review and economic evaluation hta health technology...
TRANSCRIPT
The clinical and cost-effectiveness of patienteducation models for diabetes: a systematicreview and economic evaluation
E LovemanC CaveC GreenP RoyleN DunnN Waugh
Health Technology Assessment 2003; Vol. 7: No. 22
HTAHealth Technology AssessmentNHS R&D HTA Programme
How to obtain copies of this and other HTA Programme reports.An electronic version of this publication, in Adobe Acrobat format, is available for downloading free ofcharge for personal use from the HTA website (http://www.hta.ac.uk). A fully searchable CD-ROM isalso available (see below).
Printed copies of HTA monographs cost £20 each (post and packing free in the UK) to both public andprivate sector purchasers from our Despatch Agents.
Non-UK purchasers will have to pay a small fee for post and packing. For European countries the cost is£2 per monograph and for the rest of the world £3 per monograph.
You can order HTA monographs from our Despatch Agents:
– fax (with credit card or official purchase order) – post (with credit card or official purchase order or cheque)– phone during office hours (credit card only).
Additionally the HTA website allows you either to pay securely by credit card or to print out yourorder and then post or fax it.
Contact details are as follows:HTA Despatch Email: [email protected]/o Direct Mail Works Ltd Tel: 02392 492 0004 Oakwood Business Centre Fax: 02392 478 555Downley, HAVANT PO9 2NP, UK Fax from outside the UK: +44 2392 478 555
NHS libraries can subscribe free of charge. Public libraries can subscribe at a very reduced cost of £100 for each volume (normally comprising 30–40 titles). The commercial subscription rate is £300 per volume. Please see our website for details. Subscriptions can only be purchased for the current orforthcoming volume.
Payment methods
Paying by chequeIf you pay by cheque, the cheque must be in pounds sterling, made payable to Direct Mail Works Ltdand drawn on a bank with a UK address.
Paying by credit cardThe following cards are accepted by phone, fax, post or via the website ordering pages: Delta, Eurocard,Mastercard, Solo, Switch and Visa. We advise against sending credit card details in a plain email.
Paying by official purchase orderYou can post or fax these, but they must be from public bodies (i.e. NHS or universities) within the UK.We cannot at present accept purchase orders from commercial companies or from outside the UK.
How do I get a copy of HTA on CD?
Please use the form on the HTA website (www.hta.ac.uk/htacd.htm). Or contact Direct Mail Works (seecontact details above) by email, post, fax or phone. HTA on CD is currently free of charge worldwide.
The website also provides information about the HTA Programme and lists the membership of the variouscommittees.
HTA
The clinical and cost-effectiveness ofpatient education models for diabetes:a systematic review and economicevaluation
E Loveman*
C CaveC GreenP RoyleN DunnN Waugh
Wessex Institute for Health Research and Development, University ofSouthampton, UK
*Corresponding author
Declared competing interests of the authors: none
Published August 2003
This report should be referenced as follows:
Loveman E, Cave C, Green C, Royle P, Dunn N, Waugh N. The clinical and cost-effectiveness of patient education models for diabetes: a systematic review andeconomic evaluation. Health Technol Assess 2003;7(22).
Health Technology Assessment is indexed in Index Medicus/MEDLINE and Excerpta Medica/EMBASE.
NHS R&D HTA Programme
The NHS R&D Health Technology Assessment (HTA) Programme was set up in 1993 to ensure that high-quality research information on the costs, effectiveness and broader impact of health
technologies is produced in the most efficient way for those who use, manage and provide care in the NHS.
The research reported in this monograph was commissioned by the HTA Programme on behalf of the National Institute for Clinical Excellence (NICE). Technology assessment reports are completedin a limited time to inform the appraisal and guidance development processes managed by NICE. The review brings together evidence on key aspects of the use of the technology concerned.However, appraisals and guidance produced by NICE are informed by a wide range of sources.
The research reported in this monograph was funded as project number 01/55/01.
The views expressed in this publication are those of the authors and not necessarily those of the HTA Programme, NICE or the Department of Health. The editors wish to emphasise that fundingand publication of this research by the NHS should not be taken as implicit support for anyrecommendations made by the authors.
HTA Programme Director: Professor Kent WoodsSeries Editors: Professor Andrew Stevens, Dr Ken Stein, Professor John Gabbay,
Dr Ruairidh Milne, Dr Chris Hyde and Dr Rob RiemsmaManaging Editors: Sally Bailey and Sarah Llewellyn Lloyd
The editors and publisher have tried to ensure the accuracy of this report but do not accept liability for damages or losses arising from material published in this report.
ISSN 1366-5278
© Queen’s Printer and Controller of HMSO 2003
This monograph may be freely reproduced for the purposes of private research and study and may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising.
Applications for commercial reproduction should be addressed to HMSO,The Copyright Unit, St Clements House, 2–16 Colegate, Norwich, NR3 1BQ.
Published by Gray Publishing, Tunbridge Wells, Kent, on behalf of NCCHTA.Printed on acid-free paper in the UK by St Edmundsbury Press Ltd, Bury St Edmunds, Suffolk.
Criteria for inclusion in the HTA monograph seriesReports are published in the HTA monograph series if (1) they have resulted from work commissioned for the HTA Programme, and (2) they are of a sufficiently high scientific quality as assessed by the referees and editors.
Reviews in Health Technology Assessment are termed ‘systematic’ when the account of the search,appraisal and synthesis methods (to minimise biases and random errors) would, in theory, permitthe replication of the review by others.
T
Objectives: To assess the clinical effectiveness andcost-effectiveness of educational interventions forpatients with diabetes, compared with usual care orother educational interventions.Data sources: Electronic databases, reference lists andexperts were all consulted in this study. Sponsorsubmissions to the National Institute of ClinicalExcellence were also reviewed.Review methods: Electronic databases weresearched, references of all retrieved articles werechecked for relevant studies, and experts werecontacted for advice and peer review and to identifyadditional published and unpublished references.Randomised clinical trials (RCTs) and controlled clinicaltrials (CCTs) were included if they fulfilled pre-specifiedcriteria, among which was follow-up from inception ≥ 12 months. Data were synthesised through anarrative review because the diversity of studiesprevented a meta-analysis. Results: Twenty-four studies (18 RCTs and six CCTs)that compared education with either a control groupor with another educational intervention wereincluded. The quality of reporting and methodologywas generally found to be poor by today’s standards.As part of treatment intensification, education in Type 1
diabetes (four studies) resulted in significant and long-lasting improvements in metabolic control andreductions in complications. In Type 2 diabetes (16studies) a diversity of educational programmes did notyield consistent results on measures of metaboliccontrol. Inconsistent results on metabolic control werealso found in studies of diabetes of either type (fourstudies), with studies of lower quality producingsignificant effects. Few studies evaluated quality of life.Economic evaluations comparing education with usualcare or other educational interventions were notidentified. Conclusions: Education as part of intensification oftreatment produces improvement in diabetic control inType 1 diabetes. Mixed results in Type 2 diabetes meanthat no clear characterisation is possible as to whatfeatures of education may be beneficial. Cost analysisand information from sponsor submissions indicatedthat where costs associated with patient educationwere in the region of £500–600 per patients, thebenefits over time would have to be very modest tooffer an attractive cost-effectiveness profile. Furtherresearch should focus on RCTs with clear designs basedon explicit hypotheses and with a range of outcomesevaluated after long follow-up intervals.
Health Technology Assessment 2003; Vol. 7: No. 22
iii
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Abstract
The clinical and cost-effectiveness of patient education modelsfor diabetes: a systematic review and economic evaluation
E Loveman,* C Cave, C Green, P Royle, N Dunn and N Waugh
Wessex Institute for Health Research and Development, University of Southampton, UK*Corresponding author
Contents
Health Technology Assessment 2003; Vol. 7: No. 22
v
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
List of abbreviations ........................................ vii
Executive summary ......................................... ix
1 Aim of the review ..................................... 1
2 Background ............................................... 3Description of underlying health problem ..................................................... 3Incidence and prevalence ......................... 4Current service provision .......................... 5Description of the interventions considered in this review ............................ 8
3 Methods ..................................................... 9Methods for reviewing effectiveness ......... 9Outcomes considered within clinical effectiveness sections ................................. 10
4 Effectiveness of interventions for Type 1 diabetes ..................................................... 13Background ............................................... 13Trials of self-management interventions .............................................. 13Assessment of effectiveness ....................... 16Summary of results from studies in Type 1 diabetes ...................................................... 21
5 Effectiveness of interventions for Type 2 diabetes ..................................................... 23Background ............................................... 23Trials of self-management interventions .. 23Trials of focused self-management interventions .............................................. 33Summary of results from interventions in Type 2 diabetes .......................................... 38Conclusion ................................................. 39
6 Effectiveness of interventions including patients with either Type 1 or Type 2 diabetes ..................................................... 41Trials of self-management interventions .. 41Assessment of effectiveness ....................... 43Summary of results from studies including patients with either Type 1 or Type 2 diabetes ...................................................... 44
7 Evidence from systematic reviews ............ 45Reviews of interventions in Type 1 diabetes ...................................................... 45
Reviews of interventions in Type 2 diabetes ...................................................... 45Reviews of interventions in diabetes generally .................................................... 46
8 Adverse effects .......................................... 49
9 Research progress ..................................... 51DAFNE trials ............................................. 51Other controlled trials ............................... 51Ongoing systematic reviews ...................... 51
10 Economic analysis ...................................... 53Overview of economic assessment ............ 53Methods ..................................................... 53Results of the systematic search for economic evaluations of patient education models for diabetes ................................... 53Assessing the cost-effectiveness of patient education models in diabetes .................... 56Critical appraisal of the cost-effectiveness analysis presented in the submission from the DAFNE Study Group to NICE ........... 58Southampton assessment of cost-effectiveness ....................................... 63
11 Discussion and conclusions ....................... 67Implications for other parties ................... 67Other issues and methodological concerns ..................................................... 68Strengths and limitations of the review ......................................................... 70Implications for further research .............. 70
Acknowledgements ................................... 73
References ................................................. 75
Appendix 1 Rapid review methods from the research protocol ................................. 81
Appendix 2 Sources of information, including databases searched and search terms .......................................................... 85
Appendix 3 Inclusion criteria worksheet ................................................... 89
Appendix 4 Details of excluded studies ........................................................ 91
vi
Appendix 5 Quality assessment scales for RCTs and CCTs ......................................... 97
Appendix 6 Psychological instruments used in included trials ............................... 101
Appendix 7 Data extraction: Type 1 diabetes ...................................................... 103
Appendix 8 Data extraction: Type 2 diabetes ...................................................... 119
Appendix 9 Data extraction: patients with either Type 1 or Type 2 diabetes ...................................................... 163
Appendix 10 List of reviews and systematic reviews retrieved ...................... 171
Appendix 11 Internal/external validity of economic evaluations ............................ 173
Appendix 12 Data extraction of economic evaluations ................................ 175
Appendix 13 Critical appraisal of health economic modelling studies in area of diabetes ......................................... 179
Appendix 14 Summary of educational interventions and comparators, with outline estimates on UK staffing costs ...... 185
Health Technology and Assessment reportspublished to date ....................................... 191
Health Technology and AssessmentProgramme ................................................ 199
Contents
AADE American Association of DiabetesEducators
ACD Association of ClinicalDiabetologists
ADA American Diabetes Association
ANCOVA covariance analysis
ANOVA analysis of variance
APT Automated Psychological Test
BDA British Diabetic Association(former name for Diabetes UK)
BDI Beck Depression Inventory
BG blood glucose
BGSM blood glucose self-monitoring
BMI body mass index
BMJ British Medical Journal
BP blood pressure
BDTTP basic diabetes treatment andtraining programme
CCT controlled clinical trial
CdR Condotte di Referimento
CRD Centre for Reviews andDissemination
CUA cost-utility analysis
CVD cardiovascular disease
DAFNE dose adjustment for normal eating
DCCT Diabetes Control andComplications Trial
DES diabetes education session
DKA diabetic ketoacidosis
DKNA Diabetes Knowledge Scale – form A
DLAY Diabetes Look After Yourself
DQOL diabetes quality of life measure
DSN diabetes specialist nurse
DTTP Diabetes Treatment and TeachingProgrammes
D(UK) Diabetes UK
EQ-5D EuroQol health state classificationquestionnaire
ESRD end-stage renal disease
ETDRS Early Treatment DiabeticRetinopathy Study
FBG fasting blood glucose
FF friends and family
FHQ Food Habits Questionnaire
FPG fasting plasma glucose
GFR glomerular filtration rate
GHb glycated haemoglobin
HbA1c glycated haemoglobin A1c
HDL high-density lipoprotein
Hypo hypoglycaemic episode
ICT intensified conventional treatment
IDDM insulin-dependent diabetesmellitus
Health Technology Assessment 2003; Vol. 7: No. 22
vii
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
List of abbreviations
viii
IDTTP intensive diabetes treatment andtraining programme
ITT intention-to-treat
LDL low-density lipoprotein
MANOVA multivariate analysis of variance
MRC Medical Research Council
N/A not applicable
NICE National Institute for ClinicalExcellence
NIDDM non-insulin-dependent diabetesmellitus
NR not reported
NS not statistically significant
OHA oral hypoglycaemic agent
OO one-to-one
Pt(s) patient(s)
QALY quality-adjusted life-year
QoL quality of life
QWB quality of well-being scale
RCT randomised controlled trial
SD standard deviation
SDIS Stockholm Diabetes InterventionStudy
SE standard error
SEM standard error of mean
SF-36 Short-Form 36 Health StatusQuestionnaire
SG standard gamble
SMBG self-monitoring of blood glucose
TTO time trade-off
UAER urinary albumin excretion rate
UGSM urine glucose self-monitoring
UKPDS United Kingdom ProspectiveDiabetes Study
VAS visual analogue scale
WESDR Wisconsin Epidemiologic Study ofDiabetic Retinopathy
YHEC York Health EconomicsConsortium
List of abbreviations
All abbreviations that have been used in this report are listed here unless the abbreviation is well known (e.g. NHS), or it has been used only once, or it is a non-standard abbreviation used only in figures/tables/appendices in which case the abbreviation is defined in the figure legend or at the end of the table.
Health Technology Assessment 2003; Vol. 7: No. 22
ix
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Description of the proposedserviceThis systematic review examines the clinical andcost-effectiveness of patient education models foradults with Type 1 or Type 2 diabetes.
Epidemiology and backgroundDiabetes mellitus (diabetes) is characterised by astate of chronic hyperglycaemia (raised bloodsugar). There are two main types of diabetes: Type 1and Type 2. Type 1 diabetes is an autoimmunecondition involving a process of destruction of thebeta cells of the pancreas, leading to severe insulindeficiency. About one-fifth of patients withdiabetes in England and Wales have Type 1diabetes. Type 2 diabetes is characterised byinsulin resistance and relative insulin deficiencyand is linked to being overweight or obese, and tophysical inactivity. Type 2 diabetes primarilyaffects people aged over 40 years. The basic targetin the treatment of diabetes is the normalisation ofblood glucose levels. Poor control of diabetes canin the short term result in diabetic ketoacidosis, aserious and potentially fatal condition, and in thelong term can increase the risk of complicationssuch as diabetic retinopathy and nephropathy.However, studies have shown that good diabeticcontrol is associated with a reduced risk of thesecomplications. Diabetic control is affected by bothlifestyle factors such as diet, and bypharmacological treatments, and the managementof diabetes is largely the responsibility of patients.A key component in empowering patients tomanage their own diabetes is education.
Education of patients with diabetes is considered afundamental aspect of diabetes care and aims toempower patients by improving knowledge andskills. Structured educational programmes fordiabetes self-management are often multifacetedinterventions providing patients with informationnot only about diabetes but also managementissues such as diet, exercise, self-monitoring ofblood glucose and medication use.
MethodsA systematic review of the literature and aneconomic evaluation were undertaken.
Data sourcesElectronic databases were searched, including theCochrane Library, MEDLINE, EMBASE, PubMed,Science Citation Index, Web of Science Proceedings,DARE and HTA databases, PsychINFO, CINAHL,NHS Economic Evaluation Database and EconLit.References of all retrieved articles were checkedfor relevant studies, and experts were contactedfor advice and peer review and to identifyadditional published and unpublished references.Sponsor submissions to the National Institute forClinical Excellence were reviewed.
Study selectionStudies were included if they fulfilled the followingcriteria:
� Interventions: educational interventionscompared with usual care or anothereducational intervention.
� Participants: adults with Type 1 or Type 2diabetes mellitus.
� Outcomes: must report glycated haemoglobin,hypoglycaemic episodes, diabetic complicationsor quality of life. Other reported outcomes fromincluded studies were discussed.
� Evaluation of outcomes �12 months frominception of intervention.
� Design: randomised clinical trials (RCTs), andcontrolled clinical trial (CCTs) with a concurrentcontrol were included.
� Reporting: studies were only included if theyreported sufficient detail of the intervention tobe reproducible (e.g. topics covered, whoprovided the education, how many sessionswere available).
Studies in non-English language or available onlyas abstracts were excluded.
Titles and abstracts were checked by two reviewers.Full texts of selected studies were assessed for
Executive summary
x
inclusion by one reviewer and checked by asecond. Differences in opinion were resolvedthrough discussion.
Data extraction and qualityassessmentData extraction and quality assessment wereundertaken by one reviewer and checked by asecond, with any disagreement resolved throughdiscussion involving a third reviewer if necessary.The quality of included studies was assessed inaccordance with Centre for Reviews andDissemination Report 4.
Data synthesisData on clinical effectiveness were synthesisedthrough a narrative review with tabulation of resultsfrom included studies. Studies were too diverse tobe combined in a meta-analysis. Cost-effectivenessanalyses were reported in a narrative review.
Number and quality of studiesSearches identified 24 studies comparingeducation with either a control group or withanother educational intervention. These were 18RCTs and six CCTs. Four studies included adultswith Type 1 diabetes, 16 studies included adultswith Type 2 diabetes and four studies includedadults with either Type 1 or Type 2 diabetes. Thequality of reporting and methodology of thestudies was generally poor by today’s standardswith only two RCTs reporting adequaterandomisation procedures and nonedemonstrating adequate allocation concealment.
Economic evaluationsLiterature searches identified only two studiesreporting cost-effectiveness results: one cost-utilityanalysis and one cost-effectiveness analysis usingintermediate outcomes only.
Summary of benefitsStudies of education in Type 1 diabetes suggestthat education programmes offered as a part ofintensified treatment interventions can result insignificant and long-lasting improvements inmetabolic control and reductions in complications.These are studies in which education is part of apackage of care also including treatment changes(for example diet and insulin) and therefore it isnot possible to draw conclusions about potentialeffects of education per se in Type 1 diabetes.
Diverse educational programmes in Type 2diabetes did not yield consistent results. Although
some trials reported significant improvements inmetabolic control and/or quality of life or otherpsychological outcomes, many others did notreport significant effects of educationalinterventions. No clear characterisation is possibleas to what features of education may be beneficialin this patient group.
Studies that included patients with either Type 1or Type 2 diabetes also produced mixed resultswith only poorer quality studies reportingsignificant effects.
CostsLiterature searches identified a small number ofstudies offering cost data in relation to patienteducation models. These were all studiesundertaken outside the UK and they covered avariety of methodologies. We are not able togeneralise from these studies as to the cost-effectiveness of patient education models. Patienteducation models will predominantly consist ofdirect costs for resource inputs to particulareducation packages, for example staff time(diabetes specialist nurse, dietitian and/orconsultant) and education materials. The DoseAdjustment for Normal Eating (DAFNE)intervention is estimated to cost approximately£545 per person attending.
Costs per life year gainedOwing to the absence of accurate data on healthoutcomes, we are not able to provide cost-effectiveness summary statistics. The evidence basedoes indicate that improved glycaemic control islikely to have a positive impact on the incidence oflong-term diabetic complications. Therefore,where the costs associated with patient educationare assumed to be in the region of £500–600 perpatient, the benefits over time would have to bevery modest to offer an attractive cost-effectivenessprofile for the intervention. The submission fromthe DAFNE study group predicts a scenario inwhich the DAFNE intervention results in costsavings and added health benefits over time, whencompared with usual practice.
ImplicationsThe main implication for the NHS would be stafftime, particularly of diabetes specialist nurses, butalso dietitians. Provision of increased education
Executive summary
Health Technology Assessment 2003; Vol. 7: No. 22
xi
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
may be hindered by a shortage of trained specialistnurses, which will take some years to resolve.
Future research needsThe paucity of high-quality trials that have testededucation per se in diabetes reveals a need for
more research. Such research should focus onRCTs with clear designs based on explicithypotheses and with a range of outcomesevaluated after long follow-up intervals. In orderto draw conclusions about the effects of educationalone, such trials should manipulate onlyeducation rather than confounding education withother factors.
