 0–14 yr from 1985 to 1989.Diabetes Care 15:515–517
13. LaPorte R, Tajima N, Akerblom H et al. (1985) Geographicdifferences in the risk of insulin-dependent diabetes melli-tus: the importance of registries. Diabetes Care 8 [Suppl 1]:101–107
14. LaPorte R, McCarty D, Bruno G, Tajima N, Baba S (1993)Counting diabetes in the next millenium: application ofcapture-recapture technology. Diabetes Care 16:528–534
15. World Health Organization Department of Noncommunica-ble Disease Surveillance Geneva (1999) Definition, diag-nosis and classification of diabetes mellitus and its compli-cations. Part 1: Diagnosis and classification of diabetesmellitus. Report of a WHO consultation. (http://www.who.int/ncd/dia/dia_publications.htm-accessed Dec 2003)
16. Australian Bureau of Statistics (1985–2001) Estimated res-ident population by age and sex in statistical local areas,Western Australia. Australian Government Publishing Ser-vice, Canberra, Catalog No. 3203.5
17. Australian Bureau of Statistics (2002) Population by ageand sex, Western Australia. Australian Government Pub-lishing Service, Canberra, Catalog No. 3235.5
18. Rewers M, LaPorte R, King H, Toumilheto J (1988) Trendsin the prevalence and incidence of diabetes: insulin-depen-dent diabetes mellitus in childhood. World Health Stat Q41:179–189
19. Stolwijk AM, Straatman H, Zielhuis GA (1999) Studyingseasonality using sine and cosine functions in regressionanalysis. J Epidemiol Community Health 53:235–238
20. Craig M, Howard N, Silink M, Chan A (2000) The risingincidence of childhood type 1 diabetes in New SouthWales, Australia. J Pediatr Endocrinol Metab 13:363–372
21. Karvonen M, Pitkaniemi M, Pitkaniemi J et al. (1997) Sexdifference in the incidence of insulin-dependent diabetesmellitus: an analysis of the recent epidemiological data. Diabetes Metab Rev 13:275–291
22. Green A, Gale EA, Patterson C, for the EURODIAB ACEStudy Group (1992) Incidence of childhood-onset insulin-dependent diabetes mellitus: the EURODIAB ACE study.Lancet 339:905–909
23. Toth E, Lee K, Couch R, Martin L (1997) High incidenceof IDDM over 6 years in Edmonton, Alberta, Canada. Dia-betes Care 20:311–313
24. Sebastiani L, Visalli N, Adorisio E et al. (1996) A 5-year(1989–1993) prospective study of the incidence of IDDMin Rome and the Lazio region in the age-group 0–14 years.Diabetes Care 19:70–73
25. Diabetes Epidemiology Research International Group(1990) Secular trends in incidence of childhood IDDM in10 countries. Diabetes 39:858–864
26. Rangasami JJ, Greenwood DC, McSporran B, Smail PJ,Patterson CC, Waugh NR (1997) Rising incidence of type1 diabetes in Scottish children, 1984–93. Arch Dis Child77:210–213
27. Karvonen M, Tuomilehto J, Virtala E et al. (1996) Season-ality in the clinical onset of insulin-dependent diabetesmellitus in Finnish children. Am J Epidemiol 143:167–176
28. Verge C, Silink M, Howard N (1994) The incidence ofchildhood IDDM in New South Wales, Australia. DiabetesCare 17:693–696
CO
NFID
EN
TIA
L-F
or P
eer Review
Only
�
�
�
�
�
�
�������������������� ���������������� ��� �����������
������������������������������������� ��
�
�������� ������������
������ ������ �������������
������ �������� �� ����������
������� ! ���"� ���#��$��#��� �%��
��!������& ������$��#���� '�����(�$)�� *�+� ��������,�����'��� ���(�-�"��� ����,����"�� � ����*������#������� ����������# �"�'����#��������#(�������������# �"�'����#��������#(��#��.� )��� ������/�������$������ �(�/�������$������ ���������(��0*����� ��������'����#���"���#� � ��� ����������#(�.� )��� ������1�������!�(���2��(������*������#������� ����������# �"�'����#��������#(�������������# �"�
'����#��������#(��#��.� )��� ������/�������$������ �(�/�������$������ �������(�� !��#�*�+� ��������,�����'��� ���(�-�"��� ����,����"�� � �������) �(�-� 3� ��#*�+� ��������,�����'��� ���(�-�"��� ����,����"�� � ����*������#������� ����������# �"�'����#��������#(�������������# �"�'����#��������#(��#��.� )��� ������/�������$������ �(�/�������$������ ��
4���/��"�� ������(�-� "�! ���,�(��# �"#��"�� � ����(�-�) ���!������5�������
��
�
�
Diabetes Care
CO
NFID
EN
TIA
L-F
or P
eer Review
Only
Cyclical variation in the incidence of childhood Type 1 diabetes in
Western Australia (1985 – 2010)
Short running title: Cyclical peaks in Type 1 diabetes incidence
Authors:
Aveni Haynes MBBChir1,2
, Max K Bulsara PhD3, Carol Bower PhD
2, Timothy W Jones
MD1,2
, Elizabeth A Davis FRACP1,2
Affiliations:
1 Princess Margaret Hospital, Department of Endocrinology & Diabetes, Perth, Western
Australia, Australia
2 Telethon Institute for Child Health Research, Centre for Child Health Research, The
University of Western Australia, Western Australia, Australia
3 Institute of Health and Rehabilitation Research, University of Notre Dame, Western
Australia, Australia
Corresponding Author:
Dr Elizabeth Davis,
Department of Diabetes & Endocrinology,
Princess Margaret Hospital for Children,
GPO Box D184, Perth, Western Australia, 6840
Tel: +61 8 9340 8090
Fax: + 61 8 9340 8605
Email: [email protected]
Abstract word count: 106 words including subheadings
Main text word count: 1000 words including subheadings
Number of figures: 1
Number of tables: 0
Page 11 of 20 Diabetes Care
CO
NFID
EN
TIA
L-F
or P
eer Review
Only
Abstract
Aim
To examine the incidence of childhood Type 1 diabetes in Western Australia from 1985 to
2010.
Research Design and Methods
Incidence rates were calculated for children aged 0-14 years and analysed by calendar year,
gender and age at diagnosis.
Results
There were 1873 cases and the mean incidence was 18.1/100,000 person years (95%CI:17.5–
19.2). The incidence increased by 2.3% a year (95%CI:1.6%-2.9%) with a sinusoidal 5-year
cyclical variation of 14%(95%CI:7%–22%). The lowest rate of increase in incidence was
observed in 0-4 year olds.
Conclusions
The cyclical pattern in incidence observed provides strong evidence for the role of
environmental factors in childhood Type 1 diabetes.
Page 12 of 20Diabetes Care
CO
NFID
EN
TIA
L-F
or P
eer Review
Only
Background
Between 1985 and 2002, the incidence of Type 1 diabetes in children aged 0-14 years in
Western Australia increased by an average of 3.2% a year[1]. Since then, the continuing rise
in incidence of childhood Type 1 diabetes has been reported elsewhere in Australia[2] and
worldwide[3]. This study aimed to update the data available on the incidence of childhood
Type 1 diabetes in Western Australia and to study patterns of incidence over the entire period
of observation.
Research Design and Methods
Cases were ascertained from the Western Australia Children’s Diabetes database, a
prospective population-based diabetes register established in 1987, which is estimated to be
>99% complete[1]. All children diagnosed with Type 1 diabetes using a combination of
clinical, biochemical and autoantibody assay findings, aged <15 years and resident in
Western Australia at the time of diagnosis were included in the study. Patients with Maturity
Onset Diabetes of the Young (MODY), Type 2 diabetes or secondary diabetes were
excluded. This study received approval from the Princess Margaret Hospital Ethics
Committee.
Incidence rates were calculated using annual population estimates from the Australian Bureau
of Statistics[4]. Poisson regression was used to analyse the incidence rates by calendar year,
gender, age group at diagnosis (0-4, 5-9 and 10-14 years) and to estimate the temporal trends.
To analyse the incidence for non-linear variation, sine and cosine functions were applied to
Poisson regression models for 3, 4, 5, 6, and 7 year cycles and the Akaike Information
Criterion (AIC) used to assess goodness of fit.
Page 13 of 20 Diabetes Care
CO
NFID
EN
TIA
L-F
or P
eer Review
Only
Results
There were 1873 cases (923 girls, 950 boys) of Type 1 diabetes diagnosed in children aged
<15 years in Western Australia between 1985 and 2010. Of these, 382(20%) were diagnosed
aged 0-4 years, 714(38%) between 5-9 years and 777(42%) between 10-14 years.
From 1985 to 2010, the mean age standardised incidence rate was 18.1/100,000 person years
(95%CI:17.5–19.2), with the incidence ranging from 11.3/100,000 in 1985 to 25.5/100,000 in
2003 (Figure). The incidence increased by an average of 2.3% a year (95%CI:1.6%-2.9%).
There was a significant sinusoidal cyclical variation in incidence of 14%(95%CI:7%–22%,
p<0.001) with a 5-year cycle providing the best model fit (AIC 7.518) (Figure). A sinusoidal
5-year cyclical variation in incidence was observed in both boys and girls and in all age
groups.
There was no difference in the mean incidence or incidence rate trends between boys and
girls. From 1985 to 2010, the mean age standardised incidence was 17.7/100,000 person
years (95%CI:16.9–19.3) in boys and 18.5/100,000 person years (95%CI:17.4–19.8) in girls.
There was an average annual increase of 2.9%(95%CI:2.0%–3.8%) in boys and
1.5%(95%CI:0.6%–2.4%) in girls.
From 1985 to 2010, the mean incidence in 0-4 year olds was 11.0/100,000 person years
(95%CI:10.3–12.6), which was significantly lower than the mean incidence of 21.1/100,000
(95%CI:19.5–22.6) in 5-9 and 25.5/100,000 (95%CI:20.8–23.9) in 10-14 year olds. Over the
same period, there was an average annual increase in incidence of 1.3%(95%CI: 1.0%-2.6%)
in 0-4 year olds, 2.4%(95%CI:1.4%-3.5%) in 5-9 year olds and 2.5%(95%CI:1.5%-3.5%) in
10-14 year olds.
Page 14 of 20Diabetes Care
CO
NFID
EN
TIA
L-F
or P
eer Review
Only
Conclusions
In Western Australia, the incidence of Type 1 diabetes in children aged 0-14 years increased
by an average of 2.3% a year between 1985 and 2010 and a significant 5-year cyclical pattern
in the incidence rate trend was observed.
Non-linear temporal incidence rate trends of childhood Type 1 diabetes have recently been
reported in some populations[6], and a levelling off in incidence has been reported in
others[7]. Epidemics, or peaks in the incidence of childhood Type 1 diabetes have been
reported in several populations over the past few decades[8-10]. A 4-year cyclical pattern in
the incidence of childhood Type 1 diabetes has been reported in Yorkshire, England[11], and
a 6-year cyclical pattern was recently reported in North-East England[5].