The main aim of this review is to assess theclinical effectiveness and cost-effectiveness of
educational interventions for patients withdiabetes, compared with usual care or othereducational interventions. Potential benefitsinclude improved control of blood glucose (BG)levels as reflected in glycated haemoglobin(HbA1c), fewer short- and long-term complicationsof diabetes, better self-care and improved quality
of life (QoL) or well-being. Education may alsolead to improved knowledge of diabetes, althoughthis may not necessarily affect outcomes. Thereview does not cover educational interventionsaimed at preventing Type 2 diabetes. (The HTAprogramme has commissioned a review ofinterventions targeted at weight loss in peoplewith obesity and some included studies havelooked at Type 2 diabetes.)
Health Technology Assessment 2003; Vol. 7: No. 22
1
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Chapter 1
Aim of the review
Description of underlying healthproblemDiabetes mellitus (diabetes) is a state of chronichyperglycaemia (raised blood sugar), due to anabsolute or relative deficiency of insulin, ahormone for metabolism.
There are two main types of diabetes that aredistinguished by the pathological mechanisms:
� Type 1 diabetes is a condition in which most orall of the insulin-producing cells in the pancreashave been destroyed, usually due to an auto-immune process. Patients with Type 1 diabetesare ‘insulin dependent’ and need insulin forsurvival; it was formerly called insulin-dependent diabetes mellitus (IDDM).1
Type 1 diabetes generally appears before age40 years2 and is most often diagnosed inchildren and adolescents under age 15, but itcan occur at any age. The onset of the disease isusually fairly rapid, although the underlyingprocess may be slower.
� Type 2 diabetes is caused by a defect in the waythe body responds to insulin – insulin resistance– or by a relative reduction in insulinproduction or a combination of both. Thepancreas may initially produce more insulinthan normal in order to overcome the insulinresistance, but over time the production mayfail. This type of diabetes was formerly callednon-insulin-dependent diabetes mellitus(NIDDM).1
Type 2 diabetes primarily affects people overage 40 years, and tends to have a more gradualonset.2 Type 2 diabetes may be foundincidentally, for example at routine healthchecks.
Risk factors for Type 2 diabetes include beingoverweight, having a close relative with diabetesor having gestational diabetes duringpregnancy. It is more common in some ethnicgroups, particularly Asians. It is now being seenat younger ages.3,4
Other types of diabetes, including gestationaldiabetes and less common types such as maturityonset diabetes of the young, will not be addressedin this report. Diabetes can also be secondary to
other diseases such as pancreatitis or otherendocrine disorders.
The symptoms of diabetes include increased thirst,increased urination, extreme tiredness, weight loss,genital itching and blurred vision. Thesesymptoms are usually more pronounced in Type 1diabetes.2 Type 2 diabetes may be symptomless.
ComplicationsThe adverse effects of diabetes have traditionallybeen known as ‘complications’, although this termusually refers to effects that appear over thelonger term. The effects fall into three maingroups: acute metabolic upsets such asketoacidosis or hypoglycaemia; microvasculardisorders specific to diabetes; and an increasedrisk of large vessel disease such as heart disease.
� Ketoacidosis: without adequate supplies of insulinthe body cannot use glucose effectively, and maybreak down fat and muscle for energy in aninefficient way, leading to acidosis, a disturbanceof acid–base balance. Ketoacidosis requiresprompt hospital treatment, and can result incoma and occasionally death. Ketoacidotic comais more common in Type 1 diabetes. This is themost common cause of death for people withdiabetes under the age of 20.2
� Hypoglycaemia: means that BG has fallen toolow. This is chiefly caused by the inadequacy ofcurrent methods of insulin delivery, but can alsobe also due to too high a dose of oralhypoglycaemic agents (OHAs), inadequate foodintake or sudden or sustained exercise, and itcan occur without any apparent cause. It is notseen in patients controlled by diet alone. Earlysymptoms include shakiness, sweating andirritability. If not corrected by food or sugarydrinks, these can progress to confusion,faintness, headache and disturbances of vision.Hypoglycaemia can cause loss of consciousnessand convulsions if corrective steps are nottaken. For a small proportion of patientshypoglycaemic coma can occur frequentlyenough to be incapacitating.
More long-term or ‘late’ complications frompersistently raised BG levels include damage tolarge and small blood vessels and nerves.
Health Technology Assessment 2003; Vol. 7: No. 22
3
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Chapter 2
Background
� Microvascular: damage to small blood vessels(microangiopathy) can affect the eyes (diabeticretinopathy), kidneys (nephropathy) and nerves(neuropathy). Diabetes is the single mostcommon cause of blindness among adults aged16–64 years.2 Nephropathy may be in decline atleast in Type 1 diabetes, but kidney disease maydevelop in 20–25% of people with diabetes andmay progress to kidney failure.2 The principalforms of neuropathy are sensorimotor peripheralneuropathy and autonomic neuropathy.
� Macrovascular: damage to large blood vessels(macroangiopathy) can lead to ischaemic heartdisease, cerebrovascular disease, intermittentclaudication or gangrene of the feet. Patientswith diabetes have a 2–3-fold higher risk ofcoronary heart disease in men and a 4–5-foldincreased risk in premenopausal women.2
Stroke risk is increased 2–3-fold.2
People with diabetes are prone to foot ulcerationand gangrene of the lower limb (which can resultin amputation). Other complications can affect theskin, joints and tendons, gastrointestinal tract andsexual function. Diabetes also increases the risk ofcongenital malformations (both fatal and non-fatal) in babies of women with diabetes.
Mortality is higher in people with diabetes than inpeople of similar age and sex, although diabetes isnot usually recorded as the cause of death.Therefore, the contribution of diabetes to mortalityis likely to be four to five times greater thanreported in routine mortality statistics.5 The maincause of death is heart disease.6–8
ManagementThe first goal in the treatment of diabetes is thenormalisation of BG levels. There is goodevidence to show that tight control of BG andblood pressure (BP) can prevent or delay diabeticcomplications [United Kingdom ProspectiveDiabetes Study (UKPDS)9 and Diabetes Controland Complications Trial (DCCT),10 see Appendix4]. BG levels can be controlled by diet, oralhypoglycaemic drugs and/or insulin injections.
One of the features of diabetes care is that it aimsto empower the patient to take charge of thedisease. This is because of the chronic nature ofdiabetes and the relation between BG and factorssuch as diet and exercise (i.e. lifestyle). People withdiabetes must monitor BG levels, either directly orvia urine testing, take appropriate medicationand/or insulin, eat a healthy diet aimed at bothminimising BG levels and reducing future heartdisease risk, engage in activity or exercise to
maintain a healthy weight and to improve insulinsensitivity and avoid smoking.
Diet plays a major role in the management ofdiabetes. Patients are advised to have a high-carbohydrate, high-‘viscous’ fibre, low-fat and, ifoverweight, low-calorie diet. This kind of diet isdifficult for patients to maintain. Attention tofactors such as how rapidly different foods aremetabolised (as reflected in the ‘glycaemic index’of how rapidly BG levels rise after eating) can alsohelp, but adds another complexity to the diet.
Exercise also plays an important part in diabetesmanagement. In Type 1 diabetes the balancebetween insulin, food, and exercise must bemaintained if hypoglycaemia is to be avoided.Exercise helps overweight patients with Type 2diabetes bring their weight under control. Exercisecan be used as a mechanism for glycaemic control,particularly in patients who are not taking insulin.Exercise will increase insulin sensitivity, hencereducing insulin resistance.
Insulin therapies and regimens vary. Dependingupon the goals of therapy, the frequency of insulindosing can vary. Recent evidence that tight controlof BG levels can prevent or delay seriouscomplications has led to regimens that involvemore complex patterns of daily insulin treatment.Insulin pumps may be used to provide insulin on amore continuous basis with boluses at meal times.
Oral hypoglycaemic agents are often prescribed inType 2 diabetes. Most of these are sulphonylureas.These sensitise the insulin-secreting cells and mayupregulate insulin receptors and increase theirnumber.1 Biguanides also reduce BG by anothermechanism, which shows little dependence on theresidual effectiveness of insulin-secreting cells.1
Other oral agents, such as the glitazone drugs, areavailable and are used as an adjunct tosulphonylureas and biguanides. Sometimes,insulin and biguanide drugs are used incombination (e.g. for obese patients).
Incidence and prevalenceDiabetes is one of the most common chronicdisorders, but estimates of incidence andprevalence vary. Diabetes UK [D(UK)] estimatesthat about 1.4 million people in the UK todayhave diagnosed diabetes. It is thought that at least1 million more have diabetes, but have not beendiagnosed,11 although some suggest that this maybe an overestimate.12 The Audit Commission
Background
4
estimated that diabetes affects about 3% of thepopulation, not including those who areundiagnosed.13 The number of patients withdiagnosed diabetes has been increasingsignificantly in recent years in the UK andworldwide. It has been estimated that the numberof people with diabetes will rise from 1.4 millionto 3 million by 2010.13
� Type 1: the incidence of Type 1 diabetes variesgreatly worldwide from as high as 35 per 100,000 in some Scandinavian countries to 2per 100,000 in Japan. The incidences inScandinavia and the UK are higher than thosein France and Italy.1
If approximately 3% of the population havediabetes and 10–25% of these have Type 1diabetes, then based on 1999 populationestimates about 158,000–395,000 people inEngland and Wales have Type 1 diabetes.
� Type 2: this form is far more common thanType 1, but estimates for the proportion ofpeople with diabetes who have Type 2 variesfrom 75 to 90%.2 The Audit Commissionestimates that over 80% of cases are Type 2, withover 1 million people diagnosed in the UK.13
If approximately 3% of the population hasdiabetes, then based on 1999 populationestimates and assuming that between 75 and90% of patients with diabetes have Type 2,about 1,185,500–1,422,600 people in Englandand Wales have Type 2 diabetes.
Table 1 demonstrates the prevalence of insulin-and non-insulin-treated diabetes per 1000 patientsin 1998. It is important to note that insulin-treatedpatients are likely to be a mix of patients with Type 1 diabetes and patients with Type 2 diabetes.
Diabetes is more common in older people.Diabetes may affect as many as 6% of people aged
65 and over.2 The average age of diagnosis isabout 52 in people without a family history and 51in people with a family history.
Diabetes is slightly more common in men thanwomen. Diabetes seems to remove women’snatural protection against heart disease and strokebefore the menopause.
Diabetes, especially Type 2, tends to run infamilies. There is some suggestion, however, thatconcordance between twins might also arise fromshared environments, especially foetalenvironment.
Diabetes is three to five times more commonamong people of African–Caribbean and Asianorigin living in the UK. Diabetes in these groupstends to develop at a younger age and may berelated to different underlying mechanisms.
Current service provisionThe long-term care required for people withdiabetes is organised in different ways in differentareas. Traditionally, most patients have beentreated in a hospital diabetes clinic. However, withincreasing ‘shared care’, the care of more patientsis being shared between hospitals and generalpractice teams, although this applies mainly toType 2 diabetes. There are a number of differentmodels for shared care, with varying degrees ofinvolvement from primary care teams. In someareas there are district diabetes centres that aredevoted to the care of patients with diabetesthroughout the district.
Irrespective of whether patients are cared for inprimary care or by a hospital team, it is generallythought that the best care requires a group of
Health Technology Assessment 2003; Vol. 7: No. 22
5
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
TABLE 1 Prevalence of insulin- and non-insulin-treated diabetes per 1000 patients, by age and gender in 1998
Diabetes Age (years) 0–4 5–15 16–24 25–34 35–44 45–54 55–64 65–74 75–84 85+
Insulin- Malestreated Rate/1000 0.2 1.7 3.5 4.6 6.2 7.2 10 13.3 10.9 6.8
FemalesRate/1000 0.3 1.9 3.2 4.3 5.2 5.7 9.4 12.1 9.4 5.9
Non-insulin- Malestreated Rate/1000 0 0 0.2 0.6 3.6 11.8 30.5 47.5 47.4 43.1
FemalesRate/1000 0 0 0.2 0.6 2.8 7.9 20.3 35.7 37.1 33.8
Reproduced with permission from Office of National Statistics.
health care professionals including consultantphysician, diabetes specialist nurse (DSN),dietitian, podiatrist, general practitioner (GP) andpractice nurse. The skills of clinical psychologists,ophthalmologists, nephrologists, neurologists,vascular and orthopaedic surgeons, obstetricians,midwives and other specialists may be called on asnecessary.
The goals of management for patients withdiabetes include optimisation of BG control,prevention of immediate complications andprevention of long-term complications. Thedetails of management goals should be set bypatients and professionals in consultation.
EducationEducation of patients with diabetes is considered afundamental aspect of diabetes care. Becausepatients are responsible for the day-to-day controlof their diabetes, it is critical that patientsunderstand the condition and how to treat it. Allmembers of the diabetes care team play a role ineducation. Education can be on a one-to-one basisor in groups, or both. All contacts betweenpatients and practitioners can be an opportunityfor education.
D(UK)14 has produced a list of educational needsat initial diagnosis. Patients should be instructedabout the nature of the condition and itstreatment, be given advice on adapting lifestyleand be given counselling on the implications ofdiabetes. Education, however, needs to continuebeyond initial diagnosis and to involve access toteam members as needed.
DSNs play an important role in providing care.They educate, advise and counsel people withdiabetes about all aspects of living with diabetes.They are usually based at a hospital clinic ordiabetes centre but also liaise with generalpractices and visit patients in their homes.
Practice nurses also provide education and adviceand work to coordinate care among members ofthe team. They can also provide social andpsychological support for patients and families.
For patients treated with insulin, monitoring BGlevels is necessary to try to maintain levels asconsistently near normal as possible.9,10 BG can bechecked by means of a simple blood test or,indirectly, by testing the urine. Learning when andhow to monitor and how to interpret BG is animportant aspect of self-management, particularly
for insulin-treated patients, who are at risk fromhypoglycaemia and ketoacidosis.
All of the treatment factors, diet, medication andexercise, must be carefully managed on a dailybasis by patients themselves. Patients must also beable to recognise when they need professionalhelp. Good self-management depends on initialeducation about the interaction of all thetreatment factors and ongoing support andreinforcement. Patients must also be aware of thenecessity to monitor for complications such asdiabetic retinopathy and see that they areregularly screened for these complications.
Recommendations on education fromadvisory bodiesThe National Service Framework has recentlypublished recommendations for standards indiabetes care.15 Standard 3 states that all patientswith diabetes will receive a service that encouragespartnership in decision making, supports them inmanaging their diabetes and helps them to adoptand maintain a healthy lifestyle. It goes on to statethat the provision of information, education andpsychological support that facilitates self-management is the cornerstone of diabetes careand that structured education should be tailored tothe needs of the individual and include skills-basedapproaches. Such education should be rooted inprinciples of adult learning, including anappropriate mix of “didactic information giving,active learning, problem-based learning and skillsdevelopment, as well as group teaching sessions”.However, the word ‘appropriate’ is not defined.
The National Institute for Clinical Excellence(NICE) guidelines for Type 2 diabetes recommendthat patients with diabetes should be offerededucation on an ongoing basis, and that differentapproaches should be used until there is morecertainty about the most effective methods. Theirreview of the evidence shows that educationalprovision is better than no provision, and that it isunclear which type of education (e.g. didactic,patient-centred, computer-assisted) has the mostimpact on outcomes such as metabolic control orknowledge scores. The report points out thatmany of the reported interventions have beenpoorly described, without clear evidence ofunderlying psychological, behavioural oreducational theory. Furthermore, follow-upperiods have been short, and the patients in thestudies have been somewhat heterogeneous.
The Audit Commission report on diabetesservices13 also comes out strongly in favour of
Background
6
provision of education for all patients withdiabetes, and outlines some features of high-quality provision:
� a structured programme, including a writtencurriculum
� multidisciplinary delivery (including podiatristsand dietitians)
� varied modes of delivery (including both groupand one-to-one sessions)
� access for newly diagnosed and establishedpatients
� continuous assessment and a programme forestablished patients according to their needs
� access to all patients, regardless of who deliverscare
� built-in evaluation of each patient’s knowledgeand self-care.
Various professional bodies have publishedrecommendations for both the infrastructure andthe content of diabetes education programmes.For instance, diabetes organisations in the USArecently published national standards for diabetesself-management education16 in which basicorganisational goals were outlined and referenceswere made to detailed curricula available from theAmerican Diabetes Association (ADA) and theAmerican Association of Diabetes Educators(AADE). Similarly, the AADE has published aposition statement on the scope of practice anddiabetes educators and standards of practice fordiabetes educators.17 A similar formalisation ofgoals for diabetes care including education isincluded in the guides to diabetes mellitus fromthe European Diabetes Policy Group.18
Finally, the D(UK) website (www.diabetes.org.uk)advises patients that they should be offered aprogramme of care “that suits you”, and theyshould be offered education initially (on diagnosis)and on an ongoing basis. There is an emphasis onthe diabetes care team working in tandem with thepatient and allowing shared decisions of care,based on knowledge and agreed managementgoals for each individual. A page entitled “yourresponsibilities” also states that it is theresponsibility of each patient to learn about theirdiabetes and to know how to manage the disease,and when to ask for help.
Thus, there can be little doubt that education isseen to be a pivotal part of the managementstrategy for all patients with diabetes. However,there is much less agreement as to the best methodsby which this can be achieved, owing to an apparentpaucity of rigorous research on the subject.
How effectively is diabetes educationbeing provided at present?The Audit Commission report (2000)13 foundrather variable provision in the nine hospital truststhat they visited, with a particular lack of emphasison evaluation of the services that were provided.Only five of the trusts had a structuredprogramme, with a written curriculum, and themajority did not involve podiatrists routinely,despite the recognised importance of foot care.Because the nine hospital trusts in this report werechosen to be broadly representative of the range ofhospital services available for patients with diabetesacross the country, it is likely that the situationdescribed above applies generally. The provision ofeducational services in general practice was notsurveyed in the same detail in this report, andcould well be even less comprehensive because ofthe lack of necessary skills and facilities. Probablymost GPs would expect educational services to beprovided by their local district general hospital, butthis might prove a problem where patients withdiabetes are routinely discharged back to GP care.
Examples of currently available educationprogrammesIt is thought that most patients with diabetes inEngland and Wales are offered education, at leastat the time of diagnosis. Some examples ofprogrammes that are available are detailed below,but these may not have had formal evaluation. Inaddition, the extent to which these programmesare representative of current programmes acrossthe NHS is unknown and these may reflect ‘bestpractice’.
The Diabetes and Endocrine Centre of the RoyalBournemouth Hospital has structured educationprogrammes for patients with Type 1 and Type 2diabetes. They report that the majority of patientswith Type 2 diabetes take up the offer ofeducation, but that uptake is more limited for Type1 patients – often owing to work commitments.Nevertheless, 70% of newly diagnosed patients withType 1 diabetes since 1999 have gone through theeducation programme.
The programme for Type 1 diabetes comprises fourafternoon sessions of approximately 4 hours each,led either by a consultant physician or by a DSN,with input from a dietitian. Sessions are a mixture ofdidactic teaching and practical sessions (e.g. taking ameal together in the hospital canteen and estimatingcarbohydrate intake), and cover management ofdiabetes including exercise and nutrition, why goodcontrol is important, development of complicationsand injection techniques.
Health Technology Assessment 2003; Vol. 7: No. 22
7
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
The programme for Type 2 diabetes is known asthe Focus Education Programme and consists offour group education sessions lasting 1.5 hourseach. These sessions are run by a DSN, with inputfrom a consultant physician, a podiatrist and anutritionist. Friends and relatives of the patientare encouraged to attend. Topics covered include‘what is diabetes?’, monitoring, healthy eating andcomplications. The fourth session is an optionalin-depth workshop on food labelling, cookinghints and shopping tips.
A similarly structured education programme isavailable to all new referrals to the local diabetescentre in St Helens and Knowsley. The programmeconsists of 1 hour per week of either individualand group education for five consecutive weeks.These sessions are run by the diabetes specialistnursing team and dietitians. Topics coveredinclude ‘what is diabetes?’, control andcomplications, diet and exercise and medications.
A structured education programme for Type 1diabetes is currently available in a number ofhospitals across England as part of an evaluation.The Dose Adjustment for Normal Eating (DAFNE)group educational programme incorporates skillsbased training to teach flexible insulin adjustmentto match carbohydrate in a free diet on a meal bymeal basis. The programme is based on theDiabetes Treatment and Teaching Programmes(DTTP). Developed in Europe in the 1970s, theseare often referred to as the Geneva–Düsseldorfmodels of education and consist of intensivetraining for patients with Type 1 diabetes.
The programme consists of 5 days of intensivestructured training delivered to groups of 6–8patients. Topics covered include the estimation ofthe carbohydrate content of meals and participantsare taught skills of insulin dose adjustment. Thedefinitive aim of the programme is to achievepatient autonomy. The course is taught by two orthree educators (DSNs and dietitians) in eachcentre. DAFNE is currently undergoing a processof evaluation in England and more details can befound in Appendix 4 and also in Chapter 9.
Description of the interventionsconsidered in this reviewEducation for people with diabetes aims to improvetheir knowledge and skills, enabling them to takecontrol of their own condition and to integrate self-management into their daily lives. Self-managementalso occurs within the context of overall health
management. Education is a foundation forunderstanding how (and whether) to regulate one’sown diabetic medication and often cannot beevaluated outside of the context of treatmentmodifications. For all of these reasons, it issomewhat artificial to consider the effects ofeducation alone, as the aim of education is toenable patients to use the various therapies better.We have therefore adopted a pragmatic approach inassessing the efficacy of education for diabetes, andhave included packages of care wherein education isonly one component. The methodology for thereview is detailed in Chapter 3.
The educational interventions considered in thisreview are all aimed at educating adults with Type 1or Type 2 diabetes. A number of differences canbe observed between the included interventions,such as the duration of the intervention, and thespecific topics covered. However, all can bedescribed as structured educational interventionsfor diabetes self-management, and have met anumber of criteria assessing their reproducibility(see Chapter 3). This review has subdivided theinterventions into three groups: interventions forType 1 diabetes, interventions for Type 2 diabetesand interventions aimed at either Type 1 or Type 2 diabetes.
Interventions for Type 1 diabetesThese interventions all attempt to educate patientson a wide range of topics related to diabetes self-management, including diet, self-monitoring ofblood glucose (SMBG), the effects of insulin andexercise.
Interventions for Type 2 diabetesThese trials fell into two basic categories: those inwhich the aim of the intervention was to educatepatients on a range of topics related to diabetesself-management and those in which theintervention was focused on one or two aspects ofself-management alone (e.g. diet and/or exercise).
Interventions for patients with eitherType 1 or Type 2 diabetesThese trials also fall into two basic categories:those in which the aim of the intervention was toeducate patients on a range of topics related todiabetes self-management and those in which theintervention was focused on one or two aspects ofself-management alone (e.g. diet and/or exercise).
Owing to the differences in the interventionswithin each of these groups, more detaileddescriptions will be given with the assessment ofclinical effectiveness (see Chapters 4–6).
Background
8
Methods for reviewingeffectivenessThe methods for reviewing evidence of clinicaleffectiveness and the economic evaluation aredescribed in the research protocol (Appendix 1).Expert comments were obtained from the reviewadvisory group. Although many helpful commentswere received relating to the general content ofthe research protocol and the included outcomes,there were none that identified specific problemswith the methods of the review. Some expertsexpressed reservations about the focus oncontrolled trials for the evaluation of what is oftena complex intervention, but a review whichincluded all forms of evidence, for example fromobservational and qualitative studies, would nothave been possible within the time and resourceconstraints for this review and randomisedcontrolled trial (RCT) evidence is usually the mostreliable.
The methods outlined in the protocol aresummarised below.
Search strategySources of information, search terms and aflowchart outlining the identification of studies arepresented in Appendix 2.
Studies identified by the search strategy wereassessed for inclusion through three stages. Thetitles of all identified studies were screened by onereviewer and checked by a second reviewer.Abstracts were then screened by two independentreviewers and full-text versions of relevant paperswere retrieved. Inclusion criteria were applied byone reviewer and checked by a second reviewer,any differences being resolved through discussion.Owing to the number of eligibility criteria for thereview, an inclusion worksheet was utilised for thepurpose of applying the inclusion criteria, whichcan be found in Appendix 3. Data were extractedby one reviewer using a standard data extractionform and checked by a second reviewer. At eachstage, any differences in opinion were resolvedthrough discussion. Studies excluded from thereview of clinical effectiveness are listed inAppendix 4.
Inclusion and exclusion criteriaDesignRCTs and controlled clinical trials (CCTs) thatcompared a specific educational programme withusual care or with another educational programmewere included. Because diabetes care is constantlyevolving, CCTs were required to have a concurrentcontrol group. RCTs or CCTs that comparedmodels of group education with individualeducation were included.
InterventionThe review was limited to educational interventions,that is, the dissemination of knowledge and skillsbrought about using a number of approaches,which can be carried out with the normal range ofpersonnel available in diabetes care. Trials thatonly evaluated specific, specialised psychologicalinterventions aimed at changing an individual’sperceptions, such as cognitive/behavioural orpsychoanalytic therapy, or counselling wereexcluded. Educational interventions that include apsychological component were included. Studiesof education solely about specific complications(e.g. foot care) were not included.
ReportingIn order potentially to inform practice, includedstudies were required to have been reported withsufficient detail to be reproducible. They wererequired to have described the main componentsof the educational programme, such as:
� what the intervention is with some descriptionof the topics covered
� who provides instruction (e.g. post andqualification)
� how education is delivered (e.g. in person, bycomputer)
� group or individual� length of intervention (length and number of
sessions)� target audience (e.g. Type 1, Type 2 or both;
newly diagnosed)� didactic or interactive instruction� training for the educators.
Educational interventions that were not describedin sufficient detail to replicate were not included.
Health Technology Assessment 2003; Vol. 7: No. 22
9
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Chapter 3
Methods
ParticipantsParticipants should have been diagnosed withType 1 or Type 2 diabetes using the standarddiagnostic criteria in effect at the inception of thestudy. Both newly diagnosed and patients withestablished diabetes were included. In some casesthe types of diabetes were not clearly defined intrials, in which case these were treated as aseparate sub-group of trials. Participants shouldhave been described as ‘adults’ or a minimum of80% of participants should be 18 years of age orolder.
Quality assessmentThe quality of included trials was assessed usingcriteria recommended by the NHS Centre forReviews and Dissemination (CRD) (University ofYork) (Appendix 5).19 Economic evaluations wereassessed using a modified version of the criteriarecommended by Drummond and Jefferson.20
Quality criteria were applied by one reviewer andchecked by a second reviewer. Any disagreementswere resolved through discussion.
Some changes, additions or points of clarificationwere made to the methods discussed in theoriginal protocol and these are outlined below:
� As they did not assess patient education per se,interventions that were primarily evaluations ofpatient case management were not included.
� Studies that were available only as unpublishedmaster’s theses or doctoral dissertations werenot included.
Outcomes considered withinclinical effectiveness sectionsA range of outcomes has been assessed by theincluded trials. For ease of understanding theseoutcomes will be discussed within each subsectionof the clinical effectiveness sections, in threecategories: diabetic control, diabetic end-points,and QoL and cognitive measures.
Diabetic control outcomesThese outcomes are physiological measures thatare indicative of metabolic control, lifestylemodifications or cardiovascular risk. Theseoutcomes are important indicators of self-management success and serve as surrogateindicators of the risk of long-term complications.
Glycated haemoglobin (GHb) (e.g. HbA1c) is ameasure that reflects glucose levels in the bloodover a relatively long interval (2–3 months), and
therefore provides a much better guide to diabetescontrol than simple BG measurements.
BP and blood lipids (cholesterol and triglycerides)are risk factors for cardiovascular disease.
Body mass index (BMI) and weight are measuresof obesity, which is related to the development ofproblems in glycaemic control initially and isanother risk factor for the development ofcardiovascular disease.
In Type 2 diabetes, patients may be able to controltheir BG (at least early in the disease) bymodifying lifestyle factors such as diet andexercise. Therefore, an important treatment goaland indicator of intervention success may bereductions (or lack of increases) in the level of oralhypoglycaemic agents used by patients.
Diabetic end-pointsCertain variables are indicators of the progressionof diabetes into the associated complicationsdiscussed previously or general deterioration ofhealth or diabetic status.
Episodes of hypoglycaemia or ketoacidosis:patients may have too little glucose in the systemor too much. Both of these complications havebeen discussed previously.
Retinopathy and nephropathy are long-termcomplications associated with long-term poorregulation of BG. Neuropathy can be an acute orlong-term complication.
Rates of hospital admission are an indication ofthe general health of patients and whether BG isunder control.
QoL and cognitive measuresInterventions can affect how patients feel aboutthemselves, how they are functioning in societyand their perceived control of their health status.
Some of the studies assessed these variables withinstruments that were not validated. Results usingnon-validated instruments were not data extractedand will not be discussed. Although there may besome merit in such measures, without formalvalidation instruments may not be measuring whatthey claim to measure.
QoL has been measured with a number ofvalidated instruments. These instruments aredesigned to indicate changes in how patientsperceive their QoL. Some instruments are disease-
Methods
10
specific to assess QoL in relation to diabeteswhereas others are generic measures.
Measures considered under cognitive measuresinclude attitudes toward diabetes and diabetesknowledge. Increased knowledge of diabetes maycontribute as much or more to patients’ perceivedcontrol of diabetes as to metabolic control. Patientswho are more knowledgeable may feel better abouttheir diabetes and their ability to self-manage.
Validated measures of QoL, knowledge and othercognitive measures that were used in includedstudies are described in more detail in Appendix 6.
Quality considerationsAs for most interventions it is important toconsider the effects of diabetes education relativeto a control group. Ideally, to minimise bias,patients are randomly assigned to interventionand control groups (RCTs). In this review, CCTsare also considered as long as a control group isevaluated concurrently with the interventiongroup. Although many studies of diabetesinterventions have used designs that have notemployed a control group and have relied uponbefore and after measures, this is not a satisfactoryapproach. Other factors could be confounded withthe intervention such that after measures woulddiffer from before. These differences cannot beattributed to the intervention and cannot beevaluated in uncontrolled designs.
In addition, it is important that statisticalcomparisons are made between the interventionand control groups rather than consideringwithin-group changes from baseline. If within-group changes are reported they may reflect notonly the effect of an intervention, but also theeffect of being in a study or some other factor thatis co-varying with the intervention. For instance,changes from baseline in both intervention andcontrol groups suggest something of this sort isoccurring. In newly diagnosed patients withdiabetes, it might be expected that variousmeasures will change simply as patients adjust tothe diagnosis and attempt to make recommendedadjustments to lifestyle and/or medication. Patientswith Type 1 diabetes may have a ‘honeymoonperiod’ and may even be able to stop insulininjections for a time, after which controldeteriorates again. In designs in which bothintervention and control patients might beexpected to exhibit changes in variables, it isdesirable to use statistical methods that detectrelative changes (e.g. interactions betweentreatment condition and time). Similarly, thenatural evolution of Type 2 diabetes is for diabeticcontrol to worsen over time, and methods tocompare results appropriately betweenintervention and control groups are crucial. Forexample, maintaining diabetic control in anintervention group relative to deterioratingcontrol in a control group may be a valuableoutcome.
Health Technology Assessment 2003; Vol. 7: No. 22
11
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
BackgroundDiabetes treatment aims to maintain BG levels asclose as possible to non-diabetic levels and toreduce cardiovascular risk factors includingobesity, hypertension, smoking and high bloodlipid levels. In addition, patients should haveregular ophthalmological and podiatricexaminations and maintain appropriate foot care.Most studies of educational interventions havethese treatment goals in mind and have measuredone or more related variables. In addition, there isa growing awareness of the importance of patients’QoL and a few studies have measured QoL orother more specific indicators of attitudes orpsychological well-being.
Trials of self-managementinterventionsQuantity and quality of evidenceFour studies considering education for patientswith Type 1 diabetes met the inclusion criteria forthe review (see Table 2 and Appendix 7). Two ofthe included studies were RCTs,21,22 and two wereCCTs.23,24 Only one of the studies was truly a testof an educational intervention.22 The other threetested the effects of intensified insulin treatmentthat involved an educational component.Therefore, in three of the studies the effects ofeducation are confounded with the effects ofintensified insulin treatment.
The study sample size in the RCTs varied from 37participants between four study groups in theTerent trial,22 to 102 between two groups in theReichard trial.21 Sample sizes in the CCTs were181 for three groups in the Starostina trial24 and300 between three groups in the Mühlhausertrial.23 All trials except the Terent were carriedout in secondary care. Duration of diabetes acrossthe four included trials ranged from 523 to 18years,21 with the mean ages of participants beingapproximately 28 years in all studies. The lengthof follow-ups from inception of the trial were 12months,23 18 months,22 24 months24 and 10 yearsin the Stockholm Diabetes Intervention Study(SDIS).21
The quality of reporting and methodology of theincluded studies was generally poor by today’sstandards (Tables 3 and 4). The method ofrandomisation was unknown in both RCTs, and anattempt at concealment of allocation was made inone.21 The similarity of groups at baseline and theeligibility criteria were reported in all fourincluded trials. No trial reported analysis byintention-to-treat (ITT).
Description of the interventionAll of these studies involved a full self-management approach to education meaning thatthey attempted to educate on a wide range oftopics related to diabetes self-management.However, the degree of detail in describing theeducational interventions varied among reports.In some cases certain assumptions have beenmade about the nature of the interventions basedon reported outcomes or on vague descriptions.
In the one study that specifically assessed the effectof education alone,22 four groups were randomised.Two groups received a multifaceted educationprogramme consisting of six, 1-hour sessions within1 month. These were individual sessions thatcovered the relation between food and BG, insulinand urinary glucose excretion, hypoglycaemic and hyperglycaemic episodes, foot care, injectionsand urine testing. One of the educated groups andanother group not having received the educationwere also taught about SMBG in an additionalsession. The groups performing SMBG were“encouraged to change their insulin doses toachieve preprandial values below 7 mmol/l andpostprandial values below 10 mmol/l”. A finalgroup continued with usual care. The providers forthis study were a physician and a dietitian.
Three studies were designed to test the effects ofintensified treatment. These interventions reliedon education to help patients understand therelationship between eating and insulin. Thetheory behind these interventions (and the SMBGgroups in the Terent study) is that normalmetabolic regulation is a constant interplaybetween food consumption, energy requirementsand insulin production. Therefore, theseinterventions focused on educating patients about
Health Technology Assessment 2003; Vol. 7: No. 22
13
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Chapter 4
Effectiveness of interventions for Type 1 diabetes
Effectiveness of interventions for Type 1 diabetes
14
TABLE 2 Included studies of self-management education interventions for Type 1 diabetes
Reference Intervention Participants Duration of Timing of intervention evaluation
Reichard et al.,(multiplepublications),1988–9621
(SDIS)RCT
Two groups:1. Self-management education with
intensified treatment. Physicianprovided 2 sessions of educationto individuals or pairs of 2–3 h.Regular contact over studyperiod via telephone.
2. Usual care: instructed to useSMBG and visited clinic every 4months, many had frequentcontact over study period
102 patients 2 initial education sessions thenphone calls every 2 weeks initially,later as required
1.5 years3 years5 years7.5 years10 years
Terent et al.,198522
RCT
Four groups:
1. Self-management education +SMBG
2. Self-management education
3. SMBG
4. Usual care
Groups 1 and 2 provided byphysician and dietitian for 6 hourlylessons during 1 month. SMBGgroups had additional session. Thenseen every 3rd month
Group 4 seen in clinic every 3rdmonth
37 patients 1 month 18 months
Mühlhauser et al., 198723
(Geneva–Düssel-dorf model)CCT
Three groups:1. Self-management education with
intensified treatment. Groupeducation over 5 days, run byDSNs.
2. Self-management education withsimple rules for insulinadjustment but ‘conventionaltreatment’. Group educationover 4 days, run by DSNs.
3. Usual care. Under care ofphysician
300 patients 4–5 days 12 months
Starostina et al.,199424
(Geneva–Düssel-dorf model)CCT
Three groups:
1. Self-management education withintensified treatment + SMBG
2. Self-management education withintensified treatment + urinetesting
3. Usual care
Groups 1 and 2, 5-day groupeducation provided by 2 physicians
Group 3 no details
181 patients 5 days 12 months24 months
SDIS, Stockholm Diabetes Intervention Study; SMBG, self-monitoring of blood glucose, in which patients are taught how totake a blood sample and test the glucose level.
metabolic processes and how to regulate therelation between eating, exercise and insulin doses.Contrary to a set regimen for insulin doses, thegoal was to help patients learn how to self-treatwith generally more frequent insulin doses thatwere specifically related to variations in eating.This method of constant patient self-regulation of insulin doses is designed to mimic more closelythe natural regulation of insulin production inpeople who do not have diabetes. Patients weretaught to self-monitor glucose levels and to self-adjust insulin doses in relation to their energyconsumption and energy demands. In one study21
goals for BG were set individually with an overallgoal to reduce HbA1c to 7%. The two other studiesused the Geneva–Düsseldorf model for patienteducation and self-regulation. In one of thesestudies there was a comparison between self-monitoring using BG and using urine glucose andin these two studies the potential for liberalisingdiet was emphasised in relation to self-monitoringand insulin adaptation. The SDIS study alsoincluded education on microvascularcomplications.
The SDIS programme was provided by a physicianin two sessions of 2 and 3 hours. These patientswere seen in the clinic every 2 months and hadfrequent face-to-face and telephone contact withthe physician (continuous tutoring on demand).
The control group were advised to monitor theirBG and visited the clinic every fourth month. Thisintervention lasted for 7.5 years with an additionalfollow-up at 10 years. This study was essentially anindividual intervention, rather than a group one,although the initial education was reported tosometimes be given in pairs.
The two Geneva–Düsseldorf modelledprogrammes23,24 were based on a 5-day inpatientgroup training. The Mühlhauser study23 involvedone group based on the Geneva–Düsseldorf model[intensive diabetes treatment and teachingprogramme (IDTTP)] and another group [basicdiabetes treatment and training programme(BDTTP)] who were trained over 4 days and usedurine self-monitoring using locally availablematerials (Romania). The IDTTP group wereexplicitly trained in intensified insulin treatmentwhereas the BDTTP group were instructed onsimple rules for self-adjustment of insulin but weredescribed as having conventional insulin therapy.In the Starostina study,24 one group self-monitoredBG (BGSM) and another self-monitored usingurine glucose (UGSM). In one study23 theeducation was provided by nurses and in theother24 by physicians. The control group patientsreceived usual care by their physicians (no self-adjustment of insulin doses and usual strict dietrecommendations).
Health Technology Assessment 2003; Vol. 7: No. 22
15
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
TABLE 3 Quality assessment of RCTs of education for Type 1 diabetes
Study Randomisation Concealment Baseline Eligibility Blinding of Primary ITT Missing of allocation characteristics criteria assessors outcome analysis values
results
Reichard Partial Inadequate Reported Yes Adequate Partial Inadequate Adequateet al., 1988–9621
Terent Unknown Unknown Reported Yes Adequate Partial N/A N/Aet al., 198522
Not applicable.
TABLE 4 Quality assessment of CCTs of education for Type 1 diabetes
Study Baseline Eligibility Blinding of Primary ITT Missing Representativenesscharacteristics criteria assessors outcome analysis values
results
Mühlhauser et al., 198723 Reported Yes Unknown Partial Unknown Partial Yes
Starostina et al., 199424 Reported Yes Unknown Partial Unknown Adequate No
Assessment of effectivenessOutcomes reflecting diabetic controlTable 5 shows the results for GHb for the fourstudies in Type 1 diabetes. Results are shown withRCT findings preceding CCT findings. Withinthese groups the results from the largest trials areshown first succeeded by other trials in descendingorder. The size of the study at the start is shownand the number of patients included in theanalyses is indicated with the correspondingresults. These conventions will apply throughoutthe report.
The SDIS followed patients for 7.5 years duringthe study with a final post-study follow-up at 10years. The intervention group demonstratedconsistently lower HbA1c levels at all pointsranging from 1.6% lower to 1.1% lower, p values
< 0.01. It should be noted that there was attritionacross the evaluation points, but substantial losseswere not seen until the 10-year follow-up. At thislast assessment point it may be that a non-representative group of patients remained availablefor evaluation, that is, those most concerned abouttheir illness, or those more interested in education.The decreasing HbA1c levels in the control groupover time may also reflect that the least motivatedparticipants were dropping out of the trial. Itshould also be noted that this study involved moreclinic visits for the intervention group and allowedfor telephone consultation for the interventiongroup on demand for the 7.5 years of the study.Therefore, it may be that to achieve these long-lasting results requires some continuous level ofcontact. However, between the 7.5- and 10-yearevaluations the intervention participants returnedto routine care.
Effectiveness of interventions for Type 1 diabetes
16
TABLE 5 GHb (%) findings from studies of adults with Type 1 diabetesa
Reference n Time point Intervention(s) Control Differences (mean % ± SEM unless between stated otherwise) groups
Reichard, Initial total: Baseline 9.5 9.4et al., 102 18 months 7.5 (from graph) 9.0 (est.) p < 0.011988–9621 3 y = 97 3 y 7.4 (0.1) 9.0 (0.2) p < 0.01(SDIS) 5 y = 96 5 y 7.2 (0.1) 8.7 (0.1) p < 0.01RCT 7.5 y = 89 7.5 y 7.1 (0.7) 8.5 (0.7) p < 0.01
10 y = 43 10 y 7.2 (0.6) 8.3 (1.0) p < 0.01
Terent et al., Initial total: Baseline Education SMBG Education 198522 37 (10/8/9/10) 12 months + SMBG alone aloneRCT In analysis: 18 months 12.3 (SD 3.2) 11.8 (SD 1.4) 11.2 (SD 2.0) 11.2 (SD 2.3)
37 (10/8/9/10)11.0 (SD 2.6) 10.8 (SD 1.0) 9.9 (SD 2.5) 9.5 (SD 3.2) NS
10.2 (SD 1.9) 9.8 (SD 3.0) 10.2 (SD 2.1) 10.4 (SD 2.1) NS
Mühlhauser, Initial total: Baseline IDTTP BDTTPet al., 198723 300 (100/ 12 months 12.3 (0.2) 11.7 (0.2) 12.5 (0.2) IDTTP: CCT 100/100) Control
In analysis: 9.3 (from graph) 11.2 (from graph) 12.8 (from p < 0.01287 graph) IDTTP:(98/92/93) BDTTP
p <0.01
Starostina, Initial total: Baseline UGSM BGSMet al., 199424 181 12 months 12.5 (0.2) 12.6 (0.2) 12.2 (0.2)CCT (61/60/60) 24 months
In analysis: 9.4 (0.2) 9.3 (0.2) 12.3 (0.2) Not tested165 (55/52/58) 9.2 (0.2) 9.2 (0.2) No data Not tested
a Values may represent HbA1 or HbA1c (see individual data extraction in Appendix 7 for details).SEM, standard error of mean; IDTTP, intensive diabetes treatment and teaching programme, a 5-day training with intensifiedinsulin treatment; BDTTP, basic diabetes treatment and training programme, a 4-day training with simple rules for self-adjustment of insulin; UGSM, urine glucose self-monitoring; BGSM, BG self-monitoring; SD, standard deviation; NS, notstatistically significant.