In Western Australia, a significant 5-year cyclical pattern in incidence was observed between
1985 and 2010. Of particular interest, almost identical peak and trough incidence years were
found in Western Australia as those reported in North-East England, despite the two
populations having very different demographic and climatic conditions. Being on opposite
hemispheres, these populations would experience opposite seasons and infectious disease
cycles. Consistent with findings in many European populations, a seasonal variation in the
incidence of childhood Type 1 diabetes in Western Australia has been reported, with a higher
number of patients diagnosed in the cooler winter and autumn months [1]. The data
presented in this study and that reported in North East England, are the annual incidence rates
and do not illustrate differences in seasonal variation in incidence that may occur due to the
populations experiencing opposite seasons.
Factors that have a cyclical nature, and which might modify the risk of childhood Type 1
Page 15 of 20 Diabetes Care
CO
NFID
EN
TIA
L-F
or P
eer Review
Only
diabetes in both populations include infections. Viral infections, in particular enterovirus
infections, are thought to have a role in the aetiology of Type 1 diabetes[12,13]. One early
study linked the cyclical pattern in incidence of childhood Type 1 diabetes with the cyclical
variation in number of reported mumps virus infections [14], but these findings have not been
replicated. As well as infections, the climate [15] and cyclical weather patterns may play a
role by altering lifestyle or environmental factors which modify the risk of childhood Type 1
diabetes. An alternate explanation for the cyclical pattern observed is that a trough in
incidence may follow peak years due to an exhaustion of individuals at risk of the disease.
The cyclical pattern in incidence observed in Western Australia provides strong evidence for
the role of environmental factors in childhood Type 1 diabetes. These factors may either be
environmental risk or protective factors that modify the likelihood of developing Type 1
diabetes de novo, or of progressing to clinical Type 1 diabetes in those with established
autoimmune pre-diabetes. With similar peak and trough incidence years found in Western
Australia as have been reported in North East England, the next step is to try and identify
what factors could be influencing the risk of childhood Type 1 diabetes in these distinct
populations during the same calendar years.
Author Contributions
A.H. researched data, wrote the manuscript, contributed to the discussion and edited the final
manuscript. M.B. researched data and reviewed the manuscript. C.B. contributed to the
discussion and reviewed/edited the manuscript. T.W.J. reviewed the manuscript. E.A.D
contributed to the discussion and reviewed/edited the manuscript.
Page 16 of 20Diabetes Care
CO
NFID
EN
TIA
L-F
or P
eer Review
Only
Acknowledgements
The authors thank Richard J. Q. McNally (Institute of Health and Society, Newcastle
University, England) for providing information regarding the use of Poisson regression to
assess cyclical variation of incidence rate trends in their study [5].
Guarantor
Dr Elizabeth A. Davis is the guarantor of this work and, as such, had full access to all of the
data in the study and takes responsibility for the integrity of the data and the accuracy of the
data analysis.
Conflicts of Interest
A Haynes: None declared
M.K. Bulsara: None declared
T.W. Jones: None declared
C Bower: None declared
E.A. Davis: None declared
References
1. Haynes A, Bower C, Bulsara MK, Jones TW and Davis EA. Continued increase in the
incidence of childhood Type 1 diabetes in a population-based Australian sample (1985-2002).
Diabetologia 2004; 47:866-870
2. AIHW. Incidence of Type 1 diabetes in Australian children 2000-2008. Canberra,
Australian Government Publishing Service, 2010 (Diabetes series no.13 Cat. no. CVD 51)
3. Patterson CC, Dahlquist GG, Gyürüs E, Green A and Soltész G. Incidence trends for
childhood type 1 diabetes in Europe during 1989-2003 and predicted new cases 2005-20: a
multicentre prospective registration study. Lancet 2009; 373:2027-2033
4. Australian Bureau of Statistics. Estimated Resident Population by Age and Sex in
Page 17 of 20 Diabetes Care
CO
NFID
EN
TIA
L-F
or P
eer Review
Only
Statistical Local Areas, Western Australia, 1985 - 2001 (Catalog No. 3203.5)
5. McNally RJQ, Court S, James PW, Pollock R and Blakey K. Cyclical Variation in Type 1
Childhood Diabetes. Epidemiology 2010; 21:914
6. Harjutsalo V, Sjöberg L and Tuomilehto J. Time trends in the incidence of type 1 diabetes
in Finnish children: a cohort study. Lancet 2008; 371:1777-1782
7. Berhan Y, Waernbaum I, Lind T, Möllsten A and Dahlquist G. Thirty Years of
Prospective Nationwide Incidence of Childhood Type 1 Diabetes: The Accelerating Increase
by Time Tends to Level Off in Sweden. Diabetes 2011; 60:577
8. Rewers M, LaPorte RE, Walczak M, Dmochowski K and Bogaczynska E. Apparent
epidemic of insulin-dependent diabetes mellitus in Midwestern Poland. Diabetes 1987;
36:106-13
9. Bruno G, Merletti F, Biggeri A, Cerutti F, Grosso N, De Salvia A, Vitali E, Pagano G and
Piedmont Study Group for Diabetes Epdiemiology. Increasing trend of Type 1 diabetes in
children and young adults in the province of Turin (Italy). Analysis of age, period and birth
cohort effects from 1984 to 1996. Diabetologia 2001; 44:22-25
10. Nystrom L, Dahlquist G, Rewers M and Wall S. The Swedish childhood diabetes study.
An analysis of the temporal variation in diabetes incidence 1978-1987. Int J Epidemiol 1990;
19:141-6
11. Staines A, Bodansky H, Lilley H, Stephenson C, McNally R and Cartwright R. The
epidemiology of diabetes mellitus in the United Kingdom: The Yorkshire regional childhood
diabetes register. Diabetologia 1993; 36:1282-1287
12. Hyoty H and Taylor K. The role of viruses in human diabetes. Diabetologia 2002;
45:11353-1361
13. Yeung W-CG, Rawlinson WD and Craig ME. Enterovirus infection and type 1 diabetes
mellitus: systematic review and meta-analysis of observational molecular studies. BMJ 2011;
342
14. Sultz HA, Hart BA, Zielezny M and Schlesinger ER. Is mumps virus an etiologic factor
in juvenile diabetes mellitus? Journal of Pediatrics 1975; 86:654-6
15. Nystrom L, Dahlquist G, Ostman J, Wall S, Arnqvist H, Blohme G, Lithner F, Littorin B,
Schersten B and Wibell L. Risk of developing insulin-dependent diabetes mellitus (IDDM)
before 35 years of age: indications of climatological determinants for age at onset. Int J
Epidemiol 1992; 21:352-358
Page 18 of 20Diabetes Care
CO
NFID
EN
TIA
L-F
or P
eer Review
Only
Figure Legend
Figure: Incidence of childhood Type 1 diabetes in Western Australia from 1985 to 2010
( black squares = observed annual incidence; solid line = estimated 5-yearly cyclical trend;
dotted line = estimated linear trend; white stars = peak incidence years (1992, 1997, 2003)
reported in published study from North-East England (5) for comparison)
Page 19 of 20 Diabetes Care
CO
NFID
EN
TIA
L-F
or P
eer Review
Only
�
�
��
��
��
��
��
���� ��� ���� ���� ���� ���� ��� ���� ���� ���� ���� ��� ���� ����
����
������������
� ��������� �����������
Page 20 of 20Diabetes Care
319
Appendix E: Outputs arising from Chapter 6
1. Conference and seminar presentations by Dr Aveni Haynes
2. PDF of published journal article Haynes A, Bulsara MK, Bower C, Codde JP, Jones TW and Davis EA. Independent effects of socioeconomic status and place of residence on the incidence of type 1 diabetes in Western Australia. Pediatric Diabetes, 2006. 7: 94-100
3. Co-author declarations for published work
320
321
Conference & Seminar Presentations by Dr Aveni Haynes
Oral presentations National August 2011 Temporal and spatial variation in the incidence of childhood
Type 1 diabetes in Western Australia (1985 - 2010) Australian Diabetes Society Annual Scientific Meeting, Perth
International September 2004 Independent effects of socio-economic status and place of
residence on the incidence of Type 1 diabetes in Australian children Annual Meeting of European Association for the Study of Diabetes, Munich
Local
March 2004 Analysis of incidence rates by socioeconomic status (Seminar presentation on methodological issues) Telethon Institute for Child Health Research, Perth
Poster presentations National September 2011 Temporal and spatial variation in the incidence of childhood
Type 1 diabetes in Western Australia (1985 - 2010) Australasian Epidemiological Association Annual Scientific Meeting, Perth
International November 2004 Both socio-economic status and place of residence are
associated with increased incidence of Type 1 diabetes in Western Australia International Society for Pediatric & Adolescent Diabetes Annual Meeting, Singapore
322
Original Article
Independent effects of socioeconomic statusand place of residence on the incidence ofchildhood type 1 diabetes in Western Australia
Haynes A, Bulsara MK, Bower C, Codde JP, Jones TW, Davis EA.Independent effects of socioeconomic status and place of residence on theincidence of childhood type 1 diabetes in Western Australia.Pediatric Diabetes 2006: 7: 94–100.
Objective: To analyze the incidence of type 1 diabetes in 0- to 14-yearolds in Western Australia, from 1985 to 2002, by region and socio-economic status.Methods: Primary case ascertainment was from the prospectivepopulation-based Western Australian Diabetes Register, and secondarycase ascertainment was from the Western Australian HospitalMorbidity Data System. The address at diagnosis was used to categorizecases into urban, rural and remote areas and into five socioeconomicgroups using the Index of Relative Socioeconomic Disadvantage.Denominator data were obtained from the Australian Bureau ofStatistics. Poisson regression was used to analyze the incidence rates byarea and socioeconomic status.Results: There were a total of 1143 cases (904 urban, 190 rural and 49remote). Case ascertainment was estimated to be 99.8% complete. Themean annual age-standardized incidence from 1985 to 2002 was 18.1 per100 000 person years in urban (95% CI: 16.3–19.9), 14.3 per 100 000 inrural (95% CI: 11.4–7.3) and 8.0 per 100 000 in remote areas (95% CI:5.8–10.3). The incidence was significantly higher in urban comparedwith rural (rate ratio 1.27, p ¼ 0.001) and remote (rate ratio 2.28,p < 0.001) areas. The incidence increased with higher socioeconomicstatus. The incidence in the highest socioeconomic group was 56%greater than the lowest socioeconomic group (rate ratio 1.56,p < 0.001). These differences in incidence by socioeconomic status andregion were independent of each other.Conclusions: Higher socioeconomic status and residence in the urbanarea are independently associated with an increased risk of childhoodtype 1 diabetes in Western Australia.