The Terent study22 is the only one designed to testan effect of education specifically. There were nosignificant differences in HbA1 between groups inthis study, but it was a very small study. There istherefore no indication that this educationalintervention had any effect on HbA1. Theeducation provided in this study was relativelybrief with relatively long follow-ups (11 and 17months) without additional intervention.Interestingly, the two groups who were trained toself-monitor BG and were advised to self-regulatetheir insulin also showed no signs of metabolicimprovement over the control group. However,the SMBG training was brief, consisting of only asingle session.
In the Mühlhauser study,23 the group receiving the5-day training programme and explicitlyintensified treatment (IDTTP) had lower HbA1
levels than either the control group or the groupreceiving the 4-day programme (BDTTP) andconventional insulin treatment. In the Starostinastudy24 the intervention groups appear to havelower HbA1 levels than the control group;however, between-group comparisons were notconducted. Both of these studies were CCTs.
Based on the SDIS and Mühlhauser results, itappears that educationally based intensivetreatment interventions can have long-lastingbeneficial effects on HbA1.
BPOnly one trial reported BP as an outcome. TheSDIS reported lower systolic and diastolic BP inthe intervention group at both 3- and 5-yearfollow-ups, but the differences were not compared statistically. At 10 years systolic BP waslower in intervention patients (124.9) than incontrol patients (132.2), p < 0.05. The diastolicBP in intervention patients (74.1) was alsomarginally lower than in control patients (77.3), p= 0.085. However, it should be noted that therewas considerable attrition at the 10-year follow-upand that systolic BP was higher at baseline in thepatients remaining in the control group.
BMIReduction of body weight is often not a treatmentgoal for Type 1 diabetes, but excessive increase inbody weight may be due to overinsulinisation andfrequent hypoglycaemia. None of the threestudies23–25 reporting BMI demonstrated reducedBMI in their intervention groups. At the 12-monthevaluation, the Mühlhauser study reportedsignificantly higher BMI in their IDTTP group(23.3) than in the BDTTP (22.6) or control (22.4)
groups (p < 0.05), despite similar body compositionat baseline. A similar finding occurred with higherBMIs in the intervention groups than the controlgroup in the Starostina study, but between-groupcomparisons were not performed. Intensivetreatment may result in weight gain but these donot appear to be large effects.
Outcomes reflecting diabetic end-pointsIdeally, interventions should help to prevent thecomplications associated with diabetes. These maybe short-term as in hypoglycaemic episodes orlong-term as in retinopathy or neuropathy.
Hypoglycaemic episodesTable 6 shows the reported hypoglycaemic episodesduring the intervention period in the Type 1studies.
A concern when patients are self-regulating theirinsulin doses and often increasing the doses orfrequency of doses is that their BG may fall toolow, resulting in a hypoglycaemic episode. TheDCCT, an influential large trial of the effects ofintensive treatment, concluded that there was anincreased risk of hypoglycaemia with this methodof treatment.26
In the SDIS, the intervention group had aconsistently higher percentage of patients with atleast one hypoglycaemic episode. Thesedifferences were significant at all points except atthe 10-year follow-up. The high proportions ofpatients with hypoglycaemia may be misleading asthe figures reported at each follow-up arecumulative. It appears that most of the additionalhypoglycaemic episodes in the intervention groupoccur in the first 3 years, after which there is littleif any difference between the groups. Two of thestudies reported no significant differences inhypoglycaemic episodes between study groups,22,23
although the Mühlhauser study did report thattheir IDTTP intensive treatment group hadsignificantly more patients who had at least onehypoglycaemic episode than their control group.The IDTTP group also had more patients with ahistory of severe hypoglycaemia at baseline, butthis difference was not reported to be significant.Another study24 reported fewer hypoglycaemiacases in the intervention groups than in thecontrol group at 12 months, but did notstatistically test this difference.
Across the studies there is a suggestion thathypoglycaemic episodes may be more frequent inthe first few years of intensified treatment.
Health Technology Assessment 2003; Vol. 7: No. 22
17
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
KetoacidosisThe frequency of diabetic ketoacidosis (DKA)should be reduced by effective treatments and inparticular treatments that seek to more closelymatch insulin dose with metabolic requirements.Table 7 shows the reported ketoacidotic incidentsduring the intervention period in the Type 1studies.
Two studies tested for statistical differencesbetween groups in ketoacidotic incidents. TheTerent study reported no significant differencesbetween the education plus SMBG group and theeducation group, but was likely underpowered. TheMühlhauser study reported that the control grouphad more patients with DKA and more episodes ofDKA than either of the intervention groups.
There is a suggestion that ketoacidotic incidentsmay be less frequent in the intervention groups,although the evidence is limited.
Hospital admissionsOne desirable outcome from a diabetesintervention would be reduction in hospitalisation.This would be indicative of better health.
Two studies reported hospital admission rates, butthe Starostina study did not test for between-groupdifferences. The Mühlhauser study reported thatfewer patients were hospitalised in the interventiongroups (IDTTP, 42; BDTTP, 57) than the controlgroup (84), p < 0.01. There were also lower totalhospital admissions and days admitted in theintervention groups (IDTTP, 67 admissions/630
Effectiveness of interventions for Type 1 diabetes
18
TABLE 6 Episodes of hypoglycaemia from studies of adults with Type 1 diabetes
Study Outcome n Time point Intervention Control Differences between groups
Reichard Hypoglycaemic Initial total: Baseline NR NRet al., episodes 102 18 months 48% 22% p < 0.011988–9621 (% of 3 y = 97 3 y 57% 23% p < 0.01(SDIS) patients with 5 y = 96 5 y 77% 56% p < 0.05
at least one 7.5 y = 89 7.5 y 80% 58% p < 0.05RCT episode) 10 y = 43 10 y 86% 73% NS
Terent et al., Hypoglycaemic Initial total 37 Education SMBG Education 198522 episodes (10/8/9/10) + SMBG alone alone
In analysis: BaselineRCT 37 NR NR
(10/8/9/10) 12 months 7 in SMBG groups 14 in non-SMBG NSgroups
Mühlhauser Hypoglycaemic Initial total: IDTTP BDTTPet al., 198723 episodes 300 (100/
(total 100/100) Baseline NR NR NRCCT no. of patients In analysis:
with at least 287 12 months 12 5 6 IDTTP: control one episode) (98/92/97) p < 0.05
Mühlhauser Hypoglycaemic Initial total: IDTTP BDTTPet al., 198723 episodes 300 (100/
(total 100/100) Baseline NR NR NRCCT no. of In analysis:
episodes) 287 12 months 27 5 9 NS(98/92/97)
Starostina Hypo- Initial total: UGSM BGSMet al., 199424 glycaemia 181 (61/60/60)
(cases) In analysis: Baseline 2 6 6CCT 165 (55/52/58)
12 months 2 6 8 Not tested
24 months 8 4 No data Not tested
NR, not reported.
days; BDTTP, 100 admissions/967 days; control, 173admissions/1447 days), p values < 0.01. In addition,hospitalisation was lower in the IDTTP group (5-dayeducation) than in the BDTTP group (4-dayeducation), p values < 0.05. Care is needed in theinterpretation of data on hospitalisations, as littledetail is reported as to the cause for the hospitalstay. However, these results suggest that patientswho are intensively self-treating require less hospitaltreatment than control patients.
Long-term complicationsThe rates of other complications were reportedonly in the SDIS as this was the only study with asufficiently long follow-up. These complicationswere followed in detail and all reported outcomes
can be seen in the data extraction form for thisstudy in Appendix 7. Representative outcomes arereported here.
RetinopathyThe percentage of patients who demonstratedserious retinopathy was significantly lower in theintervention group at both 7.5 years (27%) and 10years (33%) than the control group (52 and 63%,respectively), p values < 0.01. Mean retinopathylevels (using a 12 grade scale, 0.5–6.0) are shownin Table 8.
Differences in mean retinopathy level between theintervention and control groups did not becomestatistically significant until after 5 years of follow-up.
Health Technology Assessment 2003; Vol. 7: No. 22
19
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
TABLE 7 Incidents of ketoacidosis from studies of adults with Type 1 diabetes
Study Outcome n Time point Intervention Control Differences between groups
Reichard DKA (no. Initial total: Baseline NR NRet al., of patients 1021988–9621 experiencing 7.5 y = 89 7.5 y 1 2 Not tested
1 episode) 10 y = 43RCT 10 y 1 4 Not tested
Terent et al., DKA Initial total 37 Education SMBG Education 198522 (10/8/9/10) + SMBG alone alone
In analysis:RCT 37 (10/8/9/10) Baseline NR NR
12 months 2 3 NS
Mühlhauser DKA (no. of Initial total: IDTTP BDTTPet al., 198723 patients with 300 (100/ Baseline NR NR NR
at least one 100/100)CCT episode) In analysis: 12 months 2 3 13 IDTTP:
287 control, (98/92/97) p < 0.01
BDTTP: control, p < 0.05
Mühlhauser DKA (total Initial total: IDTTP BDTTPet al., 198723 no. of 300 (100/ Baseline NR NR NR
episodes) 100/100)CCT In analysis: 12 months 2 4 16 IDTTP:
287 control, (98/92/97) p < 0.01
BDTTP: control, p <0.05
Starostina DKA (cases) Initial total: UGSM BGSMet al., 199424 181 Baseline 9 10 17
(61/60/60)CCT In analysis: 12 months 1 0 16 Not tested
165 (55/52/58) 24 months 0 0 Not tested
NephropathyNephropathy was assessed by 24-hour urinaryexcretion of albumin (UAER) and by glomerularfiltration rate (GFR). These results are shown inTable 9.
The UAER was significantly higher in the controlgroup than in the intervention group at 3, 5 and7.5 years. At the end of the trial (7.5 years), onlyone patient from the intervention group hadUAER levels >200 µg/min compared with ninepatients in the control group, p = 0.01. Althoughthe mean GFR did not significantly differ betweenthe groups, by 7.5 years six control patientsdeveloped a GFR below the normal range whereasnone of the intervention patients did, p = 0.02.
NeuropathyNeuropathy was primarily assessed by self-reportsfrom patients. However, nerve conduction velocitieswere also measured and these results can be found
on the data extraction form in Appendix 7. Thenumber of patients who exhibited neuropathy isshown in Table 10.
Variable results with regard to the presence ofneuropathy can be attributed to the differingnumber of patients remaining in the evaluation atdifferent time points. At the official end of thetrial (7.5 years) there were no significantdifferences in neuropathy between interventionand control groups. However, at the 10-yearfollow-up (2.5 years after the trial had ended),among those patients who were available forevaluation there were significantly more patientswith neuropathy among the control patients.
Outcomes reflecting QoL and cognitivemeasuresQoL was not assessed using validated measures inany of the included Type 1 studies. Knowledge wasassessed with validated instruments in two studies
Effectiveness of interventions for Type 1 diabetes
20
TABLE 8 Mean (SEM) retinopathy level in SDIS trial
Time Intervention Control Difference between groups
Baseline 2.4 (0.1) 2.6 (0.1)18 months 2.8 (0.2) 3.2 (0.2)3 years 3.2 (0.2) 3.6 (0.2) NS5 years 3.5 (0.2) 4.1 (0.2) p < 0.05
TABLE 9 Mean (SEM) UAER and GFR rates in SDIS trial
Parameter Time point Intervention Control Difference between groups
Mean UAER levels (µg/min) Baseline for 3 y 1.3 (0.1) 1.4 (0.1)3 ya 1.3 (0.1) 1.6 (0.1) p < 0.05Baseline for 5 y 55.7 (26.7) 74.3 (31.0)5 y 46.0 (26.1) 239.9 (129.7) p < 0.05Baseline for 7.5 y 56 (175) 63 (206)7.5 years 45 (110) 119 (219) p < 0.05
GFR (ml/min) Baseline 122 1263 years 115 (3) 119 (3) Not tested5 years 112 (3) 115 (4) Not tested7.5 years 109 (19) 110 (27) NS10 years 110 (18) 109 (25) NS
a Appears to use a different scale from that used at 5 and 7.5 years, although reported to be µg/min.
TABLE 10 Number (percentage) of patients who exhibited neuropathy in the SDIS trial
Time points Intervention Control Difference between groups
Baselines (5 y/7.5 y/10 y) 13/5 (12)/ 2 17/8 (17)/165 y 16 34 p < 0.017.5 y 6 (14%) 13 (28%) NS10 y 14% 32% p < 0.05
and the results are shown in Table 11. A fullerdescription of the measures used in these twostudies can be found in Appendix 6.
In the Mühlhauser study, knowledge scores werehigher in the two intervention groups than in thecontrol group and were higher in the IDTTP groupthan in the BDTTP group. Although knowledgewas apparently greater in the intervention groupsof the Starostina study, differences from thecontrol group were not statistically tested.
Increased knowledge is undoubtedly a desirableoutcome that should reflect greater ability to takepart in one’s own care and greater confidence inself-care. However, there is little evidence thatknowledge alone predicts better metabolicoutcomes or reduced complications (e.g. Glasgowand Osteen27).
Summary of results from studiesin Type 1 diabetesThree included studies tested interventions thatwere built on a foundation of education, but thatfundamentally were intensified treatmentprogrammes. These interventions focused onhelping patients learn the relation between eatingand insulin requirements. The goal was to helppatients to self-regulate their insulin intake andgenerally to take more doses of insulin during theday to more closely mimic the non-diabetic state.
The SDIS trial may be of most value in that it wasan RCT that continued for a sufficiently long
period to assess not only mediating controlvariables, but also long-term complications. Thistrial was based on the belief that educationprovides the means for patients to learn to self-regulate their insulin. The initial training in thistrial was less intensive than those based on theGeneva–Düsseldorf model, but high levels ofongoing face-to-face and telephone contact wereavailable to patients meaning that for a longperiod they were effectively receiving individualisededucation. The mean contact per patient in theintervention group was 45 minutes/monthcompared with an average of 10 minutes/month forthe control group. Between 3 and 5 years after thestart of the study the contact time no longerstatistically differed between groups. Therefore, itseems that this study involved on-going educationfor approximately 3–4 years. This level ofindividualised contact with patients is not likely tobe supportable in most usual care settings.
The SDIS study demonstrated that significantreductions in HbA1c, retinopathy, nephropathyand neuropathy could be achieved. Reductions inHbA1c were long-lasting. The differences incomplications generally were not evident untilseveral years into the study, demonstrating theimportance of long follow-ups for these kinds ofstudies. However, similar to the Mühlhauser study,the intervention group had more hypoglycaemicepisodes. It should also be noted that this studyreported results at each follow-up based on thepatients who were still in the trial. Attrition levelsfor the first 7.5 years were not particularly highand there was little evidence that patientsremaining differed from those who did not.
Health Technology Assessment 2003; Vol. 7: No. 22
21
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
TABLE 11 Knowledge of diabetes from studies of adults with Type 1 diabetes
Reference n Time point Intervention (mean Control Differences between groupsscore ± SEM)
Mühlhauser Initial total: IDTTP BDTTPet al., 198723 300
(100/100/100) Baseline 16 (1) 17 (1) 16 (1)CCT In analysis:
287 (98/92/93) 12 months 32 (1) 26 (1) 24 (1) IDTTP: control, p < 0.01
BDTTP: control, p < 0.01IDTTP: BDTTP, p < 0.05
Starostina Initial total: UGSM BGSMet al., 199424 181 Baseline 11 (0.1) 11 (0.1) 11 (1)
(61/60/60)CCT In analysis: 12 months 25 (1) 26 (1) 11 (1) Not tested
165 (55/52/58) 24 months 25 (1) 26 (1) No data
However, it is possible that selective attrition mayhave left healthier and/or more motivated patientsin the intervention group.
Two CCTs also tested an intensified treatmentapproach. Although one of these studies24 did notstatistically test for differences between interventionand control groups and is therefore of limitedvalue, the results from the SDIS and the otherCCT23 suggest that such education/intensifiedtreatment programmes can have significant andlong-lasting effects. In the Mühlhauser study, onegroup was educated using the Geneva–Düsseldorfmodel with a 5-day inpatient training andintensified treatment. Because they were educatedusing a well-documented programme and usedusual glucose monitoring materials, this is themost relevant group in comparison with patientswho were receiving usual care (no self-adjustmentof insulin or self-monitoring). Results 1 year afterthe training showed that the intervention grouphad GHb 2.5% lower than the control group, aclinically significant difference. They also hadsignificantly fewer episodes of DKA, fewerhospitalisations and shorter hospital stays. Theydid, however, have significantly more episodes ofsevere hypoglycaemia and their BMI was slightlybut significantly higher than the control group.
Only one of the four included studies in Type 1diabetes incorporated a design that allowed anexplicit test of the effects of a purely educationalintervention. This study22 did not report some ofthe statistical comparisons of an education onlygroup against other interventions within the trial.However, the results presented did not indicatethat the education only intervention was effective.The education in this case consisted of 6 hours ofcontact over 1 month and covered a range ofdiabetes-related topics.
Interventions aimed at self-regulation of insulin in Type 1 diabetes do appear to have significant
and long-lasting benefits. These benefits cannot be attributed solely to the education that is offered to the patients, but are more likely due to the associated intensification of insulintreatment. The education involved in treatmentintensification programmes is fundamental totheir success.
It is of interest that the theoretical motivationbehind the intervention with both education andSMBG training in the Terent study was apparentlythe same as that of the other treatmentintensification studies (i.e. to educate aboutmetabolic processes and the relation betweeneating, exercise and insulin doses). However, thecontact time in this study was considerably lessoverall. This suggests that there may be someminimum level of intensity or overall duration ofeducation that is important to allow patients theability (perhaps made up of knowledge,experience, confidence, etc.) to achieve self-regulation of insulin that will be beneficial tometabolic control.
Although one programme of intensified treatment(SDIS) has shown long-lasting effects, it would alsobe of interest to test whether similar effects can bedemonstrated in programmes that have initiallymore intensive training, but without thecontinuing individualised educational contacts ofthe SDIS. Unfortunately, no studies using thistraining method and maintaining a control groupfor a long follow-up (>2 years) were located. TheStarostina study suggests that improved GHb,DKA, hospitalisation rates and knowledge weremaintained for 2 years following education andinception of self-regulation of insulin; however,between-group statistical comparisons were notconducted.
ConclusionIntensified treatment combined with educationimproves diabetic control and outcomes.
Effectiveness of interventions for Type 1 diabetes
22
BackgroundGenerally, the treatment goals for Type 2 diabetesare the same as those for Type 1 diabetes asoutlined in Chapter 4. Studies of educationaleffects in Type 2 diabetes have therefore tended tofocus on evaluations of metabolic control, diabeticend-points such as late complications and QoL.
There are some circumstances in which some ofthe basic treatment goals are not sought. Forinstance, in older patients the goal ofnormoglycaemia may not be as prominent. A fewstudies mentioned that glycaemic control was nota primary goal of the intervention.
In addition to the outcomes discussed previouslyas being relevant to all studies in diabetes self-management, a few outcomes are specific to Type2 diabetes. The most important of these istreatment with OHAs. Unlike patients with Type 1diabetes, patients with Type 2 diabetes are notinsulin dependent, although many may eventuallybe treated with insulin. In most patients with Type2 diabetes a treatment goal is to minimise or avoidthe use of OHAs for as long as possible. It hasbeen suggested that this is important because theinsulin-producing beta cells may desensitise overtime, lessening the effects of the agents. Inaddition, the stimulation of these overburdenedcells may contribute to their exhaustion. Drugtreatment is also more costly and has more side-effects than management with lifestyle changes(e.g. diet and exercise) alone.
Lifestyle changes are therefore a morefundamental element of self-management in Type2 than in Type 1 diabetes. More emphasis isplaced on diet, weight loss and exercise than inType 1 diabetes.
Sixteen trials that included only participants withType 2 diabetes met the inclusion criteria. Thesetrials fell into two categories: those in which theintervention was a more or less complete self-management approach and those in which theintervention was focused on one or two aspects ofself-management (e.g. diet and/or exercise). Theclinical effectiveness of these two categories oftrials will be discussed separately followed by a
summary of findings from interventions directedat Type 2 diabetes generally.
The nature of interventions aimed at Type 2diabetes is variable. There are variations in thecharacteristics of patients recruited, the focus ofthe intervention, the intensity and duration of theintervention, the theoretical foundation (if any) forthe intervention, the providers, the setting and soon. There is little consistency among studies thatallows for summarising results.
Trials of self-managementinterventionsQuantity and quality of evidenceEight studies comparing self-managementeducation for patients with Type 2 diabetes metthe inclusion criteria for the review and can beseen in Table 12 and Appendix 8. Six of theseincluded studies that were RCTs28–32 (Cooper andcolleagues; see footnote to Table 12) and twoCCTs.33,34 In the RCTs, study size varied from 5132
to 256,28 in the two CCTs study sample sizes werearound 125. These two CCTs were evaluating thesame underlying programme. Only one includedstudy compared education in more than twogroups of patients.29 The remainder all comparedan intervention group with a usual care controlgroup. Three trials were carried out in primarycare28,33,34 two in secondary care,31,32 one in auniversity clinic30 and one across both primaryand secondary care (Cooper and colleagues; seefootnote to Table 12). One trial did not report thesetting for the study.29
In two studies the duration of diabetes was within1 year of diagnosis.29,31 Duration of diabetes in theremaining trials ranged from 5 (Cooper andcolleagues; see footnote to Table 12) to 10 years.30
Mean age of participants ranged from 55 to 65years across all studies. Except for the Trento30
trial (24 months), length of follow-up frominception was 12 months.
The quality of reporting and methodology of theincluded studies was generally poor by today’sstandards (Tables 13 and 14). The method ofrandomisation was unknown in all but one RCT30
Health Technology Assessment 2003; Vol. 7: No. 22
23
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Chapter 5
Effectiveness of interventions for Type 2 diabetes
Effectiveness of interventions for Type 2 diabetes
24
TABLE 12 Included studies of self-management education interventions for Type 2 diabetes
Reference Intervention Participants Duration of Timing of intervention evaluation
Brown et al.,200228
RCT
Two groups:1. Self-management education. Team provided
group education for 52 contact hours
2. Usual care by physicians
256 patients 9 months and 3 monthsof support groupsessions = 12 months
12 months
Campbell et al., 199629
RCT
Four groups:1. Minimal instruction. Team delivered with
2 contact hours
2. Individual education. Team delivered with 8 contact hours
3. Group education. Team delivered with ~4 daystotal contact time
4. Behavioural programme. One nurse provided atleast 6 contact hours
238 patients Differed between andwithin groups. Up to12 months
12 months
Trento et al., 200130
RCT
Two groups:1. Self-management education in groups by a team.
Up to 32 contact hours over 2 years
2. Usual care. Seen by physicians every 3 months
112 patients Varied among patients;up to 2 years
24 months
Cooper et al.,unpublisheda
RCT
Two groups:1. Self-management group education. DSNs
delivered with 16 h contact
2. Usual care. No details
89 patients 8 weeks 12 months
Heller et al.,198831
RCT
Two groups:1. Self-management group education (weight loss
focus). Delivered by dietitian and DSN with 7.5contact hours
2. Usual care with physician and also saw dietitianevery 3 months
87 patients 6 months 12 months
Raz et al.,198832
RCT
Two groups:1. Self-management group education. Team
delivered. Minimum of 12 contact hours
2. Usual care. Follow-up every 2 months
51 patients 12 months 12 months
Domenech et al., 199533
CCT (groups fromsimilar medicalpractices)
Two groups:1. Self-management education. Group education by
physicians. ~7 h contact time
2. Usual care. No details
124 patients 1 month 12 months
Kronsbein et al., 198834
CCT (by medicalpractices withcontrolpractices onwait list)
Two groups:1. Self-management education. Group education by
physicians assistants. ~7 contact hours
2. Usual care with GP. No details
127 patients 1 month 12 months
a Cooper H, Booth K, Gill G. A randomised controlled study of education for people with Type 2 diabetes: unpublished work,2002.
and concealment of allocation was not reported inany. The similarity of groups at baseline, and theeligibility criteria were reported in all includedtrials. Only one study reported an analysis by ITT.28
Description of interventionsAlthough each of the trials apparently developedtheir interventions independently and withoutreference to any single theoretical foundation, theinterventions were similar in educating patientsabout a wide range of components of self-management in diabetes. Unfortunately, thedescriptions of interventions are often fairlylimited and vague. This is despite an attempt toinclude only trials that provided some detail as tothe nature of the intervention. In some casesdetails of interventions are assumed on the basis ofoutcomes that are reported or vague descriptions.
Topics that were covered in the interventionarm(s) of all of these studies included: nutrition,diet or importance of weight and self-monitoring
(blood and/or urine). The majority of studies alsodiscussed exercise or physical activity28,29,32–34
(Cooper and colleagues; see footnote to Table 12)and complications and/or management ofcomplications.28,30,31,34 Five studies covered footcare specifically28,29,33,34 (Cooper and colleagues;see footnote to Table 12), three studies includedcoverage of basic causes and treatment29,32,34 andfour how to handle sick days28,33,34 (Cooper andcolleagues; see footnote to Table 12). Twostudies33,34 trained patients to reduce or stop oralagents in the case of hypoglycaemia (MühlhauserI: personal communication, 2002). Several othertopics were incorporated into only one study each.
In most studies several people were involved inproviding the training.28,31–32 These teams weregenerally made up of physicians, nurses anddietitians. In two studies the interventions wereadministered by nurses alone29 (Cooper andcolleagues; see footnote to Table 12). A physicianwas the provider for one study33 and physician’s
Health Technology Assessment 2003; Vol. 7: No. 22
25
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
TABLE 13 Quality assessment of RCTs of education for Type 2 diabetes
Study Randomisation Concealment Baseline Eligibility Blinding Primary ITT Missing of allocation characteristics criteria of outcome analysis values
assessors results
Brown et al., Unknown Unknown Reported Yes Unknown Adequate Adequate Partial200228
Campbell Unknown Unknown Reported Yes Unknown Partial Unknown Reportedet al., 199629
Trento et al., Adequate Unknown Reported Yes Inadequate Adequate Inadequate Adequate200130
Cooper et al., Unknown Unknown Reported Yes Unknown Adequate Inadequate Partialunpublisheda
Heller et al., Unknown Unknown Reported Yes Unknown Adequate Unknown Reported198831
Raz et al., Unknown Unknown Reported Yes Adequate Inadequate Unknown Reported198832
a See footnote to Table 12.
TABLE 14 Quality assessment of CCTs of education for Type 2 diabetes
Study Baseline Eligibility Blinding of Primary ITT analysis Missing values Representativenesscharacteristics criteria assessors outcome
results
Domenech Reported Yes Unknown Partial Unknown Adequate Yeset al., 199533
Kronsbein Reported Yes Unknown Adequate Unknown Partial Noet al., 198834
assistants (no details) provided the intervention inanother.34
There was considerable variation in the number ofhours of contact for each intervention. Theinterventions also varied in whether sessions wereprovided over a short interval or were spaced outover time. The study with the least contact timeinvolved four 1-hour sessions that apparentlyoccurred at 3-month intervals.30 The most briefinterventions lasted for 4 weeks.33,34 Other studieshad interventions that involved between 8 and 52hours of contact time over periods of 3 weeks upto 2 years. Some interventions began with 2–4more intensive sessions of 90–120 minutesfollowed up with additional sessions for instance at3 and 6 months.29,31,32 One study included fourinterventions that varied in duration and othercharacteristics with the shortest intervention being2 hours and the longest approximately 30 hoursof contact.29
In all but one study interventions were provided togroups of participants. In the Campbell study,29
three of the interventions involved individualinstruction whereas one intervention was a groupintervention (intervention 3).
Most of the studies did not mention that they werebased on any particular theory of healthpsychology or behaviour change. One study wasbased on patient empowerment (Cooper andcolleagues; see footnote to Table 12). One usedcognitive-behavioural strategies in a behaviourchange intervention,29 and one developed aculturally specific intervention aimed at MexicanAmericans based on four meta-analytic reviews ofprevious diabetes education interventions.28
All of these studies attempted to address multiplecomponents of diabetes self-management, butunlike similar interventions applied in patientswith Type 1 diabetes there were no specificmanipulations of medical treatment associatedwith the educational interventions. Individualpatients were followed by their physicians ortrialists and may have had their medical treatmentvaried as deemed necessary, but patients were notbeing trained to self-regulate their ownmedication, for instance. There were alsovariations in how many patients were receivingmedications.
Participants in control groups underwent usualcare, most often provided by their physicians orlocal clinics and received clinic appointments asnecessary. In two studies28 (Cooper and
colleagues; see footnote to Table 12), the controlgroups were on the waiting list for theintervention.
Additional characteristics of the studies will bediscussed below, as are the results. Attempts will bemade to identify characteristics of the studies thatmight account for differences in obtainingsignificant effects of interventions, although suchsuggestions are largely speculative.
Assessment of effectivenessA wide variety of outcomes were measured acrossthese studies. Only those that were reported inmultiple trials or that were judged to beparticularly meaningful will be summarised here. For each study all reported outcomes can be found in the data extraction forms inAppendix 8.
Outcomes reflecting diabetic controlTable 15 shows the results for GHb for theincluded studies of self-management education inType 2 diabetes.
Only three studies reported significant differencesbetween intervention and control groups inGHb.28,30,32 All three of these were RCTs. At the12-month evaluation the intervention group in theBrown study28 had HbA1c approximately 0.75%lower than the control group. In this study thebaseline HbA1c of participants in both groups washigh. The intervention group in the Trento study30
had HbA1c 0.8% lower than the control group at24 months. Interestingly, the intervention in theTrento study seems to have prevented thedeterioration of BG levels rather than improvingBG. The intervention group’s BG remainedapproximately the same whereas the control grouphad poorer BG at the end of the trial. Theintervention group in the Raz study32 had HbA1c
1.35% lower than the control group at 12 months.The other studies of this kind reported nostatistically significant differences betweenintervention and control groups on measures ofGHb, despite what would seem to be relativelylarge differences in mean levels of GHb betweenintervention and control groups in some of thestudies.
It should be noted that although the Campbellstudy did not report significant differences in GHbbetween the three intervention groups that wereevaluated, it would appear that these interventionsdid improve BG. These findings should, however,be interpreted with caution because they cannot becompared with a control group who might also
Effectiveness of interventions for Type 2 diabetes
26
have shown improvement and because there is anextremely high attrition rate in this study. It maybe that improvements may be attributable to themost motivated patients remaining in the study.
All of the studies that demonstrated significantresults were interventions delivered by a team ofdifferent professionals, which might suggest abroader range of presented information and
provider expertise, but two studies using such teams did not produce significant differences inGHb.
With one exception (Campbell and colleagues29),all of the studies that did not report significantdifferences had longer intervals from the end ofthe intervention itself to the follow-up (rangingfrom 6 months to 48 weeks) than did those that
Health Technology Assessment 2003; Vol. 7: No. 22
27
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
TABLE 15 GHb (%) findings from studies of self-management education in adults with Type 2 diabetesa
Study n Time point Intervention Control Difference (mean ± SD unless stated otherwise) between
groups
Brown et al., Initial total: Baseline 11.81 (3.0) 11.8 (.02)200228 256 (128/128)
In analysis: 12 months 10.89 (2.56) 11.64 (.85) p < 0.05RCT 224 (112/112) adjusted 10.87 adjusted
11.66
Campbell Initial total: 238 Reported values Minimal Individual Group Behaviouralet al., 199629 (59/57/66/56) are changes education education education
In analyses: from baseline No –3.3 –3.0 –4.8 NSRCT 83 follow-up (SEM 0.9) (SEM 1.1) (SEM 0.7)
(0/25/19/39)
Trento Initial total: Baseline 7.4 (1.4) 7.4 (1.4)et al., 112 (56/56)200130 In analysis: 24 months 7.5 (1.4) 8.3 (1.8) p < 0.01
90 (43/47)RCT
Cooper Initial total: 89 Baseline 7.9 (range 4.5–11) 7.0 (range et al., In analysis 4.6–10.6)unpublishedb 78 (47/31)
12 months 7.9 (2.1) 7.2 (1.6) NSRCT
Heller et al., Initial total: Baseline 12.3 (95% CI: 11.4 to 13.2) 12.7198831 87 (40/47) (11.9–13.5)
In analysis:RCT 75 (36/39) 12 months 9.0 (95% CI: 8.2 to 9.8) 9.9 (8.9–10.9) NS
Raz et al., Initial total: Baseline 10.0 (2.7) 9.6 (2.6)198832 51 (25/26) Time ×
In analysis: 12 months 8.25 (estimated 9.6 (from group RCT 49 (23/26) from graph) graph) interaction:
p < 0.05
Kronsbein Initial total: Baseline 7.1(1.6) 6.5 (1.6)et al., 198834 127 (65/62)
In analysis: 12 months 7.1 (1.6) 6.7 (1.5) NSCCT 99 (50/49)
Domenech Initial total: Values are –0.2% (0.4) +0.8% (0.4) NSet al., 199533 124 (53/71) changes from
In analysis: baselineCCT 79 (40/39)
a Values may represent HbA1 or HbA1c (see individual data extraction in Appendix 8 for details).b See footnote to Table 12.
reported significant differences betweenintervention and control groups. Of thosereporting significant differences, the Brownstudy28 involved the most contact time overall andinvolved contact at least monthly, the Raz study32
had three education sessions every 4 months andthe Trento study30 involved four education sessionsapparently every 3 months. In other words, amongthese three studies the longest follow-up withoutany educational contact was 3 months. Theseresults suggest the possibility that potential effectsof educational programmes are either not long-lasting or that the programmes must be deliveredsuch that they are distributed over long intervals.These points are, of course, speculation unless anduntil they can be tested in experiments in whichthese interpretations are explicitly tested.
It should be noted that the Brown study,28 whichreported significant effects on HbA1c, did involvethe most contact time and it was culturally specificfor its target audience of Mexican Americans.
BPBP was reported in two studies.29,30 The results areshown in Table 16.
The intervention in the Campbell study29 thatinvolved a behavioural intervention resulted ingreater decreases in diastolic BP than in standardgroup or individual self-managementinterventions. Whether this is a meaningfuldifference or whether this effect would be
maintained in the long term is unclear. In theTrento study, more patients in the interventiongroup were no longer considered hypertensive atthe end of the study than in the control group. Asthe difference was not statistically significant, littleshould be made of this finding. However, theremay be a lack of power to detect a difference.
BMI or weightOutcomes relating to weight or BMI were reportedin all included trials and can be seen in Table 17.
Four studies31–34 reported significant differences inBMI or weight (or changes in BMI or weight)between intervention and control groups and onestudy30 reported a marginal difference in BMI. Inall four studies weight loss was greater in theintervention group than the control group.However, in the Trento study the intervention grouphad a higher BMI than the control group at bothbaseline and the 24-month evaluation. Most of theweight losses were not of great magnitude with theexception of those in the Heller study. This study,although educating on multiple aspects of self-management, was primarily directed at weight loss.This programme, starting with individualised weighttargets, did produce significant weight loss in theintervention group (5.5 kg); however the controlgroup in the study also lost an average of 3 kg.
Cholesterol and triglyceridesFour studies reported other physiologicaloutcomes,28–30,32 shown in Table 18.
Effectiveness of interventions for Type 2 diabetes
28
TABLE 16 BP findings in studies of self-management education in adults with Type 2 diabetes
Study n Time point Intervention Control Difference (mean ± SEM unless stated otherwise) between
groups
Campbell Initial total: 238 Values are Minimal Individual Group Behaviouralet al., 199629 (59/57/66/56) changes in education education education
In analysis: 64 systolic BP RCT (0/16/11/37) from baseline No –6.8 –12.4 –16.9 NS
follow-up (5.8) (6.8) (3.8)
Campbell Initial total: 238 Values are Minimal Individual Group Behaviouralet al., 199629 (59/57/66/56) changes in education education education
In analysis: 64 diastolic BP RCT (0/16/11/37) from baseline No –5.3 –5.0 –7.9 p < 0.05
follow-up (3.0) (4.0) (2.6) for individual education or group education vs behavioural
Trento et al., Initial total: No. 200130 112 (56/56) hypertensive
In analysis: Baseline 34 25RCT 90 (43/47)
24 months 26 22 NS
Only one trial reported any significant difference incholesterol or triglycerides between interventionand control groups. Trento and colleagues30
reported in the text that high-density lipoprotein(HDL) cholesterol was lower in intervention patientsat 24 months, but this is inconsistent with valuesreported in the results table in which an increase inHDL cholesterol is reported for interventionpatients between baseline and follow-up whereas it
remained the same in control participants. Thesame study reported that triglycerides weremarginally lower in the intervention patients thanin control patients. Values reported in the resultstable suggest that triglycerides were reduced in theintervention group whereas they remained the samein the control group. However, triglycerides werehigher at baseline and at follow-up for theintervention group than for the control group.
Health Technology Assessment 2003; Vol. 7: No. 22
29
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
TABLE 17 BMI or weight findings from studies of self-management education in adults with Type 2 diabetes
Study Outcome n Time Intervention Control Difference point (mean ± SD unless stated ) between
groups
Brown et al., BMI Initial total: Baseline 32.33 (5.97) 32.12200228 256 (128/128) (6.35)
In analysis: RCT 227 (113/114) 12 months 32.17 (6.45) 32.28 NS
(6.52)
Campbell BMI Initial total: 238 Values are Minimal Individual Group Behaviouralet al., (59/57/66/56) changes in education education education199629 In analysis: 96 BMI from No –2.0 –1.4 –2.6 NS
(0/30/25/41) baseline follow-up (SEM 0.4) (SEM 0.5) (SEM 0.5)RCT
Trento BMI Initial total: Baseline 29.7 (4.5) 27.8 (4.1)et al., 112 (56/56)200130 In analysis: 24 months 29.0 (4.4) 27.6 (4.2) p = 0.06
90 (43/47)RCT
Cooper BMI Initial total: 89 Baseline 32.5 (6.7) 32.1 (6.1)et al., In analysisunpublisheda 78 (47/31) 12 months 31.3 (5.7) 30.5 (3.9) NS
RCT
Heller et al., Weight Initial total: Values are (Mean and 95% CI) –3 (2–4) p < 0.05198831 (kg) 87 (40/47) change in –5.5 (4 to 6.5)
In analysis: weight from RCT 75 (36/39) baseline
Raz et al., Weight Initial total: Baseline 75.4 (11.7) 73.4 198832 (kg,12 51 (25/26) (11.5) Time/
months, In analysis: group RCT data from 49 (23/26) 12 months 73 73 interaction:
figure) p < 0.05
Kronsbein Weight Initial total: Baseline 76.5 (12.6) 75.1 Diff. in et al., 198834 (kg) 127 (65/62) (12.9) change,
In analysis: p < 0.01CCT 99 (50/49) 12 months 73.8 (12.6) 74.8
(13.2)
Domenech Weight Initial total: Values are –2.4 (0.5) –0.4 (0.5) p < 0.01et al., 199533 (kg) 124 (53/71) change in
In analysis: weight from CCT 79 (40/39) baseline
a See footnote to Table 12.