Aveni Haynesa,b,Max K Bulsarab,c,Carol Bowerb,Jim P Coddec,d,Timothy W Jonesa,b andElizabeth A Davisa,b
aDepartment of Endocrinology &Diabetes, Princess Margaret Hospital,Perth, Western Australia, Australia;bCentre for Child Health Research, TheUniversity of Western Australia, TelethonInstitute of Child Health Research, Perth,Western Australia, Australia; cSchool ofPopulation Health, The University ofWestern Australia, Perth, WesternAustralia, Australia; and dEpidemiologyBranch, Department of Health, Perth,Western Australia, Australia
Key words: Australia – childhood –epidemiology – socioeconomic factors –type 1 diabetes
Corresponding author:Dr Elizabeth Davis,Department of Endocrinology & Diabetes,Princess Margaret Hospital,GPO Box D184,Perth,Western Australia 6008,Australia.Tel.: þ61-8-9340-8090;fax: þ61-8-9340-8605;e-mail: [email protected]
Submitted 26 July 2005. Accepted forpublication 5 January 2006
There is a wide geographical variation in the incidenceof childhood type 1 diabetes both between countriesworldwide (1–3) and within some countries (4–6).The findings on regional variation in the incidencewithin countries have been inconsistent, with somestudies reporting a higher incidence in urban areas(5, 7), others reporting a higher incidence in ruralareas (4) and some reporting no regional variation(8, 9).
In Scotland, the urban/rural difference in incidenceof childhood type 1 diabetes was largely explained bydifferences in socioeconomic status between the areas(10). Several studies in the early 1980s reported agreater incidence of type 1 diabetes in individualswith higher socioeconomic status (11, 12). In contrast,a higher incidence of type 1 diabetes was observed inthe most deprived areas of northern England (13).Environmental factors are thought to have an
Pediatric Diabetes 2006: 7: 94–100 # 2006 The Authors. Journal compilation # 2006 Blackwell MunksgaardAll rights reserved
Pediatric Diabetes
94
important role in the etiology of type 1 diabetes ingenetically susceptible individuals (14, 15) and healthbehaviors, and risk factors are known to vary accord-ing to socioeconomic status in Australia (16).
Western Australia is the largest State in Australiawith a land area of over 2.5 million/km2 and an esti-mated total population of 1.9 million. The population ismainly composed of people of European descent, andIndigenous Australians account for just 3.5% of theState’s total population (17). Over 70% of the totalpopulation resides in and around the State’s capitalcity, Perth (Fig. 1). The remainder of the State is verysparsely populated, and a greater proportion of theIndigenous population resides in the rural and remoteareas.
Princess Margaret Hospital for Children is the onlytertiary pediatric hospital in the State and the onlyreferral centre for children diagnosed with diabetes.In Australia, children with diabetes are usually treatedby hospital-based pediatricians and endocrinologists.The diabetes team at Princess Margaret Hospital hasbeen running outpatient clinics in rural and remoteareas of the State since 1989. This enables case ascer-tainment levels of over 99% to be consistently achieved(18), and the complete population-based data set canbe used to accurately describe the characteristics of allnewly diagnosed patients in Western Australia.
The epidemiology of type 1 diabetes in 0- to 14-yearolds in Western Australia has previously beendescribed from 1985 to 2002, when it was found to
have increased significantly in both sexes and acrossall age groups (18). Preliminary analysis suggested agreater increase in the incidence in rural areas.Therefore, the aims of this study were to analyzethe incidence of type 1 diabetes in Western Australiain 0- to 14-year-old children from 1985 to 2002 byregion and socioeconomic status.
Methods
The primary source of cases was the WesternAustralian Children’s Diabetes database, a prospect-ive population-based register, which was establishedat Princess Margaret Hospital in 1987 and has beendescribed previously (19). Princess Margaret Hospitalis the only referral centre for children diagnosed withdiabetes in Western Australia, and the case ascertain-ment level of this database is over 99% (18, 20).
Secondary case ascertainment was from the WesternAustralian Hospital Morbidity Data System, which con-tains name-identified information on discharges fromall public and private sector, acute-care hospitals in theState. As the reporting of cases to these two sources iscompletely independent, the capture-recapture methodwas used to estimate completeness of ascertainment (21).
All patients diagnosed with type 1 diabetes by apediatric physician, according to the World HealthOrganisation’s definition (22), under the age of 15years and resident in Western Australia at the timeof diagnosis, were included in the study. The diagnosis
PERTH
0 400 800
WesternAustralia
NorthernTerritory
Queensland
SouthAustralia New South
Wales
Victoria
Tasmania
N
ACT
kilometers
Fig. 1. Population distribution of Western Australia. Each dot represents 1000 people. (produced by Epidemiology Branch, HIC, Departmentof Health, Western Australia in August 2004. Source: Australian Bureau of Statistics 2003 Estimated Resident Population).
Childhood type 1 diabetes in Australia
Pediatric Diabetes 2006: 7: 94–100 95
of type 1 diabetes was based on clinical presentationand the presence of islet cell autoantibodies and/orautoantibodies to glutamic acid decarboxylase. Thecontinuation of the clinical care of these patientsunder the same team until 16 years provides theopportunity to review and reconsider the type of dia-betes. Patients with diabetes secondary to other causessuch as cystic fibrosis, maturity-onset diabetes of theyoung and type 2 diabetes were excluded from thestudy. The date of diagnosis was defined as the dateof commencement of insulin therapy.
Analysis by place of residence
The residential postcode at the time of diagnosis was usedto categorize cases into urban, rural or remote areas, asdefined by the Western Australian Department ofHealth. The area definitions are based on populationdensity, as well as factors such as community infra-structure and service availability. In 2001, the populationdensity in urban areas of Western Australia was 172people/km2 compared with 0.43 people/km2 in ruralareas and 0.07 people/km2 in remote areas (23).
As there is a very low incidence of type 1 diabetes inchildren of Indigenous origin in Western Australia(24), and the Indigenous population is unevenly dis-tributed throughout the State, incidence rates for eacharea were calculated using both the total populationand the non-Indigenous population as denominators.
Annual population estimates, based on census dataand published by the Australian Bureau of Statistics(25), were obtained from the Western AustralianDepartment of Health by area and used as the totalpopulation denominator data.
Non-Indigenous population estimates were derived bysubtracting Indigenous population estimates, obtainedfrom the Department of Health’s Epidemiology Branch,from the annual population estimates published by theAustralian Bureau of Statistics. Indigenous populationestimates were made by the Epidemiology Branch basedon 2001 census counts, with annual adjustments forbirths and deaths (Codde JP, personal communication).
Analysis by socioeconomic status
The Index of Relative Socioeconomic Disadvantagewas used to analyze the incidence of childhood type 1diabetes in Western Australia by socioeconomic status.
This index, which is published every 5 years by theAustralian Bureau of Statistics, is based on censusdata and provides a summary measure of the socio-economic status of a geographical area. It is con-structed from variables such as the proportion ofindividuals in an area who are on a low income, areof Indigenous descent and are unemployed or in rela-tively unskilled occupations (26).
The census collection district is the smallest geo-graphical area for which the Index of RelativeSocioeconomic Disadvantage is available. It repre-sents the area covered by an individual census officer,which is approximately 250 dwellings in urban areasand proportionately less in rural and remote areas, asthe distance between dwellings increases (26).
The address at the time of diagnosis was usedto map each case to a census collection districtand obtain the Index of Relative SocioeconomicDisadvantage score for the census year closest to theyear of diagnosis. Cases were then categorized intofive socioeconomic groups using index scores thatdivided the total population of 0- to 14-year olds inWestern Australia into quintiles of disadvantage.
Census population counts from the 1991, 1996 and2001 censuses were obtained from the AustralianBureau of Statistics by sex, age and census collectiondistrict. Linear regression using the census counts wasused to interpolate the annual intercensus populationcounts, which were used as the denominator data tocalculate the incidence by socioeconomic status.
Ethics approval for this study was obtained fromthe Princess Margaret Hospital Ethics Committee andthe Western Australian Department of Health’sConfidentiality of Health Information Committee.Informed consent was obtained from all patientsprior to their data being stored on the diabetes data-base at Princess Margaret Hospital.
Statistical analyses
Incidence rates were calculated for each area andsocioeconomic group, using cases from both sourcesas the numerator data. Confidence intervals were esti-mated assuming a Poisson distribution of cases.
Poisson regression models were used to analyze thedifferences in incidence rates and incidence rate trends,using calendar year, area and socioeconomic groupin the models. To examine potential interactions, theanalysis of socioeconomic status was stratified by area.
Due to the small number of cases in rural andremote areas, cases from these areas were combinedfor some analyses and referred to as being non-urban.
A p-value of less than 0.05 was considered statisti-cally significant. Data analysis was performed usingthe STATA statistical software package (27).
Results
Ascertainment
From 1985 to 2002, there were a total of 1143 cases.As described previously (18), the case ascertainmentfor the study period was estimated to be 99.8%complete.
Haynes et al.
96 Pediatric Diabetes 2006: 7: 94–100
Place of residence
Of the total 1143 cases in Western Australia, 904occurred in the urban area, 190 occurred in ruralareas and 49 occurred in remote areas.
There were highly significant differences in themean age-standardized incidence of type 1 diabetesbetween all areas (Table 1). The overall incidence inthe urban area was 27% higher than in rural areas[incidence rate ratio (IRR) 1.27 (95% CI: 1.09–1.48),p ¼ 0.001] and 128% higher than in remote areas[IRR 2.28 (1.71–3.05), p < 0.001]. The overall inci-dence in rural areas was 80% higher than in remoteareas [IRR 1.80 (95% CI: 1.31–2.47), p < 0.001].
After adjusting for sex and age group, the incidencein the urban area was 44% higher than in non-urbanareas (IRR 1.44 (95% CI: 1.25–1.66), p < 0.001].
From 1992 to 2002, less than 0.01% of cases in thisstudy were of Indigenous ethnic origin. In 2001, inthe urban area, 3% of 0- to 14-year olds were ofIndigenous origin compared with 7% in the ruralareas and 26% in remote areas (28). To examine theimpact of this uneven distribution, the mean incidencewas calculated using just the non-Indigenous popula-tion as the denominator, and there was still a highlysignificant difference in the mean incidence betweenall areas (Table 1).
From 1985 to 2002, the incidence increased by anaverage of 2.5% a year (95% CI: 1.2–3.8%) in theurban area (p < 0.001) compared with 5.0% a year(95% CI: 2.4–7.6%) in the non-urban area (p < 0.001)(Fig. 2). This difference in incidence-rate trends overthe whole study period was not statistically significant(p ¼ 0.08). However, analyzed just over the last 10years, the rate of increase in incidence in the non-urban areas was significantly higher than that in theurban areas (p ¼ 0.02).
Socioeconomic status
Socioeconomic index values were obtained for 99%(900/904) of cases in the urban area and 94% (225/239) of cases in non-urban areas. Cases for which asocioeconomic index value could not be obtainedincluded those with an onset address that could notbe mapped to a census collection district, and those
that occurred in a census collection district for whichthe Australian Bureau of Statistics did not publish anindex value, a greater proportion of which are inremote areas (26).
There was a significant increase in the incidence oftype 1 diabetes with higher socioeconomic status(Fig. 3). For each increase in socioeconomic group,the incidence increased by an average of 9.9% (95%CI: 5.4–14.5%, p < 0.001). The incidence in socio-economic group 5, the least disadvantaged group,was 56% higher than in socioeconomic group 1,the most disadvantaged group [IRR 1.56 (95% CI:1.29–1.88), p < 0.001].