Effectiveness of interventions for Type 2 diabetes
30
TABLE 18 Cholesterol and triglyceride findings from studies of self-management education in adults with Type 2 diabetesa
Study Outcome n Time Intervention Control Difference point (mean ± SD unless stated ) between
groups
Brown Cholesterol Initial total: Baseline 21.7 (2.5) 11.3 et al., (mmol/l) 256(128/128) (2.7)200228 In analysis: 12 monthsRCT 225 (112/113) 10.5 (2.0) 10.4 NS
(2.4)
Campbell Cholesterol Initial total: 238 Values are Minimal Individual Group Behaviouralet al., (mmol/l) (59/57/66/56) changes education education education199629 In analysis: 76 from No 0.12 0.16 –0.33RCT (0/23/19/34) baseline follow-up (SEM (SEM (SEM NS
0.20) 0.16) 0.15)
Campbell HDL Initial total: 238 Values are Minimal Individual Group Behaviouralet al., cholesterol (59/57/66/56) changes education education education199629 (mmol/l) In analysis: 64 from No 0.02 0.18 0.06RCT (0/21/16/27) baseline follow-up (SEM (SEM (SEM NS
0.04) 0.10) 0.08)
Campbell Cholesterol Initial total: 238 Values are Minimal Individual Group Behaviouralet al., risk ratio (59/57/66/56) changes education education education199629 (total/HDL) In analysis: 61 from No –0.25 –0.35 –0.59
(0/21/15/25) baseline follow-up (SEM (SEM (SEM NS0.03) 0.46) 0.20)
Trento Total Initial total: Baseline 5.8 (1.1) 5.5 (0.9)et al., cholesterol 112 (56/56)200130 (mmol/l) In analysis: 24 months 5.7 (1.2) 5.6 (1.2) NS
90 (43/47)
Trento HDL Initial total: Baseline 1.2 (0.3) 1.3 (0.3)et al., cholesterol 112 (56/56)200130 (mmol/l) In analysis: 24 months 1.4 (0.4) 1.3 (0.3) p < 0.05RCT 90 (43/47)
Raz et al., Mean blood Initial total: Baseline 12.5 (2.4) 12.2 (3.1)198832 cholesterol 51 (25/26)RCT (mmol/l) In analysis: 12 months 11.8 (2.1) 12.5 (3.4) NS
49 (23/26)
Raz et al., HDL Initial total: Baseline 2.6 (0.2) 2.5 (0.2)198832 cholesterol 51 (25/26)RCT (mmol/l) In analysis: 12 months 2.7 (0.2) 2.5 (0.2) NS
49 (23/26)
Brown Triglyceride Initial total: Baseline 11.9 (7.2) 10.8 (6.6)et al., (mmol/l) 256(128/128)200228 In analysis: 12 months 11.9 (10.8) 11.0 (8.2) NSRCT 226 (113/113)
Trento Triglyceride Initial total: Baseline 2.6 (0.7–11.5) 1.7 et al., (mmol/l) 112 (56/56) (0.5–5.2)200130 In analysis:RCT 90 (43/47) 24 months 2.1 (0.7–6.9) 1.7
(0.6–3.9) p = 0.053
continued
Oral hypoglycaemic treatmentStopping oral agent therapy was an explicitobjective of the programme in two studies.33,34
Both reported significant differences in the use ofmedications between intervention and controlgroups. In the Kronsbein study34 no patients inthe intervention group were on insulin at follow-up whereas 10 of 49 patients in the control groupwere. In the same study the proportion of patientsnot using glucose-lowering medications in theintervention group rose from 32 to 62% betweenbaseline and evaluation whereas it remained at39% in the control group. In the Domenechstudy33 intervention patients had reduced theiraverage daily intake of OHAs (–1.4 ± 0.2 tablets)while the control group had increased intake(+0.9 ± 0.2 tablets).
Interestingly, these studies were both CCTs ratherthan RCTs. In the Kronsbein study theintervention patients came from practices in whichtheir physician chose to participate immediately inthe programme. Although the physicians of bothintervention and control patients had attended atraining session, it is possible that those physicianswho chose to start the programme immediatelywere more motivated to change the treatment oftheir patients. In the Domenech study theintervention and control patients were treated bythe same physicians, however, there was noblinding as to which patients were in which group.Surprisingly, these two interventions were also themost brief, consisting of only 6–8 hours ofeducation over 4 weeks.
Outcomes reflecting diabetic end-pointsVery few of these studies included complications asoutcomes, usually because the follow-up in thesestudies was too short. However, those that werereported are shown in Table 19.
There were no differences between interventionand control groups for any of these outcomes.
Outcomes reflecting QoL and cognitivemeasuresIt is possible that interventions may affect the QoLof patients either in conjunction with or instead ofeffects on physiological or behavioural measures.However, few studies included measures of QoL orknowledge using validated instruments. ReportedQoL and knowledge effects using validatedinstruments are shown in Table 20 and details ofthese measures are given in Appendix 6.
Only one study30 reported on QoL using avalidated scale. This scale used questions that wereto be answered on a Likert scale such that loweroverall scores reflect higher satisfaction. This studyreported results from 2 years of follow-up frominception; however, educational sessions wereconducted every 3 months throughout the 2-yearperiod. This intervention did apparently improvepatients’ QoL whereas QoL appeared todeteriorate in the control group.
Two of three studies30,34 reporting results forknowledge measures demonstrated thatintervention patients had higher knowledge ofdiabetes than the control patients. This isdesirable as patients who are more knowledgeableare better able to communicate with theirphysicians and likely to feel in better control oftheir own health. However, it is unclear whetherknowledge of diabetes alone has any effect onmetabolic control (e.g. Glasgow and Osteen27).
Only one trial reported any additional validatedQoL measures (Cooper and colleagues; see footnoteto Table 12). That study reported significantly betterattitudes to diabetes and its treatment in theintervention group at 12 months [baseline 72.8 (SD13.2), 12 months 75.1 (SD 11.0)] than the controlgroup [baseline 76.7 (SD 14.2), 12 months 70.5 (SD11.0)], p < 0.01. This test measures the integrationof diabetes and its treatment into the lifestyle andpersonality of the patient. Higher scores indicatebetter psychological adjustment to diabetes.
Health Technology Assessment 2003; Vol. 7: No. 22
31
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
TABLE 18 Cholesterol and triglyceride findings from studies of self-management education in adults with Type 2 diabetes (cont’d)
Study Outcome n Time Intervention Control Difference point (mean ± SD unless stated ) between
groups
Raz et al., Blood Initial total: Baseline 12.8 (1.8) 11.7 (1.9)198832 triglycerides 51 (25/26)RCT (mmol/l) In analysis: 12 months 11.8 (1.3) 11.3 (1.7) NS
49 (23/26)
a Cholesterol levels are presented here in mmol/l with the conversion rate of 1 mg/dl = 0.0555 mmol/l having been used.
Effectiveness of interventions for Type 2 diabetes
32
TABLE 19 Diabetic end-points from studies of self-management education in adults with Type 2 diabetes
Study Outcome n Time Intervention Control Difference point between
groups
Trento Diabetic Initial total: Baseline 42/8/6 38/13/5et al., retinopathy 112 (56/56)200130 (none/mild/ In analysis: 24 months 35/5/3 33/7/7 NS
more severe) 90 (43/47)RCT
Trento, Foot ulcers Initial total: Baseline 54/0/2 53/2/1et al., (never/past/ 112 (56/56)200130 active) In analysis: 24 months 42/1/0 45/1/1 NS
90 (43/47)RCT
Campbell Proportion Initial total: 238 Baseline Minimal Individual Group Behaviouralet al., consulting (59/57/66/56) education education education199629 ophthal- In analysis: 122 12 months NR NR NR NR
mology (%) (0/38/37/47) No 97 95 89 NSRCT follow-up
Campbell Proportion Initial total: 238 Baseline Minimal Individual Group Behaviouralet al., consulting (59/57/66/56) education education education199629 podiatry (%) In analysis: 103 12 months NR NR NR NR
(0/31/30/42) No 55 73 74 NSRCT follow-up
TABLE 20 QoL and knowledge from studies of self-management education in adults with Type 2 diabetes
Study Outcome n Time Intervention Control Difference point (mean ± SD unless stated between
otherwise) groups
Trento DQOL Initial total: Baseline 67.6 (19) 66.7 (25)et al., 112 (56/56)200130 In analysis: 24 months 55.6 (15.9) 80.8 p < 0.01
90 (43/47) (31.5)RCT
Campbell Knowledge Initial total: 238 Values are Minimal Individual Group Behaviouralet al., (59/57/66/56) changes education education education199629 In analyses: 90 from No 4.4 4.2 5.6 NS
(0/29/26/35) baseline follow-up (SEM 0.6) (SEM 0.5) (SEM 0.6)RCT
Trento Knowledge Initial total: Baseline 14.9 (7.9) 20.2 (7.4)et al., 112 (56/56)200130 In analysis: 24 months 24 (6.6) 17.4 (8.6) p < 0.01
90 (43/47)RCT
Kronsbein Knowledge Initial total: Baseline 9 (3) 9 (3)et al., 127 (65/62)198834 In analysis: 12 months 13 (4) 10 (4) p < 0.01
99 (50/49)CCT
DQOL, diabetes quality of life measure.
The QoL and knowledge results suggest that some of these programmes may affect thepsychological well-being of patients with diabetes,although these effects are by no means universal.
Interim summaryOf the studies designed to instruct patients aboutmultiple components of self-management for Type2 diabetes, the majority compared a singleintervention with a usual care control group over12 months. One study followed up patients for 24 months and another made a comparison offour different educational interventions over 12 months. In general, findings demonstratedlimited impact on outcomes.
Some effect of education on diabetic control, asmeasured by HbA1c, was demonstrated; however,these appear to be mostly attributable to longerterm interventions with a shorter duration fromthe intervention’s conclusion to the evaluation.There was little effect on weight loss. Two studiesreported reduced usage of OHAs in theintervention groups.
Very few studies reported outcomes relating todiabetic end-points. No significant effects weredemonstrated.
Patients’ QoL was assessed with a validatedmeasure in only one trial. QoL was better in theintervention group than the control group.Knowledge was found to be higher amongparticipants in the intervention groups in twostudies.
Trials of focused self-managementinterventionsRather than educating patients on all aspects ofdiabetes self-care as in the studies just discussed,the following studies have attempted to addressspecific, limited topics in diabetes self-management.
Quantity and quality of evidenceEight studies (seven RCTs, one CCT) comparingmore focused self-management education forpatients with Type 2 diabetes met the inclusioncriteria for the review and can be seen in Table 21and Appendix 8. These interventions focused ondiet and exercise (four studies35–37,42), diet,38
exercise,40 weight versus self-regulation41 or weightversus SMBG.39 Study sample sizes were generallysmall, varying from 2041 to 104.42 Three of the
included studies compared education in morethan two groups of patients.35,38,42 All trials thatreported the study setting carried out the trial inprimary care. Two trials did not report thesetting.35,41 Duration of diabetes was not widelyreported. In the four trials that report durationthis ranged from newly diagnosed36 to 13 years.37
The majority of trials followed up theirparticipants for 12 months from inception, thefollow-up was 18 and 24 months in the Kaplanand Uusitupa trials, respectively.
The quality of reporting and methodology of theincluded studies was poor by today’s standards(Tables 22 and 23). No details of an adequatemethod of randomisation or concealment ofallocation was reported in any of the includedtrials. The similarity of groups at baseline, and theeligibility criteria were reported in all sevenincluded RCTs. No trial reported analysis byintention to treat.
Description of interventionsThese interventions, owing to their focusednature, are more self-explanatory than those thatincluded a range of diabetes-related topics.However, as in the previous group ofinterventions, it is often difficult to describe theexact nature of the interventions as publishedreports are vague or incomplete. Someassumptions as to the interventions have beenmade based on outcomes or vague descriptions.
Interventions for diet and exerciseFour studies focused on diet and exercise.35–37,42
Detailed dietary education was provided in each ofthese studies and two of the four36,37 usedindividualised dietary programmes. Another35
used the ADA exchange diet. Little detail of thenature of the dietary education was reported inthe fourth study.42
Exercise programmes were individualised in two ofthe studies35,37 and in one other36 exercise wasrecommended at a particular intensity andfrequency for all. Little detail of the nature of theexercise programme was reported in the fourthstudy.42 Three of these interventions usedbehaviour modification principles to greater orlesser extents. One study35 required a monetarydeposit that was returned with the meeting ofgoals and meeting attendance. One usedcontracts37 and the other36 used food records.
These studies all involved at least some group work.
Health Technology Assessment 2003; Vol. 7: No. 22
33
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Effectiveness of interventions for Type 2 diabetes
34
TABLE 21 Included studies of focused self-management education for Type 2 diabetes
Reference Intervention Participants Duration of Timing of intervention evaluation
Kaplan et al.,198735
RCT
Four groups:1. Group diet education. Dietitian delivered. 20 contact hours
2. Group exercise education. Contact hours not given
3. Group diet and exercise education over 5 weeks, no detailsof contact time
4. Control education in group with team – each gave a lecture.~14 contact hours
87 patients 10 weeks 18 months
Uusitupa et al.,1992–636
RCT
Two groups:1. Diet and exercise education. Provided by a team. Contact =
6 clinic visits (duration not given)
2. Usual care control. Local health centre visits every 2–3months + outpatient clinics
(both groups given basic diabetes education)
86 patients 12 months 24 months
Ridgeway et al., 199937
RCT
Two groups:1. Group diet and exercise education. Nurse and dietitian
delivered. 9 contact hours
2. Usual care control. No details
56 patients 6 months 12 months
Wing et al.,198538
RCT
Three groups:1. Diet – behaviour modification
2. Nutrition education
3. Usual care (with nutrition education)
Groups 1 and 2 = group education provided by psychologistand nutritionist. Contact = 16 weekly sessions
Group 3 = content identical with (2) but only 4 monthlymeetings
53 patients 16 weeks 16 months
Wing et al.,198639
RCT
Two groups:1. Diet – weight control. Contact time not given
2. Diet – SMBG. Contact time ~20 meetings
50 patients 12 months 62 weeks
Samaras et al., 199740
RCT
Two groups:1. Exercise education. Group sessions provided by a team.
Contact time ~6 h
2. Usual care. Routine clinic visits + 3 assessment visits (nodetails of duration)
26 patients 6 months 12 months
Wing, et al.,198841
RCT
Two groups:1. SMBG with education on meaning of SMBG (self-regulation),
13 sessions
2. SMBG (self-monitoring). Contact time not given
20 patients 10 months 68 weeks
Gilliland et al.,200242
CCT
Three groups:1. Friends and family (FF). Group culturally appropriate diet
and exercise education with support 5 sessions, one every 6weeks, for ~2 h
2. One-on-one (OO). Individual culturally appropriate diet andexercise education. 5 sessions, once every 6 weeks for ~45 minutes.
3. Usual care control (some education but not culturallyappropriate and no details given)
104 MexicanAmericanpatients
10 months 12 months
Providers of the interventions varied but generallyinvolved teams of specialists such as dietitians,nutritionists, DSNs and physicians. In theGilliland study a trained community mentorprovided the intervention.
The duration and intensity of the interventionsvaried. Two interventions involved approximately9 hours of contact.36,37 One of these involved sixmonthly sessions and the other six sessionsbimonthly, and another35 involved 20 hours ofcontact in ten 2-hour meetings over 10 weeks. Thegroup intervention in the Gilliland study42
involved approximately 12 contact hours over 10months and the individual interventionapproximately 4 hours over the same period.
In studies with a control group, participantsunderwent usual care, most often provided bytheir physicians or local clinics and received clinicappointments as necessary.
Other focused interventionsFour other studies involved focused interventionsthat were each unique.
One study (Samaras and colleagues, 199740) usedan exercise intervention. This intervention wastheoretically motivated using the ‘proceed–precede’health promotion model, which is built on thenotion that health and health risks are determinedby multiple factors.43 The intervention involvedgroup sessions focusing on barriers to exercise,
Health Technology Assessment 2003; Vol. 7: No. 22
35
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
TABLE 22 Quality assessment of RCTs of focused education for Type 2 diabetes
Study Randomisation Concealment Baseline Eligibility Blinding of Primary ITT Missing of allocation characteristics criteria assessors outcome analysis values
results
Kaplan Unknown Unknown Reported Yes Unknown Inadequate Unknown Reportedet al., 198735
Uusitipa Unknown Unknown Reported Yes Unknown Adequate Inadequate Unknownet al., 199336
Ridgeway Unknown Unknown Reported Yes Unknown Inadequate Inadequate Adequateet al., 199937
Wing Unknown Unknown Reported Yes Unknown Partial Unknown Partialet al., 198538
Wing Unknown Unknown Reported Yes Adequate Partial Unknown Reportedet al., 198639
Samaras Unknown Unknown Reported Yes Unknown Partial N/A N/Aet al., 199740
Wing Unknown Unknown Reported Yes N/A Partial Inadequate Partialet al., 198841
TABLE 23 Quality assessment of CCT of focused education for Type 2 diabetes
Study Baseline Eligibility Blinding of Primary ITT Missing Representativenesscharacteristics criteria assessors outcome analysis values
results
Gilliland Reported Yes Unknown Adequate Inadequate Partial Noet al., 200242
diabetes and exercise, self-esteem, goal-setting,etc. Education sessions were followed by a groupaerobic exercise session. The intervention formallyinvolved 6 months of sessions, but exercisesessions were available after 6 months.
One study (Wing and colleagues, 198538)compared a diet intervention with a weight loss-focused intervention. This study did not reportany between-group differences and therefore willnot be discussed further.
A second study (Wing and colleagues, 198639)compared a group who focused on the relationbetween weight loss and BG control with a groupwho focused on weight control. This study usedbehaviour modification for weight control withself-monitoring of calories by diaries. Patientdeposits were returned on the basis of meetinggoals and attendance. There were 12 weeks ofweekly meetings followed by monthly meetings forthe next 6 months and follow-up sessions at 9 and12 months.
Another study (Wing and colleagues, 198841) wassimilar to the previous one using a behaviouralweight control programme. The two groups in thisstudy differed in what they were taught aboutSMBG. One group (self-regulation) was taught howto use SMBG information to regulate behaviourusing behaviour modification principles. The othergroup (self-monitoring) was taught how to do SMBGbut not how to use the information. The interventioninvolved 13 sessions in 16 weeks with follow-upeducation sessions lasting until 10 months.
Assessment of effectivenessOutcomes reflecting diabetic controlTable 24 shows the results for GHb for theincluded studies that considered focusedinterventions.
The Kaplan intervention involving combined dietand exercise35 produced significantly lower HbA1c
than in a control group who received only didacticeducation. The diet plus exercise interventionproduced a sizeable reduction in HbA1c whereas thedrop was small in the diet group and HbA1c
increased from baseline in the exercise group andeducation group. The diet plus exercise interventionwas the most intensive intervention involving 20hours of contact, but it lasted only 10 weeks.Therefore, this effect was reasonably long lastingas the outcome was measured at 18 months.
In the Uusitupa study,36 mean levels of HbA1c didnot differ between intervention and control groups
(although there was a marginal difference at 12months), but the proportion of patients with HbA1c
≤ 7.0% was greater in the intervention group.This was true at both the 12- and 24-monthevaluations. Again, this was a long-lasting effect asthe intervention ceased at 12 months. In theGilliland CCT,42 despite all groups seeing anincrease in HbA1c the two intervention groupscombined showed a significantly smaller rise thanthe control group.
The Samaras exercise study40 reported no overallsignificant differences in HbA1c betweenintervention and control patients. However, HbA1c
levels among patients who were treated withmetformin or diet alone rose less in interventionpatients (change +0.4) than in control patients(+1.5%), p < 0.05. The fact that HbA1c rose inboth groups is not encouraging.
The remaining four studies did not report anydifferences in measures of GHb betweenintervention and control groups (Ridgewaystudy37) or between different interventions (Wingstudies38,39,41).
BPOnly two studies36,42 reported BP results. Therewere no significant differences between interventionand control groups in the Uusitupa study, whereasthere was a significant difference in diastolic BPbetween the two intervention groups combined[FF –6.5 (2.0), OO –0.4(1.7)] and the controlgroup [–0.3 (2.1)] in the Gilliland CCT.
BMI/weightFive studies reported either BMI or weight.36,37,40–42
In none of these studies was there a significantdifference between intervention and controlgroups. In one study42 there was a significantdifference in weight between the two interventiongroups combined [FF –2.0 (1.5), OO –1.8(1.5)]compared with the control group [+1.7 (1.8)].
Cholesterol and triglyceridesFour studies reported cholesterol and triglyceridelevels.36,37,40,42 There were no reported differencesin cholesterol, HDL cholesterol, low-densitylipoprotein (LDL) cholesterol or triglycerides inthese studies.
Treatment intensityUusitupa and colleagues36 reported the percentageof patients taking glucose-lowering drugs. At 24months 12.5% of intervention patients were takingdrugs whereas 34.8% of control patients were, p < 0.01. Wing and colleagues39 reported no
Effectiveness of interventions for Type 2 diabetes
36
Health Technology Assessment 2003; Vol. 7: No. 22
37
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
TABLE 24 GHb (%) findings from studies of focused education in adults with Type 2 diabetes
Study Outcome n Time Intervention Control Difference point (mean ± SD unless stated between groups
otherwise)
Kaplan HbA1 % Initial total: 87 Diet Exercise Diet + Education Overall difference et al., In analysis: 76 Baseline exercise between groups, 198735 8.97 8.16 9.18 8.21 p < 0.10; diet + RCT (2.82) (3.44) (2.46) (1.54) exercise differs from
education, p < 0.0518 months 8.51 9.46 7.70 8.57
Uusitupa HbA1c % Initial total: Baseline 7.1 (1.8) 7.8 (2.0)et al., 86 (40/46)1992, 92, In analysis 12 months 6.6 (1.6) 7.5 (1.7) p = 0.0693, 93, (24 months): 94, 9636 82 (38/44) 24 months 7.2 (1.9) 8.0 (1.6) NSRCT
Uusitupa HbA1c % Initial total: Baseline 7.4 7.8et al., (adjusted) 86 (40/46)1992, 92, In analysis 12 months 6.7 7.3 NS93, 93, (24 months): 94, 9636 82 (38/44) 24 months 7.4 7.9 NSRCT
Uusitupa HbA1c % Initial total: Baseline NR NRet al., patients 86 (40/46)1992, 92, with In analysis 12 months 74.4% 47.8% p < 0.0193, 93, ≤ 7.0% (24 months): 94, 9636 82 (38/44) 24 months 55.3% 31.8% p < 0.05RCT
Ridgeway GHb Initial total: Baseline 12.3 (2.2) 12.3 et al., 56 (28/28) (SD3.0)199937 In analysis:RCT 38 (18/20) 12 months 11.52 11.64 NS
Gilliland HbA1c % Initial total: 159 Reported FF OO +1.2 (0.4) Between 3 groups, et al., (adjusted) In analysis: values are p < 0.05200242 104 (32/39/33) changes +0.5 (0.3) +0.2 (0.3)CCT from Between FF/OO
baseline combined and control, p < 0.05
Other focused interventionsWing HbA1 Initial total: Baseline Weight control Glucose monitoringet al., 50 (25/25) 10.86 (2.0) 10.19 (2.51)198639 In analysis: 12 monthsRCT 45 (22/23) 10.44 (2.16) 10.19 (2.29)Weight vs SMBG
continued
significant differences in medication decreasesbetween patients trained in weight control andthose trained in glucose self-monitoring.