When the 904 cases in the urban area were analyzedseparately, the same trend was observed. In the urbanarea, for each increase in socioeconomic group, theincidence increased significantly by an average of10.7% (95% CI: 5.7–15.9%, p < 0.001). However, inthe non-urban area, no significant trend was observed[IRR 0.90 (95% CI: 0.81–1.01), p ¼ 0.06].
Independent effects of place of residence andsocioeconomic status
The effect of area on the incidence of type 1 diabetesin Western Australia was independent of differencesin socioeconomic status. After adjusting for
Table 1. Number of cases, total person years at risk, non-Indigenous person years at risk and mean incidence rate per100 000 person years (95% CI confidence level) in Western Australia, from 1985 to 2002, by geographical area
Urban Rural Remote
Number of cases 904 190 49Total person years at risk 4 943 792 1 318 302 612 259Mean incidence in total population 18.1 (16.3–19.9) 14.3 (11.4–17.3) 8.0 (5.8–10.3)Non-Indigenous person years at risk 4 798 106 1 228 998 450 985Mean incidence in non-Indigenous population 18.7 (16.8–20.6) 15.4 (12.2–18.7) 10.9 (7.8–14.0)
Year
1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
Inci
denc
e ra
te p
er 1
00 0
00 p
erso
n ye
ars
0
5
10
15
20
25
30
Fig. 2. Annual age-standardized incidence of type 1 diabetes in0- to 14-year olds in urban (~) and non-urban (*) areas ofWestern Australia.
Childhood type 1 diabetes in Australia
Pediatric Diabetes 2006: 7: 94–100 97
socioeconomic status, the incidence in the urban areawas 48% higher than in non-urban areas [IRR 1.48(95% CI: 1.27–1.71), p < 0.001].
In addition, the effect of socioeconomic status onthe incidence of childhood type 1 diabetes in WesternAustralia was independent of place of residence. Afteradjusting for area, for each increase in socioeconomicgroup, the incidence increased by an average of 7.3%(95% CI: 2.9–11.9%, p ¼ 0.001), and the incidence insocioeconomic group 5, the least disadvantagedgroup, was 42% higher than in socioeconomic group1, the most disadvantaged group [IRR 1.42 (95% CI:1.17–1.72), p < 0.001].
Due to the small number of cases in the non-urbanarea, adjustment for sex and age group was notpossible.
Discussion
This study reports significant independent effects of placeof residence and socioeconomic status on the incidence ofchildhood type 1 diabetes in the complete, predominantlyCaucasian, population of Western Australia.
The incidence in urban areas of Western Australia issignificantly higher than that in rural and remoteareas. This is in keeping with findings in Italy (5)and Lithuania, where the urban/rural difference wasfound to be most marked in 0- to 9-year olds (29). Incontrast, another Australian study in New SouthWales from 1990 to 1991 reported no significant geo-graphical variation in the incidence of childhood type1 diabetes (30). The difference in findings between ourstudy and that in New South Wales may be due to theuse of different definitions of urban/rural, or due todifferences in the urban/rural areas of the two States,which differ in their population distribution, topogra-phy and climate.
In Western Australia, the incidence was highest in theurban areas, which have the highest population density,and lowest in the sparsely populated remote areas.Conversely, a higher incidence of childhood type 1 dia-betes in rural compared with urban areas was reportedin Finland, where an inverse relationship between theincidence and population density was found (4).
Findings on the relationship between the incidence ofchildhood type 1 diabetes and population density areinconsistent, with a higher incidence found in the moredensely populated regions of Norway (7), but a lowerincidence found in areas with higher population densityin Northern Ireland (31) and Yorkshire, England (32).The relationship between the incidence of childhoodtype 1 diabetes and population density may be con-founded by socioeconomic status. In Scotland, forexample, a lower incidence in urban areas was explainedby increased deprivation in these areas (10).
In Western Australia, the incidence of childhoodtype 1 diabetes increased significantly with increasingsocioeconomic status, independent of place of resi-dence. A higher risk of childhood type 1 diabetes inthe least deprived, or most wealthy, social groups hasbeen reported by other studies (11, 12). An ecologicalanalysis of the incidence of childhood type 1 diabetesin Europe found that the incidence was positively cor-related with national indicators of prosperity, such asgross domestic product and low infant mortality (33).Compared with other European countries, the inci-dence of childhood type 1 diabetes increased morerapidly in the former socialist countries in CentralEurope, possibly related to an increase in wealth (34).No association between the incidence of childhoodtype 1 diabetes and social class was found in a studyin Southwest England (35) and Pittsburgh (36).
Comparison of the findings from different studieson the relationship between the incidence of type 1diabetes and socioeconomic status is difficult, due tothe use of different measures of socioeconomic statusthat may reflect different underlying environmentalrisk factors. The measure of socioeconomic statusused in this study (Index of Relative SocioeconomicDisadvantage) is a summary measure of the socio-economic status of a geographical area, based on thecharacteristics of individuals living in that area, ratherthan a measure of socioeconomic status at the indivi-dual level (26). The index was assigned to individualsat the census collection district level, which has beenshown to provide a more accurate measure of socio-economic status than if assigned at the larger, post-code level (37). Heterogeneity in the socioeconomicstatus of individuals living in the same geographicalarea would, if anything, result in an underestimationof the true relationship between socioeconomic statusand the incidence of type 1 diabetes.
The difference in incidence between urban, ruraland remote areas of Western Australia is not due to
0
5
10
15
20
25
1 2 3 4 5Socioeconomic group
Inci
denc
e pe
r 10
0 00
0
Fig. 3. Mean incidence of type 1 diabetes in 0- to 14-year olds inWestern Australia from 1985 to 2002, by socioeconomic group(1 ¼ most disadvantaged group, 5 ¼ least disadvantaged group).Error bars represent the upper 95% CI confidence level limit.
Haynes et al.
98 Pediatric Diabetes 2006: 7: 94–100
differing ascertainment levels, as case ascertainmentlevels have been comparable in all areas of WesternAustralia since the mid-1980s, and there are no differ-ences in accessibility to services provided by the dia-betes department at Princess Margaret Hospital.
In keeping with previous findings in WesternAustralia (24), this study reports a very low incidenceof type 1 diabetes in children of Indigenous descent.This study demonstrates that the regional variation inthe incidence of childhood type 1 diabetes in WesternAustralia is not explained by the differing proportionof Indigenous Australians, who have a reduced risk ofthe disease, in the population of these areas. In con-trast to the low incidence of type 1 diabetes inIndigenous Australians, there is an increased riskof childhood type 2 diabetes in this ethnic group(38). This may be indicative of differences in geneticpredisposition or environmental exposures betweenIndigenous and non-Indigenous Australians.
The differences in incidence of childhood type 1diabetes in Western Australia found by place of resi-dence and socioeconomic status are likely to be theresult of differences in environmental factors. Health-related behaviors have been shown to vary both bysocioeconomic group and by area in Australia (39).Potential risk factors for type 1 diabetes that may varyby socioeconomic status and place of residenceinclude diet and lifestyle, breast-feeding practices,increased maternal age, as well as factors in supportof the hygiene hypothesis (40), such as smaller familysize and different child-care practices.
There has been a significant increase in the inci-dence of type 1 diabetes in both urban and non-urban areas of Western Australia since 1985.Interestingly, there was some evidence of a greaterrate of increase in the incidence of type 1 diabetes innon-urban compared with urban areas, although thisonly reached statistical significance when analyzedover the last 10 years. The greater rate of increase inthe incidence of type 1 diabetes in non-urban areasmay reflect changes in the lifestyle or environment ofthese areas, as urban lifestyle factors become moreprevalent or accessible. Further studies are requiredto investigate regional differences in potentiallyimportant environmental exposures during childhood,including antenatal and perinatal risk factors.
Continued monitoring of the incidence of type 1diabetes in Western Australia is important for evalu-ating variations in the incidence and incidence ratetrends in this complete population-based data set,and investigating potential etiological factors.
Acknowledgments
We thank Hoan Nguyen and Gillian Smith (Centre for ChildHealth Research, University of Western Australia, Telethon
Institute for Child Health Research) for their invaluable assis-tance with mapping the cases to census collection districts andobtaining the appropriate socioeconomic index values. We alsoacknowledge Maree Grant (Department of Endocrinology &Diabetes, Princess Margaret Hospital) for management of theWestern Australian Children’s Diabetes database.
References
1. KARVONEN M, VIIK-KAJANDER M, MOLTCHANOVA Eet al. Incidence of childhood type 1 diabetes worldwide.Diabetes Care 2000: 23: 1516–1526.
2. EURODIAB ACE STUDY GROUP. Variation and trends inincidence of childhood diabetes in Europe. Lancet 2000:355: 873–876.
3. KARVONEN M, TUOMILEHTO J, LIBMAN I, LAPORTE RFOR THE WORLD HEALTH ORGANIZATION DIAMOND
PROJECT GROUP. A review of the recent epidemiologicaldata on the worldwide incidence of type 1 (insulin-dependent) diabetes mellitus. Diabetologia 1993: 36:883–892.
4. RYTKONEN M, MOLTCCHANOVA E, RANTA J, TASKINEN O,TUOMILEHTO J, KARVONEN M. The incidence of type 1diabetes among children in Finland – rural-urbandifference. Health Place 2003: 9: 315–325.
5. CHERUBINI V, CARLE F, GESUITA R et al. Large incidencevariation of type 1 diabetes in central-southern Italy1990–1995: lower risk in rural areas. Diabetologia1999: 42: 789–792.
6. YANG Z, WANG K, LI T et al. Childhood diabetes inChina. Enormous variation by place and ethnic group.Diabetes Care 1998: 21: 525–529.
7. JONER G, STENE LC, SOVIK O. Nationwide, prospectiveregistration of type 1 diabetes in children aged <15years in Norway 1989–1998: no increase but significantregional variation in incidence. Diabetes Care 2004: 27:1618–1622.
8. CASU A, PASCUTTO C, BERNARDINELLI L, SONGINI M.Type 1 Diabetes among Sardinian children is increasing:the Sardinian diabetes register for children aged 0–14years (1989–1999). Diabetes Care 2004: 27: 1623–1629.
9. HAMMAM R, GAY E, CRUICKSHANKS K et al. ColoradoIDDM registry incidence and validation of IDDM inchildren aged 0–17 yr. Diabetes Care 1990: 13: 499–506.
10. PATTERSON C, WAUGH NR. Urban/rural and depriva-tional differences in incidence and clustering of child-hood diabetes in Scotland. Int J Epidemiol 1992: 21:108–117.
11. SIEMIATYCKI J, COLLE E, CAMPBELL S, DEWAR R,AUBERT D, BELMONTE M. Incidence of IDDM inMontreal by ethnic group and by social class and com-parisons with ethnic groups living elsewhere. Diabetes1988: 37: 1096–1102.
12. TARN AC, GORSUCH AN, SPENCER KM, BOTTAZZO G,LISTER J. Diabetes and social class. Lancet 1983: 2:631–632 (Letter).
13. CROW YJ, ALBERTI KGMM, PARKIN JM. Insulin depen-dent diabetes in childhood and material deprivation innorthern England, 1977–1986. BMJ 1991: 303: 158–160.
14. SOLTESZ G. Diabetes in the young: a paediatric andepidemiological perspective. Diabetologia 2003: 46:447–454.
15. DAHLQUIST G. The aetiology of type 1 diabetes: anepidemiological perspective. Acta Paediatr Suppl 1998:425: 5–10.
16. TURRELL G, MATHERS C. Socioeconomic status andhealth in Australia. Med J Aust 2000: 172: 434–438.
Childhood type 1 diabetes in Australia
Pediatric Diabetes 2006: 7: 94–100 99
17. Epidemiology Branch, HIC, Department of Health,Western Australia. Population characteristics of resi-dents of the State. December 2002.
18. HAYNES A, BOWER C, BULSARA MK, JONES TW, DAVIS
EA. Continued increase in the incidence of childhoodtype 1 diabetes in a population-based Australian sample(1985–2002). Diabetologia 2004: 47: 866–870.
19. KELLY HA, BYRNE GC. Incidence of IDDM in WesternAustralia in children aged 0–14 yr from 1985 to 1989.Diabetes Care 1985: 15: 515–517.
20. KELLY HA, RUSSELL MT, JONES TW, BYRNE GC.Dramatic increase in incidence of insulin dependentdiabetes mellitus in Western Australia. Med J Aust1994: 161: 426–429.
21. LAPORTE R, MCCARTY D, BRUNO G, TAJIMA N, BABA S.Counting diabetes in the next millenium: application ofcapture-recapture technology. Diabetes Care 1993: 16:528–534.
22. WORLD HEALTH ORGANIZATION DEPARTMENT OF
NONCOMMUNICABLE DISEASE SURVEILLANCE GENEVA.Definition, diagnosis and classification of diabetes mel-litus and its complications. Part 1: diagnosis and classi-fication of diabetes mellitus. Report of a WHOconsultation. 1999, Accessed 10 September 2004, fromwww.who.int/diabetes/currentpublications/en/
23. AUSTRALIAN BUREAU OF STATISTICS. Census ofPopulation and Housing: Usual Residents Profile(Catalogue No. 2004.0). Canberra: AustralianGovernment Publishing Service, 2003.
24. GLATTHAAR C, WHITTALL D, WELBORN T et al. Diabetesin Western Australian children: descriptive epidemiol-ogy. Med J Aust 1988: 148: 117–123.
25. AUSTRALIAN BUREAU OF STATISTICS. Estimated ResidentPopulation by Age and Sex in Statistical Local Areas,Western Australia (Catalog No. 3203.5). Canberra:Australian Government Publishing Service, 1985–2001.
26. AUSTRALIAN BUREAU OF STATISTICS. Information Paper:Census of Population and Housing. Socio-economicIndexes for Areas. Australia 2001 (Catalog No.2039.0). Canberra: Australian Government PublishingService, 2001.
27. STATA CORPORATION. STATA statistical software (ver-sion 8.0).
28. Epidemiology Branch, HIC, Department of Health,Western Australia. Rates Calculator (version 9.1.3).
29. PUNDZIUTE-LYCKA A, URBONAITE B, OSTRAUSKAS R,ZALINKEVICIUS R, DAHLQUIST G. Incidence of type 1diabetes in Lithuanians aged 0–39 years varies by the
urban-rural setting, and the time change differs for menand women during 1991–2000. Diabetes Care 2003: 26:671–676.
30. VERGE C, SILINK M, HOWARD N. The incidence ofchildhood IDDM in New South Wales, Australia.Diabetes Care 1994: 17: 693–696.
31. PATTERSON C, CARSON D, HADDEN D and ON BEHALF OF
THE NORTHERN IRELAND DIABETES STUDY GROUP.Epidemiology of childhood IDDM in NorthernIreland: low incidence in areas with highest populationdensity and most household crowding 1989–1994.Diabetologia 1996: 39: 1063–1069.
32. STAINES A, BODANSKY H, MCKINNEY P et al. Smallarea variation in the incidence of childhood insulin-dependent diabetes mellitus in Yorkshire, UK: linkswith overcrowding and population density. Int JEpidemiol 1997: 26: 1307–1313.
33. PATTERSON C, DAHLQUIST G, SOLTESZ G, GREEN A andON BEHALF OF THE EURODIAB ACE STUDY GROUP. Ischildhood-onset type 1 diabetes a wealth-related dis-ease? An ecological analysis of European incidencerates. Diabetologia 2001: 44 (Suppl. 3): B9–B16.
34. GREEN A, PATTERSON C and ON BEHALF OF THE EURODIAB
TIGER STUDY GROUP. Trends in the incidence of child-hood-onset diabetes in Europe 1989–1998. Diabetologia2001: 44 (Suppl. 3): B3–B8.
35. BAUMER JH, HUNT LP, SHIELD JP. Social disadvantage,family composition, and diabetes mellitus: prevalenceand outcome. Arch Dis Child 1998: 79: 427–430.
36. LAPORTE R, ORCHARD T, KULLER L et al. The PittsburghInsulin Dependent Diabetes Mellitus Registry: the rela-tionship of insulin dependent diabetes mellitus incidenceto social class. Am J Epidemiol 1981: 114: 379–384.
37. HYNDMAN J, HOLMAN C, HOCKEY R, DONOVAN R, CORTI
B, RIVERA J. Misclassification of social disadvantagebased on geographical areas. comparison of postcodeand collector’s district analyses. Int J Epidemiol 1995:24: 165–176.
38. MCMAHON SK, HAYNES A, RATNAM N et al. Increase intype 2 diabetes in children and adolescents in WesternAustralia. Med J Aust 2004: 180: 459–461.
39. AUSTRALIAN INSTITUTE OF HEALTH and WELFARE.Australia’s Health 2004. Canberra: AIHW. Accessed1 October 2004, from http://www.aihw.gov.au/publications/index.cfm/title/10014
40. KOLB H, ELLIOTT R. Increasing incidence of IDDM aconsequence of improved hygiene? Diabetologia 1994:37: 729 (Letter).
Haynes et al.
100 Pediatric Diabetes 2006: 7: 94–100
327
Appendix F: Outputs arising from Chapter 7
1. Conference and seminar presentations by Dr Aveni Haynes
2. PDF of published journal article Haynes A, Bower C, Bulsara MK, Finn J, Jones TW and Davis EA. Perinatal risk factors for childhood Type 1 diabetes in Western Australia-a population-based study (1980-2002). Diabetic Medicine, 2007. 24(5): 564-570
3. Co-author declarations for published work
328
329
Conference & Seminar Presentations by Dr Aveni Haynes
Oral presentations International August 2005 Perinatal risk factors for childhood Type 1 diabetes in Western
Australia - a population based study (1980 – 2002) Short poster presentation International Society for Pediatric & Adolescent Diabetes Annual Meeting, Krakow
Local March 2006 Perinatal risk factors for childhood Type 1 diabetes in Western
Australia - a population based study (1980 – 2002) Telethon Institute of Child Health Research, Perth
September 2005 Perinatal risk factors for childhood Type 1 diabetes in Western
Australia - a population based study (1980 – 2002) Research and Advances Conference, Princess Margaret Hospital, Perth
March 2005 Linkage of the Western Australia Children’s Diabetes Database to the Midwives’ Notification System (1980 – 2002)
(Seminar presentation on methodological issues) Telethon Institute for Child Health Research, Perth
Poster presentations National September 2006 Perinatal risk factors for childhood Type 1 diabetes in Western
Australia - a population based study (1980 – 2002) Australasian Paediatric Endocrine Group Annual Scientific Meeting, Hobart
September 2005 Perinatal risk factors for childhood Type 1 diabetes in Western
Australia - a population based study (1980 – 2002) Australian Diabetes Society Annual Scientific Meeting, Perth
330
© 2007 The Authors.
564
Journal compilation © 2007 Diabetes UK.
Diabetic Medicine
,
24
, 564–570
DIABETIC
Medicine
DOI: 10.1111/j.1464-5491.2007.02149.x
Introduction
Although the aetiology of Type 1 diabetes (T1DM) remainsunknown, it is generally accepted to be the result of immune-mediated destruction of pancreatic B-cells caused by complexinteractions between genetic and environmental factors [1,2].Over the past two decades, the incidence of childhood T1DMin Western Australia has increased by an average of 3% a year
[3], similar to the rate of increase reported in many countriesworldwide [4,5]. The increase in incidence is too rapid to bedue to shifts in the population gene pool and is more likely tobe the result of changes in still unknown environmentalfactors. Environmental exposures that may have a role includeviruses, reduced exposure to microbial agents in early childhoodand early introduction of cow’s milk protein [6,7].
More recently, several studies have reported a greater rateof increase of T1DM in children under 5 years of age [4,8],suggesting that factors in early life, including the perinatalperiod, may be important in the aetiology of T1DM [6,7]. Manystudies have examined the relationship between the incidence
Correspondence to
: Dr Elizabeth A. Davis, Department of Endocrinology & Diabetes, Princess Margaret Hospital, GPO Box D184, Perth, Western Australia 6008. E-mail: [email protected]
Abstract
Aims
To investigate perinatal risk factors for childhood Type 1 diabetes inWestern Australia, using a complete population-based cohort.
Methods
Children born between 1980 and 2002 and diagnosed with Type 1diabetes aged
<
15 years (
n
=
940) up to 31 December 2003 were identifiedusing a prospective population-based diabetes register with a case ascertainmentrate of 99.8%. Perinatal data were obtained for all live births in WesternAustralia from 1980 to 2002 (
n
=
558 633) and record linkage performed toidentify the records of cases.
Results
The incidence of Type 1 diabetes increased by 13% for each 5-yearincrease in maternal age [adjusted incidence rate ratio (IRR) 1.13, 95% confidenceinterval (CI) 1.05, 1.21], by 13% for every 500-g increase in birth weight(adjusted IRR 1.13, 95% CI 1.04, 1.23). The incidence decreased with increasingbirth order (adjusted IRR 0.89, 95% CI 0.82, 0.96) and increasing gestationalage (adjusted IRR 0.84, 95% CI 0.77, 0.93). A higher incidence of Type 1diabetes was associated with an urban vs. non-urban maternal address at thetime of birth (adjusted IRR 1.38, 95% CI 1.18, 1.63), but no association wasfound with socio-economic status of the area.
Conclusions
A higher incidence of Type 1 diabetes was associated with increasingmaternal age, higher birth weight, lower gestational age, lower birth order andurban place of residence at the time of birth.
Diabet. Med. 24, 564–570 (2007)
Keywords
birth order, birth weight, gestational age, maternal age, Type 1 diabetes
Abbreviations
IRR, incidence rate ratio; T1DM, Type 1 diabetes
Blackwell Publishing, Ltd.Oxford, UKDMEDiabetic Medicine0742-3071Blackwell Publishing, 200724EpidemiologyEpidemiologyPerinatal risk factors for childhood Type 1 diabetes A. Haynes
et al
.