Outcomes reflecting QoL and cognitivemeasuresOne study35 considered QoL effects using avalidated measure (see Appendix 6). In this studyQoL was significantly better in diet (+0.03) anddiet plus exercise groups (+0.06) than in adidactic education control group (–0.0.04). Thedifferences are small, but placed on an overallscale of 0–1.0, they may be meaningful to patients.
Summary of results frominterventions in Type 2 diabetesA wide variety of interventions have been designedto impact on self-management of diabetes inpatients with Type 2 diabetes. Many haveattempted to instruct patients about the multiplefacets of self-care required whereas others havefocused on changing major lifestyle characteristicsthat have a negative impact on blood glucosecontrol (e.g. diet and/or exercise). There have alsobeen limited attempts to tailor interventions toparticular cultural sub-groups of the population(e.g. Mexican Americans).
Generally, these programmes have had a limitedimpact on outcomes that indicate control ofdiabetes (e.g. HbA1c), QoL, or long-term end-points (e.g. complications).
Arguably the most important indicator of diabeticcontrol is GHb. Multifaceted interventions that
affected GHb seemed either to be delivered overlong intervals or to require frequent contactbetween patients and trainers. None of themultifaceted interventions produced long-lastingeffects on GHb with limited, short-terminterventions. However, there were two focusedinterventions that did result in long-lasting effectson GHb and one CCT reported smaller increasesin HbA1c in intervention groups than in a controlgroup. Speculatively, it may be that focusedinterventions can result in longer lasting effectsbecause patients can remain focused on a singlegoal. Culturally appropriate interventions mayalso have a limited positive impact.
Reductions in the need for OHAs may also be animportant measure of the success of an intervention.This may be particularly true if GHb levels arealready relatively low in patients. Two multifacetedinterventions demonstrated reduced use of oralagents33,34 as did one focused intervention.36
From the results of these studies, it is difficult tosay what characteristics of an educationally basedintervention may be crucial to successful metaboliccontrol in Type 2 diabetes. The two multifacetedinterventions that reduced the use of oral agentswere based on the same basic programme.Surprisingly, these interventions were limited incontact (6–8 hours).
Most studies were far too short to allow for themeasurement of end-points such as diabeticcomplications. None of the studies testingparticipants with Type 2 diabetes reportedsignificant effects on end-points such as short-termcomplications or hospital admissions.
Effectiveness of interventions for Type 2 diabetes
38
TABLE 24 GHb (%) findings from studies of focused education in adults with Type 2 diabetes (cont’d)
Study Outcome n Time Intervention Control Difference point (mean ± SD unless stated between groups
otherwise)
Samaras HbA1c Initial total: 12 months + 0.86 (SEM 0.29) + 0.86 NSet al., (reported 26 (13/13) (SEM 0.27)199740 values are In analysis:RCT changes 26 (13/13)Exercise from
baseline)
Wing HbA1 Initial total: Baseline Self-regulation Self-monitoringet al., 20 (10/10) 10.57 (SEM 0.44) 10.54 (SEM 0.55)198841 In analysis: 12 monthsRCT 17 (9/8) 10.8 (SEM 0.8) 9.71 (SEM 0.78) NSself-regulation vs self-monitoring
Health Technology Assessment 2003; Vol. 7: No. 22
39
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
One study of a multifaceted intervention reporteda significant improvement in QoL.30 Again, thiswas an intervention that involved multiple sessionsspaced over the entire evaluation period and mayreflect the effects of continual contact. Anotherstudy reported significant improvements inattitudes toward diabetes in intervention patients.Any improvements in patients’ QoL or perceivedcontrol of disease are certainly desirable. However,interventions for diabetes self-management aregenerally aimed also at improving diabetic control. If an intervention only produces QoLeffects then it may well be that other interventionsfocused on QoL may produce far greater benefits in this realm (e.g. psychologicalinterventions).
Two studies reported significant improvements inpatients’ knowledge of diabetes. It is notsurprising that educational programmes shouldaffect knowledge. If anything, it is perhapssurprising that more studies did not report sucheffects. Some studies did not test for knowledgechanges or did not use a validated measure to doso. Improved knowledge is again desirable, but itsrelation to metabolic control is questionable.27
Most of the interventions aimed at Type 2 diabeteswere group interventions. The included designsdo not allow for any strong conclusions about themerits of group versus individual interventions.
However, generally those studies that reportedsignificant results used group interventions.Groups have the advantages that patients ingroups can serve as support for one another andmay form a sort of behaviour modification milieueven if the intervention itself is not formallyoriented toward behaviour modification. Inaddition, group interventions are generally lesscostly and allow staff to use the time they devoteto patient education more efficiently.
ConclusionOverall, the results of educational interventionsaimed at patients with Type 2 diabetes are difficultto interpret. There were positive effects ofinterventions in each of the Types of outcomesconsidered. However, many studies reported fewor no significant effects of educationalinterventions. It is impossible on the basis of thelimited significant intervention effects todetermine which specific characteristics of diabeteseducation for patients with Type 2 diabetes willreliably produce significant impacts on any of thereported outcomes. Because of the variations ininterventions and their impacts and also themethodological limitations of these studies, nofirm conclusions are possible about possibleeducational interventions that would havesignificant, long-lasting effects.
Trials of self-managementinterventionsA few studies have included patients with eitherType 1 or Type 2 diabetes. Although, practically,many diabetes education programmes may includepatients with both types of diabetes, these studiesare limited in their usefulness because they do notreport results separately for patients with Type 1versus Type 2 diabetes. Because of the differentaetiology, differing risk of certain complications(e.g. ketoacidosis) and different treatment optionsit would seem better to educate and evaluate thesegroups separately.
Quantity and quality of evidenceThree RCTs,44–46 apparently included patients witheither Type 1 or Type 2 diabetes. Two of thesestudies were undertaken within secondary care andin one the setting was unclear.46 One CCT,47
undertaken in primary care, does not report thetype of diabetes (see Table 25 and Appendix 9).Study sizes in the three RCTs were 206, 302 and106, respectively.44–46 Two trials compared anintervention group with a usual care control44,45
and one trial46 compared two different educationalinterventions. In this final trial, the twoeducational interventions were also compared witha non-randomised convenience control group and
Health Technology Assessment 2003; Vol. 7: No. 22
41
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Chapter 6
Effectiveness of interventions including patients with either Type 1 or Type 2 diabetes
TABLE 25 Included studies of self-management education for patients with either Type 1 or Type 2 diabetes
Reference Intervention Participants Duration of Timing of intervention evaluation
Bloomgardenet al., 198744
RCT
Two groups:1. Self-management education. Group education
provided by nurse and dietitian. No details ofnumbers of sessions
2. Usual care. Usual contact, no details provided
302 insulin-treated Type 1or Type 2patients
1.6 ± 0.3 y ineducation groupand 1.5 ± 0.3 yin control group.
1.6 ± 0.3 y and1.5 ± 0.3 y
Glasgow et al., 199745
RCT
Two groups:1. Brief dietary education. Provided by researcher
to individual patients. 20 minutes initial contactwith computer assessment then telephonecontact at weeks 1, 3, 12 + 24
2. Usual care. Clinic visits every 4 months, plustelephone call at 3 and 24 weeks
206 Type 1 or 2diabetes patients
9 months 12 months
Raji et al.,200246
Groups 1 & 2RCT, group 3matched butnon-randomised
Three groups:1. Intensive education. Team provided group
education over 3.5 days
2. Passive education. Educational materials mailedto patient’s home
3. Usual care. No details
106 patients inRCT (type notdefined) + 56matched usualcare control(those decliningparticipation)
Intervention 1,3.5 days;intervention 2,once every 3months for 12months
12 months
Gilden et al.,199247
CCT (usualcare groupnon-randomisedmatched)
Three groups:1. Self-management education
2. Self-management education plus support
3. Usual care
Groups 1 and 2: team provided group educationonce a week for 6 weeks. Group 2 had supportgroup sessions monthly for 18 months
Group 3: no details
32 patients (typenot defined)
6 weeks foreducation onlygroup.Support = 18months
24 months
any reported results from comparisons with thisgroup are effectively from a CCT. In the CCT 32patients were divided between three study groups.In three studies44,45,47 the duration of diabetes wasbetween 10 and 13 years and in one46 duration ofdiabetes was not reported. In two RCTs theproportion of patients with Type 2 diabetes was76%.44,45 This proportion was not reported for theother RCT. Mean ages within the trials rangedfrom 56 to 68 years. Trial duration differedamongst the four studies; this was 12 months inthe Glasgow44 and Raji46 studies, approximately19 months in the Bloomgarden trial,44 and 24months in the Gilden CCT.47
The quality of reporting and methodology of theincluded studies was generally poor (Tables 26 and27). The method of randomisation was reported inonly one RCT45 and concealment of allocation wasnot reported in any trial. The similarity of groupsat baseline (only HbA1c in the Raji study), and theeligibility criteria were reported in RCTs, but werenot reported in the CCT. One of the studiesreported an analysis by ITT.46
Description of interventionThree of the studies were full self-managementprogrammes44,46,47 whereas one focused on diet.45
Details of the educational interventions havevaried between reports.
The topics covered in the studies of full self-management programmes included general
knowledge about diabetes, nutrition and self-caretechniques. The Bloomgarden study44 includedonly insulin-treated patients and therefore alsocovered insulin administration. It also includedindividualised diet instruction and discussion ofmacrovascular disease. The Gilden study47
offered the same education to two groups, but in one group included social support.Therefore, topics for this group included socialwork support services and stress management. Athird group in this study received no intervention.The Raji study46 used education based on the ADA recommendations and included, in additionto the above, discussion of coronary artery disease.
The providers for the Bloomgarden study were anurse educator and a nutritionist whereas teamsprovided education in the Gilden and Raji(intervention 1) studies. The social support aspect of the Gilden study was self-directed bypatients.
The Bloomgarden study involved nine groupsessions and lasted for approximately 1.5 years.The Gilden study involved six weekly sessions for the education group and 6 weeks ofeducation plus 18 monthly sessions for theeducation plus support group. The Raji studyinvolved 3.5 days of group education for theintensive education group and mailedinformation every 3 months (four mailings) forthe passive education group.
Effectiveness of interventions including patients with either Type 1 or Type 2 diabetes
42
TABLE 26 Quality assessment of RCTs of education for either Type 1 or Type 2 diabetes
Study Randomisation Concealment Baseline Eligibility Blinding Primary ITT Missing of allocation characteristics criteria of outcome analysis values
assessors results
Bloomgarden Unknown Unknown Reported Yes Unknown Adequate Unknown Partialet al., 198744
Glasgow Adequate Unknown Reported Yes Unknown Partial Unknown Unknownet al., 199745
Raji et al., Unknown Unknown Unknown Yes Unknown Inadequate Adequate Reported200246
TABLE 27 Quality assessment of CCT of education for either Type 1 or Type 2 diabetes
Study Baseline Eligibility Blinding of Primary ITT Missing Representativenesscharacteristics criteria assessors outcome analysis values
results
Gilden et al., Unknown No Unknown Adequate Unknown Unknown No199247
One study45 used a diet intervention that involvedpatient-centred goal setting. This interventioninvolved 20 minutes of initial contact withtelephone follow-ups at 3 weeks and 3 and 6months. This intervention was led by a researcher.
Assessment of effectivenessOutcomes reflecting diabetic controland diabetic end-pointsThere were no significant differences between theintervention and control groups on any control orend-point measures in the Bloomgarden study.Likewise, the two intervention groups [intensive(8.0%) and passive education (8.0%)] did not differin HbA1c in the Raji study. A group of patientswho had declined participation in the Raji trialwere matched to the passive education group. Thetwo education interventions combined resulted ina significantly greater decrease from baselineHbA1c than in this non-randomised control group(HbA1c at 12 months: 8.0 ± 1.4% versus 8.6 ±1.8%), p < 0.05.
In the Gilden CCT both of the education groups(education and support, 6.6%; education alone,6.5%) had lower HbA1c than the control group(8.4%), p < 0.05, after 2 years. The two educatedgroups did not differ from one another.
Neither HbA1c nor BMI were significantlydifferent for intervention and control patients inthe Glasgow study that focused on diet.45 Thisstudy reported that serum cholesterol wassignificantly lower in intervention patients (208)than in control participants (226), p < 0.05.Results from a food habits questionnaire were alsosignificantly better in the intervention (2.06) thancontrol patients (2.26), p < 0.05. The
questionnaire measured four dimensions of fat-related dietary habits.
Outcomes reflecting QoL and cognitivemeasuresQoL was assessed using a validated measure onlyin the CCT (see Appendix 6 for details). Thisstudy47 tested QoL using a scale that had twosubscales. The QLa subscale measured moredemanding and intensive lifestyle changes due todiet, exercise and other general factors. QLbreflected less demanding behaviour includingmedication compliance and self-testing. Higherscores reflect better knowledge and perception ofQoL. Both aspects of QoL as well as total QoLscore were better in the group receiving botheducation and support than in the control group[total QoL scores (mean ± SEM): education +support = 78 ± 5; education = 71 ± 6; control =64 ± 3]. The education and support group alsohad higher total QoL scores than the educationalone group. Unfortunately, it is not clear whetherthe group receiving education alone wasstatistically compared with the control group.
KnowledgeKnowledge about diabetes was assessed using avalidated instrument in two studies.44,47 Theseknowledge findings can be seen in Table 28 anddescription of the knowledge measure inAppendix 6.
Two studies reported that interventions improvedknowledge scores. In the Gilden study47 theeducation plus support group scored better thanboth the education group and the control group.It should be noted that part of the supportsessions involved continuing education. In theBloomgarden study44 intervention patients hadhigher knowledge scores than the control patients.
Health Technology Assessment 2003; Vol. 7: No. 22
43
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
TABLE 28 Knowledge from studies of adults with either Type 1 or Type 2 diabetes
Reference n Time point Intervention (mean ± SEM Control Differences between unless stated otherwise) groups
Bloomgarden Initial total: Baseline 5.3 (SD1.6) 5.3 (SD1.7)et al., 198744 302 (145/157)
In analysis: ~19 months 5.8 (SD 1.6) 5.3 (SD 1.7) p < 0.01RCT 266 (127/139)
Gilden Initial total: Education and support Education aloneet al.,199247 32 (11/13/8)
In analysis: Baseline 36 (4) NR NRCCT 32 (11/13/8) Education/support:
24 months 38 (1) 36 (1) 34 (1) education, p < 0.05Education/support: control, p < 0.05
These effects, although statistically significant, donot appear to be large. As noted previously, thereis little indication that improved knowledge aloneis related to better overall self-management.
Summary of results from studiesincluding patients with eitherType 1 or Type 2 diabetesOn measures of diabetic control, the results of twoincluded studies46,47 suggest that it is possible tolower HbA1c levels following an educationalintervention. Both of these results involvedcomparisons between non-randomised interventionand control groups and one CCT was very small.Perhaps surprisingly, one RCT46 reported thatintensive group education and passive education(mailings) were equally effective in reducing HbA1c
when compared with a non-randomised controlgroup. It should be emphasised that despite thetwo intervention groups being randomised, thecontrol group was not. Noteworthy too is the lackof information about whether treating physicianswere blinded as to patients’ participation in thestudy. It is possible that participating patients weretreated more intensively than those who were notparticipating. This study also did not report anyinformation on the duration of diabetes and it istherefore possible that large numbers of newlydiagnosed patients might have lowered theirHbA1c simply in response to the diagnosis (this isconsistent with a substantial decrease in HbA1c inthe control group as well as the interventiongroups). Finally, this study with 99% males did notinclude a representative patient sample.
In one CCT the effects on HbA1c were long lastingas the intervention lasted for only 6 weeks and thefollow-up was at 2 years. This study was a CCTrather than an RCT. It also included only maleparticipants and had very few participants. The
degree to which these results may generaliseshould be scrutinised. It is also unfortunate thatthe statistical tests in this study are not sufficientlywell described to determine whether the educationalone group was specifically compared with thecontrol group for all measures. The inclusion ofthese comparisons could have answered importantquestions about the potential impact of educationalone.
The two remaining included RCTs did notdemonstrate significant differences in HbA1c
between the intervention and control groups. TheBloomgarden trial involved ongoing educationthroughout the study period and the time fromthe end of the intervention to the follow-up wasonly 3 months in the Glasgow trial.
There may also be an impact on QoL byeducational interventions for diabetes;unfortunately, however, only one of these includedstudies assessed QoL with a validated measure. Inthis study patients who received both educationand support reported a higher QoL than patientswho received education alone or than controlpatients. This is not surprising as the supportcomponent of this intervention was specificallyaimed at QoL.
Two studies reported significant effects onknowledge. This would be expected fromeducational interventions. Although the effectswere statistically significant they were not large.
ConclusionOverall, the evidence for effects of educationwithin mixed groups of patients is fairly limited.As in the interventions for Type 2 diabetes, itwould be difficult to draw any firm conclusionsabout what interventions or characteristics ofinterventions have a substantial impact in groupsof patients with either Type 1 or Type 2 diabetes.
Effectiveness of interventions including patients with either Type 1 or Type 2 diabetes
44
Anumber of systematic reviews of educationalinterventions in diabetes were identified (see
Appendix 10 for a list). In addition, a largenumber of literature reviews that did not usesystematic methods were located but will not bediscussed further (see Appendix 10).
The systematic reviews did not use the sameinclusion criteria as those set out for the currentreview. In particular, most did not impose anyrequirement for a long-term follow-up. Inaddition, many allowed a wider range of studydesigns, including single-group, pre-test, post-testdesigns. Owing to these differences, the reviewshave not been data extracted and will not bediscussed in detail. Instead, the bibliographies ofthese reviews have been used as sources of studiesthat meet our inclusion criteria.
Brief summaries are provided below.
Reviews of interventions in Type 1 diabetesNo systematic reviews were located that consideredinterventions only in patients with Type 1 diabetes.
Reviews of interventions in Type 2 diabetesFive systematic reviews of interventions in Type 2diabetes were located.48–52
In the review by Norris and colleagues,48 72 studiesof self-management training were included. Theyreported short-term positive effects (<6 months)for knowledge, frequency and accuracy of SMBG,self-reported dietary habits and glycaemic control.“With longer follow-up, interventions that usedregular reinforcement throughout follow-up weresometimes effective in improving glycaemiccontrol” (p. 561). This review concluded that self-management training in Type 2 diabetes iseffective in the short term, but that furtherresearch is needed.
A second review by Norris and colleagues49 wasbased on the search strategy of the previous review
and discussed a subset of the same trials includedin the above review. Thirty-one studies wereassessed to evaluate the effects of self-managementeducation on glycaemic control. As in the previousreview, studies with shorter follow-up periods thanin the current review were included. The findingswere similar to those reported above. “Self-management education improved GHb levels atimmediate follow-up, and increased contact timeincreases the effect. The benefit declines 1–3months after the intervention ceases, however,suggesting that learned behaviours change overtime” (p. 1159). Improvements in GHb averagedonly 0.26% in studies with follow-ups of ≥4months, suggesting that it is difficult to maintainimprovements in glycaemic control withoutmaintenance of educational/supportive contact.
Norris and colleagues50 also reviewed theeffectiveness and economic efficiency of self-management interventions for people with Type 2diabetes in community settings. Thirty trials metthe inclusion criteria and evaluated a variety ofoutcomes, over a range of follow-up periods. Self-management education was demonstrated to beeffective in community gathering places (e.g.community centres, libraries) in terms ofglycaemic control at 6 months. Evidence wasinsufficient for outcomes such as dietary intake,physical activity and BP and was also inadequateto assess the effects of interventions in theworkplace or at home.
A review was also conducted by the AlbertaHeritage Foundation for Medical Research.51 Thisreview stated that reliable conclusions could not bemade as to which types of programmes orcomponents are most effective in improving self-management in Type 2 diabetes or which categoryof patients might benefit most. “There is noconsistent pattern of effect across outcomes basedon type of intervention, length of educationalintervention, core team composition or type ofeducational setting; and there is no standardmethod to describe formal patient diabeteseducation programmes and interventions, thusmaking it difficult to replicate studies” (p. ii).