Perinatal risk factors for childhood Type 1 diabetes in Western Australia—a population-based study (1980–2002)
A. Haynes*†, C. Bower†, M. K. Bulsara†‡, J. Finn‡, T. W. Jones*† and E. A. Davis*†
*Department of Endocrinology & Diabetes, Princess Margaret Hospital, †Telethon Institute of Child Health Research, Centre for Child Health Research and ‡School of Population Health, The University of Western Australia, Perth, Australia
Accepted 21 December 2006
dme_2149.fm Page 564 Friday, April 13, 2007 6:15 PM
© 2007 The Authors.Journal compilation © 2007 Diabetes UK.
Diabetic Medicine
,
24
, 564–570
565
Epidemiology
DIABETIC
Medicine
of T1DM and various perinatal factors, but findings betweenstudies have been inconsistent.
Maternal age at delivery has been associated with anincreased risk of T1DM in several cohort studies [9–11].Higher birth weight has also been associated with an increasedrisk [9,12], but some studies have found no associationbetween the risk of T1DM and birth weight [13,14]. Pretermbirth has been associated with an increased risk of T1DM[9,13], but several studies have found no association betweenthe risk of T1DM and gestational age [10,12]. Similarly, adecreased [10] and increased [9,15] risk of T1DM has beenreported in first-born children, as well as no association withbirth order [16].
Western Australia presents a unique opportunity to investigate,in detail, perinatal factors associated with T1DM as complete,population-based data are available on all children diagnosedwith T1DM under the age of 15 years, and for all births in theState since 1980.
Perinatal data collection has been mandatory for all births inWestern Australia since 1980. Princess Margaret Hospital isthe only tertiary paediatric hospital in the State and the onlyreferral centre for children diagnosed with diabetes. Childrendiagnosed with T1DM can be identified from the WesternAustralia Children’s Diabetes Database, a prospective population-based diabetes register with a case ascertainment rate of99.8% [17].
Data from this register have shown an increase in theincidence of childhood T1DM in Western Australia of 3% ayear from 1985 to 2002 [3], and that higher socio-economicstatus and residence in urban areas at the time of diagnosis areindependently associated with an increased risk of T1DM inWestern Australian children [17]. This raises the question ofwhat environmental factors may explain these findings andwhether the effect of such factors on the risk of childhoodT1DM might vary according to timing of exposure. For example,do socio-economic status and place of residence at the time ofbirth have a similar association with the risk of T1DM thatwas found at the time of diagnosis?
Therefore, the aim of this study was to investigate theassociation between perinatal factors and childhood T1DM inWestern Australia, including socio-economic status and placeof residence at the time of birth, using a complete population-based cohort born between 1980 and 2002.
Patients and methods
Cases were defined as all children born in Western Australiabetween 1980 and 2002 who were diagnosed with T1DM up to31 December 2003 and were aged
<
15 years at diagnosis. Thediagnosis of T1DM was based on clinical presentation togetherwith the presence of one or more islet autoantibodies [glutamicacid decarboxylase (GAD
65
) and protein tyrosine phosphotase-likeprotein IA-2 (IA-2) determined by radioimmunoassay, or isletcell antibodies determined by an immunofluorescence assay].These patients are managed by the same team until they are atleast 16 years old, providing the opportunity to review and
reconsider the type of diabetes. Patients with maturity-onsetdiabetes of the young, Type 2 diabetes or diabetes secondaryto other causes such as cystic fibrosis were excluded from thestudy. Eligible cases were identified using the previously describedprospective, population-based Western Australian Children’sDiabetes Register at Princess Margaret Hospital [3].
Perinatal data were obtained from the Midwives’ Notifica-tion System for all live births in Western Australia from 1980to 2002. The Midwives’ Notification System contains data onover 99% of all births in Western Australia [18]. The data arecollected using a standard form, by midwives at the time of thebirth, and include details on the mother, infant, pregnancy,labour and delivery. In Australia, the definition of live birth is
≥
20 weeks’ gestation or birth weight
≥
400 g [19]. Birth orderwas determined from the number of previous births reported bythe mother at the time of birth of the index child. Data are avail-able on the presence of maternal diabetes prior to pregnancyand gestational diabetes occurring during the current pregnancy.However, no differentiation is made between pre-existingmaternal Type 1 and Type 2 diabetes. Maternal ethnicity wasself-reported at the time of birth. In the Midwives’ NotificationSystem, all people of Caucasoid descent (European heritage)are classified as being Caucasian. The non-Caucasian popula-tion in Western Australia is principally composed of indigenousAustralians and people of Asian descent.
Record linkage was performed by the Western AustraliaData Linkage Unit to identify perinatal records of the cases,using probabilistic matching of name, sex, date of birth andaddress variables, combined with an extensive manual reviewof doubtful links.
Study cohorts
Due to the potential confounding effects of ethnicity, pluralityand maternal diabetes [20,21], the effects of these factors on theincidence of T1DM were analysed for the whole cohort, consistingof all live births occurring during the study period.
The analysis of the incidence of T1DM by year of birth, birthweight, gestational age, maternal age and birth order was thenrestricted to singleton pregnancies of Caucasian mothers, withoutpre-existing or gestational diabetes.
Place of residence and socio-economic status at time of birth
The maternal address at the time of birth was used to analysethe incidence by area and socio-economic status of that area.
The post code was used to categorize births into urban andnon-urban areas, as defined by the Western Australia Depart-ment of Health. In Western Australia, the urban area consists ofthe metropolitan areas surrounding the capital city of Perth,with the remainder of the State being classified as non-urban.
In addition, the maternal address was used to analyse theincidence by socio-economic status. The Index of RelativeSocio-economic Disadvantage, published every 5 years by theAustralian Bureau of Statistics, is based on census data andprovides a summary measure of the socio-economic status of ageographical area [22]. It is constructed from variables such asthe proportion of individuals in an area who are on a low income,are of indigenous descent and are unemployed or in relatively
dme_2149.fm Page 565 Friday, April 13, 2007 6:15 PM
© 2007 The Authors.
566
Journal compilation © 2007 Diabetes UK.
Diabetic Medicine
,
24
, 564–570
DIABETIC
Medicine Perinatal risk factors for childhood Type 1 diabetes •
A. Haynes et al.
unskilled occupations. Each record was mapped to its corre-sponding census collection district and the Index of Relative Socio-economic Disadvantage score for this geographical area obtainedfrom the census year closest to the year of birth. Births werethen categorized into quintiles of socio-economic status basedon the scores of the study cohort. Data on socio-economicstatus were available only for births between 1984 and 1998.
These methods are similar to those used previously for theanalysis of these factors at the time of diagnosis [17].
Ethical approval for this study was obtained from thePrincess Margaret Hospital Ethics Committee and the WesternAustralia Confidentiality of Health Information Committee.Written, informed consent was obtained from all patients priorto their data being stored on the diabetes database at PrincessMargaret Hospital.
Statistical analysis
All live births in Western Australia between 1980 and 2002contributed to person time under observation from birthuntil diagnosis of Type 1 diabetes, the age of 15 years or31 December 2003, whichever occurred earlier. Deaths underthe age of 15 years occurring during the study period werecensored.
Poisson regression modelling with aggregated data was usedto analyse incidence rates for differences between and trendsacross variable categories, and to test for interactions. Statisticaladjustment of potential confounders was carried out usingstandard Poisson regression methods [23]. Covariate selectionfor the multivariate analyses was based on known biologicalinterrelationships and optimal model fit. To test for differencesin effect according to age at disease onset, the interactionterm between each perinatal factor and age group at diagnosis(
<
5 years, 5–14 years) was also examined.Seasonality of month of birth was analysed using the method
described by Stolwijk
et al
. [24], which is based on rates andtherefore controls for seasonal variation of births in the generalpopulation.
A two-tailed
P
-value of
<
0.05 was considered statisticallysignificant. Data analysis was performed using STATA statis-tical software (Stata Corp., College Station, TX, USA; version9.0).
Results
Data linkage
There were 940 eligible cases identified from the diabetesregister at Princess Margaret Hospital. Of these, 926 (98.5%)were successfully linked to the Midwives’ Notification System.Perinatal data were obtained for 926 cases and 557 707non-cases (Table 1).
All live births
Pre-existing maternal diabetes, gestational diabetes, ethnicity,
plurality
A significantly higher incidence of T1DM was found in babiesborn to mothers with pre-existing diabetes [incidence rateratio (IRR) 4.74, 95% confidence interval (CI) 2.93, 7.66;
P
<
0.001] and those with Caucasian compared with non-Caucasian mothers (IRR 3.73, 95% CI 2.61, 5.34;
P
<
0.001).The increased incidence of T1DM in babies born to motherswith gestational diabetes was not significant, even after adjustingfor maternal age and year of birth (IRR 1.47, 95% CI 0.89,2.42;
P
=
0.13).No significant difference in the incidence of T1DM was
found between babies from singleton compared with multiplepregnancies (IRR 0.86, 95% CI 0.58, 1.27;
P
=
0.45).
Month of birth
No significant seasonal variation was found when the incidenceof T1DM was analysed by month of birth for both genders
Table 1 Number of births by case, gender and age group at diagnosis for the cohort of all live births in Western Australia between 1980 and 2002, and the subcohort of singleton, Caucasian live births with no maternal pre-existing or gestational diabetes
All live birthsSingleton Caucasian live births,no maternal diabetes
Singleton Caucasian live births, no maternal diabetes(with complete data on all perinatal variables)
Cases n = 926 n = 840 n = 835Boys
0–4 years 147 135 1345–9 years 163 148 14710–14 years 143 132 129Total 453 415 410
Girls0–4 year 127 114 1145–9 year 217 199 19910–14 years 129 112 112Total 473 425 425
Non-cases n = 557 707 n = 466 004 n = 462 184Boys 286 584 239 268 237 314Girls 271 123 226 736 224 870
dme_2149.fm Page 566 Friday, April 13, 2007 6:15 PM
© 2007 The Authors.Journal compilation © 2007 Diabetes UK.
Diabetic Medicine
,
24
, 564–570
567
Epidemiology
DIABETIC
Medicine
combined (
χ
2
[3]
=
3.94,
P
=
0.27), or for boys and girlsanalysed separately.
Singleton, Caucasian live births, no maternal diabetes
Due to the effects of ethnicity, plurality and maternal diabeteson birth weight and gestational age, the remainder of theanalyses were restricted to singleton live births to Caucasianmothers with no pre-existing or gestational diabetes (840cases, 466 004 non-cases) (Table 1). The adjusted rate ratios
are based on the analysis of 835 cases and 462 184 non-casesfor which complete data on all covariates were available.
Without adjusting for other factors, the incidence of T1DMincreased with increasing maternal age and decreased withincreasing gestational age (Table 2).
After adjusting for the other variables, a higher incidence ofT1DM was associated with older maternal age, higher birthweight, lower gestational age and lower birth order (Table 2).The incidence of T1DM increased by an average of 13% forevery 5-year increase in maternal age, and by an average of
Table 2 Number of cases, total person-years at risk, incidence of T1DM (per 100 000 person-years), unadjusted and adjusted incidence rate ratios by maternal age, birth weight, gestational age, birth order and year of birth
T1DM cases(n = 840)
Totalperson-years
T1DM incidence(per 100 000)
Unadjusted Adjusted*
IRR (95% CI) P† IRR (95% CI) P
Maternal age at delivery (years)< 20 44 269 220 16.3 1.00 (0.73, 1.38) NS 0.92 (0.66, 1.27) NS20–24 169 1 181 992 14.3 0.88 (0.73, 1.06) NS 0.87 (0.72, 1.05) NS25–29 307 1 884 950 16.3 1.00‡ 1.0030–34 234 1 226 668 19.1 1.17 (0.99, 1.39) NS 1.18 (0.99, 1.40) NS≥ 35 86 425 598 20.2 1.24 (0.98, 1.58) NS 1.28 (1.00, 1.64) NSMissing 0 226Trend across categories 1.11 (1.04, 1.18) 0.003 1.13 (1.05, 1.21) 0.001
Birth weight (g)< 3000 144 958 126 15.0 0.90 (0.74, 1.09) NS 0.79 (0.64, 0.98) 0.043000–3499 311 1 853 775 16.8 1.00‡ 1.003500–3999 281 1 606 120 17.5 1.04 (0.89, 1.23) NS 1.09 (0.92, 1.28) NS≥ 4000 104 570 332 18.2 1.09 (0.87, 1.36) NS 1.19 (0.95, 1.49) NSMissing 0 302Trend across categories 1.06 (0.99, 1.14) NS 1.13 (1.04, 1.23) 0.003
Gestational age (weeks)< 37 53 273 348 19.4 1.17 (0.88, 1.56) NS 1.43 (1.04, 1.96) 0.0337–38 215 1 115 961 19.3 1.17 (0.99, 1.37) NS 1.24 (1.04, 1.47) 0.0139–40 438 2 652 022 16.5 1.00‡ 1.00> 40 134 915 656 14.6 0.89 (0.73, 1.08) NS 0.86 (0.71, 1.05) NSMissing 0 31 667Trend across categories 0.89 (0.82, 0.97) 0.008 0.84 (0.77, 0.93) < 0.001
Birth order1st 356 1 976 492 18.0 1.00‡ 1.002nd 265 1 679 817 15.8 0.88 (0.75, 1.03) NS 0.81 (0.69, 0.95) 0.013rd 149 771 166 17.1 0.95 (0.78, 1.15) NS 0.83 (0.68, 1.01) NS4th+ 65 435 980 14.9 0.83 (0.64, 1.08) NS 0.68 (0.52, 0.90) 0.01Missing 5 25 199Trend across categories 0.95 (0.89, 1.02) NS 0.89 (0.82, 0.96) 0.004
Year of birth§1980–1985 253 1 746 900 14.5 1.00‡ 1.001986–1991 326 1 782 314 18.3 1.26 (1.07, 1.49) 0.005 1.20 (1.02, 1.42) 0.031992–1997 213 1 112 727 19.1 1.32 (1.10, 1.59) 0.003 1.22 (1.01, 1.46) 0.041998–2002 48 346 713 13.8 0.96 (0.70, 1.30) NS 0.85 (0.62, 1.16) NSMissing 0 0Trend across categories 1.07 (0.99, 1.15) NS 1.03 (0.95, 1.11) NS
*Adjusted for all other listed variables.†NS, Non-significant (P > 0.05).‡The most numerous category was used as the baseline group for all variables except year of birth, for which the earliest birth cohort was used as the baseline.§1980–1985 birth cohort has complete 15 years of follow-up; follow-up to 15 years of age is increasingly incomplete with successive birth cohorts as case ascertainment is to 31 December 2003.
dme_2149.fm Page 567 Friday, April 13, 2007 6:15 PM
© 2007 The Authors.
568
Journal compilation © 2007 Diabetes UK.
Diabetic Medicine
,
24
, 564–570
DIABETIC
Medicine Perinatal risk factors for childhood Type 1 diabetes •
A. Haynes et al.
13% for every 500-g increase in birth weight. The incidence inbabies born
<
37 weeks was 43% higher than in those born at39–40 weeks, and the incidence decreased by an average of16% for each additional completed week of gestation. Afteradjusting for year of birth, maternal age and birth order, largepremature babies had the highest incidence of T1DM (Fig. 1).There was a significant decrease in incidence by an average of11% with each increase in birth order.
The incidence was significantly higher for births in 1986–1991 and 1992–1997 compared with those in 1980–1985(Table 2). However, no statistically significant trend in theincidence of T1DM was observed with year of birth for thestudy cohort overall. This may be due to incomplete 15-yearfollow-up of births after 1998, since cases in births up to 2002were included and follow-up ceased on 31 December 2003.
There was no significant interaction between age group atdiagnosis (
<
5 years, 5–14 years) and maternal age, birthweight, gestational age, birth order or year of birth, indicating nodifference in the effects of these variables by age of disease onset.
No statistically significant interactions were found bet-ween birth weight or gestational age and maternal age or birthorder. There were inadequate case numbers in all categories toassess the interactions between maternal age and birth order,and between birth weight and gestational age.
Place of residence at time of birth
The incidence of T1DM was 18.4 per 100 000 person-yearsin births with an urban (644 cases), and 13.3 per100 000 person-years in those with a non-urban (191 cases)maternal address at the time of birth. The incidence was 38%higher in births with an urban compared with non-urbanmaternal address (IRR 1.38, 95% CI 1.18, 1.63;
P
<
0.001),and this difference remained significant after adjusting for theperinatal variables examined in this study (IRR 1.33, 95% CI1.13, 1.56;
P
=
0.001).As significant interactions were found between place of
residence and year of birth, as well as birth order, models were
performed for each area separately. Associations between theperinatal variables and the incidence of T1DM, similar tothose described above for the whole cohort, were observed forbirths with an urban maternal address. Overall, similar trendswere also observed for births with a non-urban maternaladdress, but these did not reach statistical significance.
Socio-economic status at time of birth
Due to the limited availability of data on socio-economicstatus, the analysis of the incidence of T1DM by socio-economicstatus at birth was further restricted to births between 1984and 1998. Socio-economic Indexes for Area scores wereobtained for 90% of the singleton live births to Caucasianmothers with no diabetes that occurred during this time period(588 cases, 279 825 non-cases). There was an equal proportionof missing scores for cases and non-cases. Possible explana-tions for missing scores include having an address that couldnot be mapped to a census collection district, or being in acensus collection district for which the Australian Bureau ofStatistics did not publish an index value [22].
There was no significant difference in the incidence ofT1DM between the five quintiles of socio-economic status,based on the Index of Relative Socio-economic Disadvantage,and no significant trend across the categories (IRR 1.00, 95%CI 0.95, 1.06;
P
=
0.93). After adjusting for maternal age,birth weight, gestational age, birth order and year of birth,the effect of socio-economic status remained unchanged (IRR0.98, 95% CI 0.93, 1.04;
P
=
0.60). In addition, after adjustingeach of the perinatal variables for socio-economic status, theireffects on the incidence of T1DM remained unchanged.
Discussion
Using complete population-based data collected over 20 years,this study reports several significant associations betweenperinatal factors and the incidence of childhood T1DM inWestern Australia.
A fivefold higher incidence of T1DM was observed in babiesof mothers with pre-existing diabetes. While this probablyreflects an increased genetic susceptibility in these children,intrauterine exposure to hyperglycaemia may also contributeto their increased risk of T1DM [25]. A small, but non-significantincrease in the incidence of T1DM in babies of mothers withgestational diabetes, who would also be exposed to hypergly-caemia, was found in this study. This may be due to differencesin genetic susceptibility between babies born to mothers withpre-existing compared with gestational diabetes, or in thetiming and degree of intrauterine exposure to hyperglycaemia.
The incidence of T1DM was almost fourfold higher in theoffspring of Caucasian compared with non-Caucasian mothersin Western Australia [19]. This may be a reflection of either anincreased genetic susceptibility to T1DM in Caucasians [26] orprotection in non-Caucasians, but could also be a marker ofethnic differences in environmental and lifestyle factors includingdiet, antenatal care, breast feeding and child rearing practices.
FIGURE 1 Relationship between gestational age and birth weight and the incidence of Type 1 diabetes (�, < 3000 g; �, 3000–3499 g; ∆, 3500–3999 g; �, ≥ 4000 g; . . . ., linear trend).
dme_2149.fm Page 568 Friday, April 13, 2007 6:15 PM
© 2007 The Authors.Journal compilation © 2007 Diabetes UK.
Diabetic Medicine
,
24
, 564–570
569
Epidemiology
DIABETIC
Medicine
For singleton, Caucasian births to mothers without diabetesin this study, an increased risk of T1DM was found with oldermaternal age, higher birth weight, lower gestational age andlower birth order. These factors have been previously found tobe associated with T1DM in other populations. For example,older maternal age has been associated with a higher incidenceof T1DM in several populations [10,11,27]. This could be dueto increasing maternal age being a marker for factors associatedwith an increased risk of T1DM such as accumulated toxinsor exposure to infections [11], chromosomal aberrations [28]and complications of pregnancy.
Findings on the association between birth weight andT1DM have been inconsistent between studies in otherpopulations. Higher birth weight has been associated with anincreased risk of T1DM in several studies [9,10,12], althoughothers have found no association between birth weight andT1DM [13,14]. In this study, higher birth weight was asso-ciated with an increased incidence of T1DM and the highestrisk of T1DM was observed in large, premature babies. Areduced risk of T1DM has been previously observed in babiesborn small for gestational age [29]. The association betweenbirth weight and T1DM may be due to genetic factors that areassociated with both higher birth weight and risk of T1DM[30,31]. Alternatively, higher birth weight babies may havean altered metabolic profile that affects their future risk ofT1DM. For example, hyperinsulinaemia has been found inlarge-for-gestational-age babies of mothers without diabetesduring pregnancy [32]. The increased pancreatic B-cell activityassociated with hyperinsulinaemia may increase the risk ofimmune-mediated destruction of these cells [12].
Consistent with the findings of some studies [9,33], a lowerincidence of T1DM was found with increasing gestational agein this study. However, other studies have found no associa-tion between gestational age and the risk of T1DM [10,12].
Similarly, this study reports a lower incidence of T1DMwith increasing birth order which has also been previouslyreported [9,15], although again, others have found no associa-tion between birth order and the risk of T1DM [13,27]. Anincreased risk of T1DM in first-born compared with later bornchildren may be due to reduced exposure of first-born childrento infections, which are thought to be important for the appro-priate maturation of the developing immune system [34]. Inaddition, the association may be due to variation in factorssuch as feeding practices and child care arrangements whichvary with family size [11].
As it has been suggested that some exposures have a greatereffect in early-onset and others in later onset T1DM [9,35], theassociations between perinatal factors examined in this studyand T1DM were analysed for differences by age of diseaseonset. No significant difference was found in this study inthe effects of maternal age, birth weight, gestational age, birthorder or year of birth and the incidence of T1DM by age atdisease onset.
Residence in urban areas and higher socio-economic statusat the time of diagnosis have been shown to be independently
associated with an increased risk of T1DM in WesternAustralian children [17]. In this study, residence in urban areasat the time of birth was associated with a significantlyincreased incidence of T1DM. However, no association wasfound with socio-economic status at the time of birth.
The observed difference in incidence by place of residenceat the time of birth is not due to differing case ascertainmentbetween these areas, as children diagnosed with T1DM through-out the State are referred to Princess Margaret Hospital [17].The higher incidence of T1DM found in births with an urbanmaternal address at the time of birth is probably due todifferences in environmental exposures between urban andnon-urban areas of Western Australia. The association betweenT1DM and place of residence at both the time of diagnosis andthe time of birth suggests that unknown environmental factorsmay be acting to increase the risk of T1DM either
in utero
or the early postnatal period, as well as around the time ofdiagnosis.
The association previously found in Western Australia withsocio-economic status at time of diagnosis, but not found inthis study at the time of birth, is counter-intuitive. Perhapsthe area-based measure of socio-economic status used in thesestudies is an inadequate marker of factors related to T1DMrisk that vary by socio-economic status during the perinatalperiod, but is an adequate marker for environmental exposuresinvolved around the time of diagnosis.
Inconsistent findings between studies investigating therelationship between perinatal factors and T1DM may be dueto differences in study design or the use of sample sizes withinsufficient power to detect small effects. This study was basedon complete population-based data routinely recorded at thetime of birth for all live births in the State and so was notsubject to either recall or selection bias. However, a limitationof using routinely collected data is the lack of information onvariables which may be important confounders and could notbe accounted for such as smoking during pregnancy, paternalfactors such as age and ethnicity, maternal weight gain andindividual measures of socio-economic status.
The findings from this study are based on complete totalpopulation-based data over an extended period and providefurther evidence that factors during the perinatal period havean important role in the aetiology of childhood T1DM. Furtherstudies are required to identify factors which may be contributingto the increasing incidence of childhood T1DM in WesternAustralia and to elucidate possible causal pathways.
Competing interests
None to declare.
Acknowledgements
The authors thank Diana Rosman (Western Australia DataLinkage Unit), Vivien Gee (Manager, Midwives’ NotificationSystem), Peter Cosgrove and Margaret Wood (Centre for
dme_2149.fm Page 569 Friday, April 13, 2007 6:15 PM
© 2007 The Authors.
570
Journal compilation © 2007 Diabetes UK.
Diabetic Medicine
,
24
, 564–570
DIABETIC
Medicine Perinatal risk factors for childhood Type 1 diabetes •
A. Haynes et al.
Child Health Research, University of Western Australia,Telethon Institute for Child Health Research) for assistancewith the data linkage and advice regarding the perinatal dataused in this study. They also thank Hoan Nguyen (Centre forChild Health Research, University of Western Australia,Telethon Institute for Child Health Research) for providingthe appropriate socio-economic index values and Maree Grant(Department of Endocrinology & Diabetes, Princess MargaretHospital) for management of the Western AustralianChildren’s Diabetes Database. This study was funded by a2005 Research Grant awarded by the Diabetes AustraliaResearch Trust (DART). C.B. is partly funded by NHMRCFellowship no. 353628.
References
1 Atkinson MA, Eisenbarth GS. Type 1 diabetes: new perspectives ondisease pathogenesis and treatment.
Lancet
2001;
358
: 221–229.2 Devendra D, Liu E, Eisenbarth GS. Type 1 diabetes: recent develop-
ments.
BMJ
2004;
328
: 750–754.3 Haynes A, Bower C, Bulsara MK, Jones TW, Davis EA. Continued
increase in the incidence of childhood Type 1 diabetes in a population-based Australian sample (1985–2002).
Diabetologia
2004;
47
:866–870.
4 EURODIAB ACE Study Group. Variation and trends in incidence of childhood diabetes in Europe.
Lancet
2000;
355
: 873–876.5 The DIAMOND Project Group. Incidence and trends of childhood
Type 1 diabetes worldwide 1990–1999.
Diabet Med
2006;
23
: 857–866.
6 Soltesz G. Diabetes in the young: a paediatric and epidemiologicalperspective.
Diabetologia
2003;
46
: 447–454.7 Dahlquist G. The aetiology of type 1 diabetes: an epidemiological
perspective.
Acta Paediatrica Supplement
1998;
425
: 5–10.8 Gardner SG, Bingley PJ, Sawtell PA, Weeks S, Gale EA, the Bart’s-
Oxford Study Group. Rising incidence of insulin dependent diabetesin children aged under 5 years in the Oxford region: time trend analysis.
BMJ
1997;
315
: 713–717.9 Cardwell CR, Carson DJ, Patterson CC. Parental age at delivery,
birth order, birth weight and gestational age are associated with therisk of childhood Type 1 diabetes: a UK regional retrospective cohortstudy.
Diabet Med
2005;
22
: 200–206.10 Dahlquist G, Patterson C, Soltesz G, the EURODIAB Substudy 2
Study Group. Perinatal risk factors for childhood type 1 diabetes inEurope: the EURODIAB Substudy 2 Study Group.
Diabetes Care
1999;
22
: 1698–1702.11 Stene L, Magnus P, Lie R, Sovik O, Joner G and the Norwegian
Childhood Diabetes Study Group. Maternal and paternal age atdelivery, birth order, and risk of childhood onset type 1 diabetes:population based cohort study.
BMJ
2001;
323
: 369–371.12 Stene L, Magnus P, Lie R, Sovik O, Joner G and the Norwegian
Childhood Diabetes Study Group. Birth weight and childhood onsettype 1 diabetes: population based cohort study.
BMJ
2001;
322
:889–892.
13 Bache I, Bock T, Volund A, Buschard K. Previous maternal abortion,longer gestation and younger maternal age decrease the risk of Type1 diabetes in male offspring.
Diabetes Care
1999;
22
: 1063–1065.14 Kyvik KO, Bache I, Green A, Beck-Nielsen H, Buschard K. No
association between birth weight and Type 1 diabetes mellitus—atwin-control study.
Diabet Med
2000;
17
: 158–162.15 Wadsworth EJK, Shield JPH, Hunt LP, Baum JD. A case–control
study of environmental factors associated with diabetes in the under5s.
Diabet Med
1997;
14
: 390–396.16 Jones M, Swerdlow A, Gill L, Goldacre M. Pre-natal and early life
risk factors for childhood onset diabetes mellitus: a record linkagestudy.
Int J Epidemiol
1998;
27
: 444–449.17 Haynes A, Bower C, Bulsara M, Jones T, Davis E. Independent effects
of socioeconomic status and place of residence on the incidence oftype 1 diabetes in Western Australia.
Pediatr Diabetes
2006;
7
: 94–100.
18 Stanley FJ, Croft ML, Gibbins J, Read AW. A population databasefor maternal and child health research in Western Australia usingrecord linkage. Paediatr Perinat Epidemiol 1994; 8: 433–447.
19 Gee V, Green T. Perinatal Statistics in Western Australia, 2003.Twenty-first Annual Report of the Western Australian Midwives’Notification System. Perth, Western Australia: Department of Health2004.
20 Rosenberg TJ, Garbers S, Lipkind H, Chiasson MA. Maternalobesity and diabetes as risk factors for adverse pregnancy outcomes:differences among 4 racial/ethnic groups. Am J Public Health 2005;95: 1545–1551.
21 Hummel M, Schenker M, Ziegler A-G. Influence of perinatal factorson the appearance of islet autoantibodies in offspring of parents withtype 1 diabetes. Pediatr Diabetes 2001; 2: 40–42.
22 Australian Bureau of Statistics. Information Paper: Census of Popu-lation and Housing. Socio-economic Indexes for Areas. Australia2001 (Catalog Number 2039.0). Canberra: Australian GovernmentPublishing Service 2001.
23 Kianifard F, Gallo PP. Poisson regression analysis in clinical research.J Biopharm Stat 1995; 5: 115–129.
24 Stolwijk AM, Straatman H, Zielhuis GA. Studying seasonality usingsine and cosine functions in regression analysis. J Epidemiol CommunityHealth 1999; 53: 235–238.
25 Weiss P, Scholz H, Haas J, Tamussino K, Seissler J, Borkenstein M.Long-term follow-up of infants of mothers with type 1 diabetes:evidence for hereditary and non-hereditary transmission of diabetesand precursors. Diabetes Care 2000; 23: 905–911.
26 Dorman J, McCarthy B, McCanlies E, Kramer M, Vergona R, StoneR et al. Molecular type 1 diabetes epidemiology: Internationalstudies. Diabetes Res Clin Pract 1996; 34 (Suppl.): S107–116.
27 McKinney PA, Parslow R, Gurney KA, Law GR, Bodansky HJ,Williams R. Perinatal and neonatal determinants of childhood type 1diabetes: a case–control study in Yorkshire, UK. Diabetes Care 1999;22: 928–932.
28 Jongbloet PH, Groenewoud HMM, Hirasing RA, Buuren SV,Morrison E. Seasonality of birth in patients with childhood diabetesin the Netherlands. Diabetes Care 1998; 21: 190.
29 Dahlquist G, Bennich SS, Kallen B. Intrauterine growth pattern andrisk of childhood onset insulin dependent (type I) diabetes: popula-tion based case–control study. BMJ 1996; 313: 1174–1177.
30 Larsson HE, Lynch K, Lernmark B, Nilsson A, Hansson G, AlmgrenP et al. Diabetes-associated HLA genotypes affect birthweight in thegeneral population. Diabetologia 2005; 48: 1484–1491.
31 Weedon MN, Frayling TM, Shields B, Knight B, Turner T, MetcalfBS et al. Genetic regulation of birth weight and fasting glucose by acommon polymorphism in the islet cell promoter of the glucokinasegene. Diabetes 2005; 54: 576–581.
32 Evagelidou EN, Kiortsis DN, Bairaktari ET, Giapros VI, CholevasVK, Tzallas CS et al. Lipid profile, glucose homeostasis, bloodpressure, and obesity-anthropometric markers in macrosomic off-spring of nondiabetic mothers. Diabetes Care 2006; 29: 1197–1201.
33 Dahlquist G, Kallen B. Maternal–child blood-group incompatibilityand other perinatal events increase the risk for early-onset type-1(insulin-dependent) diabetes-mellitus. Diabetologia 1992; 35: 671–675.
34 Bach J-F. The effect of infections on susceptibility to autoimmune andallergic diseases. N Engl J Med 2002; 347: 911–920.
35 Leslie RDG, Castelli MD. Age-dependent influences on the origins ofautoimmune diabetes: evidence and implications. Diabetes 2004; 53:3033–3040.
dme_2149.fm Page 570 Friday, April 13, 2007 6:15 PM