A review by Huang and colleagues52 focused oncardiovascular outcomes. This review included
Health Technology Assessment 2003; Vol. 7: No. 22
45
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Chapter 7
Evidence from systematic reviews
trials that varied treatment intensity and the use ofcholesterol-lowering and BP-loweringinterventions. As they were not specifically trials ofpatient education, this review will not be discussedfurther.
Reviews of interventions indiabetes generallyTen reviews included studies that recruitedpatients with either Type 1 or Type 2 diabetes.Although these reviews may be useful in relationto practical programmes that may include patientswith either type of diabetes, it does seem moreuseful to consider each type of diabetes separatelyfor reasons discussed previously.
A series of reviews by Brown53–55 and another byPadgett and colleagues56 seem to be mostfrequently cited and influential. The originalBrown meta-analysis included 47 studies that werewidely variable across a range of characteristics(e.g. design, intervention). Despite this, resultswere pooled to determine overall effects ofeducational intervention with the result thateducation was deemed to yield positive results.However, the usefulness of combining suchdisparate studies across multiple outcomes is veryquestionable. There was no indication as towhether positive results were long-lasting.
In a follow-up to the original meta-analysis,Brown54 included further studies and moreoutcomes of a psychological nature, including 82studies. The methods and results differed littlefrom the original review. It was concluded thateducation led to positive results for knowledge,self-care behaviours, insulin injection and weightloss, metabolic control and psychologicaloutcomes. Again, there was no indication as towhether these results were from trials withreasonably long follow-up periods, and there wasalso no differentiation as to results for patientswith Type 1 versus Type 2 diabetes. Data from thesecond Brown meta-analysis were re-analysed55 toconsider more closely the effects of study andpatient characteristics on patient outcomes. Thisreview included 73 studies and concluded thateducation was more effective in younger patients,particularly for knowledge outcomes. HbA1c levelsimproved in the short term (up to 6 months), butimprovements were lost after 6 months. In thisanalysis length of the intervention did not appearto influence outcomes. Generally, smaller effectswere found in studies with more rigorousmethods.
A further meta-analysis by Brown and Hedges57
was focused on testing a particular theoreticalmodel for predicting metabolic outcomes. Owingto the very specific nature of this analysis it willnot be discussed further.
A systematic review by Padgett and colleagues56
included 93 studies. This review focused onevaluating the nature of the intervention. Thereview concluded that there was an overallmoderate positive effect of educationalintervention. Effects were greatest for physicaleffects (although this outcome was not definedand could include a wide variety of measures) andfor knowledge. Diet and social learninginterventions were most effective. Generally,patient characteristics and type of interventionwere not correlated with effect sizes. Again, theseresults are combined across widely divergentstudies including studies of children andadolescents in addition to adults. There is littleindication as to whether effects were long-lasting,but separate analyses on a small number of studiesindicated that effects diminished over time. Forinstance, an effect size in four studies of +0.36 forHbA1c at 6 months was reduced to +0.03 at 12months (this supports the inclusion of trials with aminimum follow-up of 12 months in the currentreview).
Six additional systematic reviews were located.58–63
One of these63 was a review of computerisededucation and included only five trials in diabetes.Another focused on computer-based systemsprimarily oriented toward patient management.60
These will not be discussed further here. Albanoand Jacquemet58 included 37 papers and focusedprimarily on how interventions are reported. Theyconcluded that educational interventions are notwell described and that interventions focus on avery narrow range of possible outcomes. A reviewby Fain and colleagues61 included 78 studies, butfailed to offer summary statements aboutoutcomes, instead again lamenting the narrowrange of outcomes evaluated and poordescriptions of interventions. Whittemore62
included 71 studies in her review. This reviewagain concluded that there were positive outcomesassociated with programmes that focused on self-management, emphasised behavioural strategiesand provided culturally relevant information.Once again, however, a very diverse set of studiesare combined and we are left with little idea as tospecific intervention strategies that are effectiveand whether effects are long-lasting. Griffin andcolleagues59 in a report to the British DiabeticAssociation (BDA) [D(UK)] reviewed 57 trials and
Evidence from systematic reviews
46
seven meta-analyses of a variety of interventions,including some of practitioner education. Theyalso concluded that educational programmes arebeneficial for patients across a range of outcomes.However, they also stress that limitations of theresearch methods reduce the strength of theevidence provided.
A number of worrying methodologicalshortcomings of studies in diabetes education werenoted in the systematic reviews (e.g. inadequatedescription, lack of theoretical model, attrition).Most of these correspond with the shortcomings ofthe studies discussed in this report in the section‘Other issues and methodological concerns’ (p. 68).
Health Technology Assessment 2003; Vol. 7: No. 22
47
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Reviews of clinical interventions such as surgical or pharmacological interventions
would include an explicit discussion of the adverse effects associated with the intervention. In the case of an intervention such as patienteducation the definition of adverse effects is not soclear.
It has been mentioned in the context of trials ofintensified treatment in Type 1 diabetes that theseinterventions may increase the risk ofhypoglycaemic episodes. This elevated risk wasalso reported in early trials of intensifiedtreatment such as the DCCT. However, it has beendisputed that intensified treatment necessarilyleads to an increased risk of hypoglycaemia (e.g.Berger and Mülhauser64).
Just as the potential benefits of intensifiedtreatment programmes cannot be simplisticallyattributed to the education that provides thefoundation for the programmes, the education isnot necessarily linked to adverse effects. Educationitself is not likely to be responsible for any potentialincrease in hypoglycaemia. It is more likely that theincreased use of insulin is responsible for increasesin hypoglycaemic episodes.
The included studies did not report any otheradverse effects associated with patient education.It should be pointed out, however, that manystudies did have high rates of attrition. One canonly speculate as to whether there are adverseevents such as anxiety or stress that contribute topatient drop-out.
Health Technology Assessment 2003; Vol. 7: No. 22
49
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Chapter 8
Adverse effects
Anumber of research projects of a variety ofeducational interventions for patients with
diabetes are under way.
DAFNE trialsThe DAFNE evaluation (see details in Appendix4) has been expanded and extended for another12 months to include seven more centres and upto 1000 participants. The aim is also to learn moreabout how DAFNE courses can be implementedacross the NHS. Work is also under way to developa new DAFNE programme for children with Type1 diabetes and in the future it is hoped to developa programme for people with Type 2 diabetes.
Other controlled trialsTwo other controlled trials have been identifiedfrom searches of the National Research Register.
A randomised comparative trial of group educationand distance learning in the self-management ofType 2 diabetes is currently underway in Bolton,Lancashire. The study aims to evaluate whichpatients benefit from distance learning and which
benefit from group education. Outcomes includelifestyle measures, confidence, emotionaladjustment, weight concerns, barriers to diet andmedical outcomes. The trial is expected to end inSeptember 2004 but there are no details as to thelength of follow-up.
A controlled comparison of the effectiveness oftwo education programmes for patients with Type2 diabetes is currently underway. The study aimsto evaluate whether a short, 2.5-hour session withor without exercise or a 6-week programme withor without exercise is more beneficial. Outcomesinclude GHb, BP, weight and QoL. The trial wasexpected to end in 2002 but there are no details asto the length of follow-up.
Ongoing systematic reviewsTwo systematic reviews of relevance are under wayfor the Cochrane collaboration; both reviews areexpected to be published in 2003. One of these isa review of psychological interventions forimproving glycaemic control in patients withdiabetes and the other is a review of group-basedself-management strategies in people with Type 2diabetes.
Health Technology Assessment 2003; Vol. 7: No. 22
51
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Chapter 9
Research in progress
Overview of economic assessmentThe aim of this chapter is to assess the cost-effectiveness of patient education models fordiabetes. Our economic analysis includes asystematic review of the cost-effectivenessliterature relating to patient education models fordiabetes, a review of the economic analysissubmitted to NICE by the DAFNE Study Groupand the submission to NICE from the Associationof Clinical Diabetologists (detailing experience atPoole Hospital NHS Trust). In addition, literaturerelating to the assessment of the cost-effectivenessof treatments for diabetes and the literatureconcerning modelling for diabetes have beenconsidered for comparative purposes.
MethodsA systematic literature search was undertaken foreconomic evaluations of patient education modelsfor diabetes. Methodological details of this searchare presented in Appendix 2.
A more general search of the literature wasundertaken to identify model-based economicassessments of treatment of diabetes.
Results of the systematic searchfor economic evaluations ofpatient education models fordiabetesThe literature search identified only two studies,both from the USA, that consider the economicevaluation of education models for diabetes.45,65
Kaplan and colleagues65 present a cost–utilityanalysis (CUA) alongside the findings from anRCT in Type 2 diabetes. Glasgow and colleagues45
present cost-effectiveness findings based onintermediate outcomes, alongside a RCT inpatients with Type 1 and Type 2 diabetes. Webelieve the two cost-effectiveness studies identifieddo not offer a basis on which we can assess thecost-effectiveness of patient education models fordiabetes in the context of this review, but forcompleteness they are discussed below. A numberof other studies were identified that presented
findings on costs related to patient educationmodels, and we also discuss these below.24,66–68
Economic evaluationsKaplan and colleagues65 evaluated the cost–utilityof behavioural interventions in an experimentalstudy of 76 adult patients with Type 2 diabetes.The study is based on findings from a RCT thathas been discussed in Chapter 5 of this report.35
The study reports on four groups, using twogroups for comparison in the CUA. The CUA isbased on comparison of an education controlgroup and a group undergoing a diet plus exerciseprogramme, where significant improvements inhealth status (from diet plus exercise) werereported over an 18-month period.
The education group, used as the control, wasexposed to healthcare specialists (e.g.endocrinologist, dietitian, ophthalmologist), whoeach offered a 2-hour presentation over a 10-weekperiod. The exercise and diet group receiveddetailed instruction on these two aspects over thesame time period (2-hour sessions over a 10-weekperiod). Kaplan and colleagues65 estimate costs ofthe diet and exercise intervention (1986 prices) atapproximately US$1000. Costs comprised historyand physical examination, laboratory charges,charges for behaviour modification sessions andcharges for medical consultations. No side-effectswere reported in the study, therefore costs for theseitems were not included. Benefits were estimatedbased on the reported scores on the Quality of Well-Being scale (QWB), and scores were used to reflectwell-years; QWB scores range on a continuum ofhealth from 0 (death) to 1 (asymptomatic function).Over an 18-month period, using estimates from theQWB, the diet and exercise intervention wasreported to offer 0.06 additional units of well-being(compared with baseline), and the educationcontrol was reported to result in a reduction of–0.04 units of well-being (compared with baseline);a difference of 0.092 units of well-being is reportedbetween the comparator groups at the 18-monthassessment. A cost–utility estimate of US$10,870per well-year (1986 prices) was presented by theauthors, where a 1-year benefit rate is calculatedbased on the difference between treatment andcontrol groups at each assessment point (3, 6, 12and 18 months) weighted by duration of stay (this
Health Technology Assessment 2003; Vol. 7: No. 22
53
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Chapter 10
Economic analysis
calculated 1-year rate is reported to be 0.092 unitsof well-being). The actual difference in QWB scoresat 12 months is reported to be 0.043 units.Sensitivity on the effectiveness parameter resultedin a range of cost–utility estimates ofUS$21,740–5435 per well-year.
The study by Kaplan and colleagues haslimitations. It is based on one experimental studywith very small numbers of self-selected patientsrandomly assigned across four different groups(numbers in groups are not reported); the study isdiscussed under clinical effectiveness in Chapter 5.The way in which benefits have been assessed aspart of the CUA, using indirect modelled tariffvalues for the QWB scores (detailed health statedata/scores are not provided), is open to criticism,as is the weighted 1-year benefit used in the CUA.Scores are not those of the patients themselves,but reflect scores modelled from responses fromsamples of the general public. The model for theQWB assigns a well-being score based on aclassification of study participants according to theQWB descriptive scales (i.e. mobility, physicalactivity, social activity) and a reporting ofsymptoms. The QWB uses decrements in well-being based (from a position of 1.0 reflectingasymptomatic/optimum function) on weightsderived from the general population for healthstates described using the three QWB descriptivescales, and additional decrements based onreported symptoms. For example, where patientsreport under symptoms ‘general tiredness,weakness, or weight loss’ (this is QWB symptomnumber 10), the QWB tariff reduces well-being by–0.259 (on a scale of 0–1). Given the smallnumbers of patients in intervention groups (i.e. 76patients randomised across four groups), it ispossible that average benefits could be influencedby variations in the two comparator groups, oradverse events (e.g. onset of complications) ineither group (neither baseline characteristics noradverse events are reported in the study). Furtherdetail on the study is presented in Appendices 11and 12. In the context of this review, it is notedthat the control group is an education group,albeit information only and not behaviouralstrategies, with the intervention being directed atfocused education on diet and exercise andparticipation in group exercise sessions.
Glasgow and colleagues45 report the findings froma RCT examining an intervention focused onbehavioural issues related to dietary self-management, compared with usual care, in adultswith diabetes (both types). The study findings havebeen discussed in Chapter 6 of this report and
further detail is presented in Appendix 9. Glasgowand colleagues present estimates of cost-effectiveness based on intermediate healthoutcomes (i.e. percentage reduction in dietary fat,percentage reduction in saturated fat andreduction in serum cholesterol). Benefits are basedon findings from the trial, using a dietary self-management questionnaire to identify differencesin dietary intake and physiological measures ofserum cholesterol. Costs were calculated for thecomputer-based intervention package. Cost itemsincluded computer hardware and software,delivery materials (e.g. handouts, pamphlets),supplies, labour costs for health educators, nurses,physicians and support staff, postage andtelephone charges. Capital costs were depreciatedover year 1 in the base case analysis, and base caseanalysis did not include facility space and labourcosts for training (of educators); these wereconsidered in sensitivity analyses.
Glasgow and colleagues estimate costs for thedelivery of the dietary self-managementintervention to be US$137 per participant. Costswere combined with outcomes data on fatconsumption, saturated fat consumption andserum cholesterol (there were no significant effectson HbA1c). The marginal cost per unitimprovement in these outcomes were: US$62 perreduction of each per cent in dietary fat; US$105per percentage reduction in saturated fat; and,US$8 per mg/dl reduction in serum cholesterol.Cost-effectiveness estimates were also presentedfor three different-sized potential patient groups,to reflect economies of scale (these were similar tothe study estimates above). Further detail on thisstudy is presented in Appendices 11 and 12.
Costing studiesThe literature search identified five studies, allbased outside the UK, that presented some dataon costs associated with various patient educationmodels for diabetes. These studies are discussed inoutline for information only (note: all except thestudy by Starostina and colleagues did not meetinclusion criteria specified in the review of clinicaleffectiveness).
Starostina and colleagues24 present findings froma Russian prospective controlled study to assessBG self-monitoring in Type 1 diabetes. The studyhas been discussed in Chapter 4 of this report.The intervention comprised methods prescribedby Mühlhauser and colleagues23 (discussed inChapter 4) and comprised a 5-day inpatienttreatment and training programme. The authorspresent cost estimates for the intervention in
Economic analysis
54
roubles (Rb), with costs for materials and drugsalso presented in German marks (DM). The directcosts for the hospitalisation associated with theintervention are reported at 4200 Rb (assume1994 prices; not stated in paper), with the authorspresenting cost offsets (reduced hospitalisations,and reductions in lost productivity), to establish anet cost saving associated with the intervention.Methodological uncertainties over the studyreporting also give rise to concerns (see Appendix 7).
De Weerdt and colleagues68 provide findings froma Dutch study involving 6-month follow-up ininsulin-treated patients with diabetes. QoL wasassessed using a Dutch version of the BradburnAffect-Balance Scale and a subjective rating system(where overall QoL was rated on a scale from 0 to10, low to high). The study did not identify anystatistically significant differences in outcomes(e.g. QoL, HbA1c, adverse events) and thereforeno cost-effectiveness analysis was undertaken. Theauthors do provide some insight to the costsassociated with the intervention. The interventionwas outpatient based and consisted of four weeklygroup sessions of 3 hours duration, for groups ofabout 10 patients. The programme was structuredand consisted of video, written and practicematerials, with relevant aspects of self-carediscussed throughout. The education sessions wereled by a trained nurse, a dietitian or a patient withdiabetes, with a physician present at the beginningof each session. The authors present estimates ofthe cost of the education programme. Each singleeducation programme involved 4 hours ofphysician time, 14 hours for the session leader(healthcare worker or patient) and 18 hours foreach participant. Costs per education programmewere estimated at NLG 1325 (US$795), andestimated costs per patient were NLG 165(US$100), based on an average of eight patientsper programme. With costs of other educationmaterials taken into account the cost per patientincreased to NLG 240 (US$144). These costsinclude the cost for participants’ time. In theoverall assessment of cost the authors found nosignificant differences in the use of health services,no significant difference in the number of sickdays for patients, no differences in insulin doseand that the frequency of BG monitoringincreased in the experimental groups. Informationis not presented for the costs of the control group.
Pieber and colleagues67 report findings from aprospectively controlled trial to assess the efficacyof a treatment and teaching programme inpatients with Type 2 diabetes in Austria. The
intervention group comprised 53 patientsundergoing a structured DTTP and the controlconsisted of 55 patients without the programme.The DTTP consisted of four weekly teachingsessions (90–120 minutes each) for groups of 4–8patients. The follow-up was 6 months anddifferences were detected in outcomes related toglycaemic control. The authors do not presentdisaggregated cost analysis. They report that theDTTP reduced routine health care costs by anaverage of 594 Austrian schillings (UK £33) perpatient per year due to the reduced prescriptionof OHAs. The cost for glycosuria self-monitoringin the intervention group was 8% and the learningmaterial 6% of the routine diabetes treatmentcosts. Similarly, Gagliardino and Etchegoyen66
present findings from an observational study in asample of Type 2 diabetic patients (n = 446) in 10Latin American countries. The interventioncomprised a structured educational model,covering four weekly teaching units (90–120minutes each) and a reinforcement session at 6months. The authors present some findings on thecosts associated with the intervention, althoughthey do not present estimates of the actualintervention costs. Findings from Gagliardino andEtchegoyen indicate that the intervention resultedin a decrease in drug use, as there was asignificant reduction in the percentage of patientstaking OHAs, antihypertensive drugs andcholesterol lowering agents (all with p < 0.05).However, we must remain aware of methodologicalconcerns with respect to these studies that mayintroduce bias in various forms.
Gruesser and colleagues69 report a German studyto evaluate the practicability and efficacy of astructured treatment and teaching programme fornon-insulin-treated patients with Type 2 diabetesin a primary care setting. This involved a survey ofphysicians, and their office staff, who hadparticipated in a training course related to thedelivery of patient education to patients withdiabetes. The course covered materials on patienteducation methods as prescribed by Mühlhauserand colleagues23 (i.e. the Geneva–Düsseldorfmodel). The study also describes a retrospectivedata analysis for patients from 17 randomlyselected physicians’ records (physicians whoparticipated in the training course). The authorspresent limited data on costs of the educationprogramme. The authors report remunerationdata covered by health insurance for educationprogramme costs. Education costs are reported atUS$49 per patient (assuming 1992 costs), with anadditional patient cost for self-monitoring ofabout US$34 per patient. The study does not offer
Health Technology Assessment 2003; Vol. 7: No. 22
55
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
further detail on the actual cost components forthe education programme. There was a substantialreduction in the prescription of oral antidiabeticagents (e.g. glibenclamide) in patients undergoingeducation programmes.
The literature is not very clear on the costing ofeducational interventions and is characterised byheterogeneous methods for costing andpresentation.
Assessing the cost-effectiveness ofpatient education models indiabetesThis review is interested in assessing the additionalcosts associated with patient education models fordiabetes, and the additional benefits attributable tothe education models, when compared with usualcare, in order to consider the cost-effectiveness ofthe education models. Such an assessment iscomplex owing to the nature of diabetes and tothe format of patient education models, which areoften part of a wider package of care, involvingother aspects of treatment for diabetes (e.g.alterations in insulin and oral medications). Therehas been a great deal reported on the merits ofintensive insulin therapy versus conventionalinsulin therapy, and studies such as the DCCT10
and UKPDS9,70 have demonstrated that intensivetherapy is a clinically and cost-effective treatmentoption. We are not examining the benefits ofintensive versus conventional therapy in thisreview; we seek to assess the benefits of patienteducation models, and care must be taken toensure that patient education models under revieware considered on their merits, regardless of theknown benefits of more intensive diabetic therapy.Generally, patients will use the education modelsto self-manage their existing insulin treatment (i.e.either conventional or intensive therapy), or tomanage their treatment of Type 2 diabetes.However, it may also be that owing to patienteducation (and subsequent treatmentintensification) some patients will cross over fromconventional therapy to intensive therapy. Onthese occasions it is not the change in therapeutictreatment option that an economic evaluationshould seek to assess, but the role of the patienteducation model, which, due to difficulties indisentangling the costs and benefits of combinedcomponents of treatment (i.e. education andmedication), proves a difficult task.
The benefits of treatment in diabetes are primarilyassessed using clinical measures of glycaemic
control, for example HbA1c (discussed in Chapters 2 and 3), with secondary outcomemeasures often related to QoL and the incidenceof longer term diabetic complications. In theclinical review the main benefits from patienteducation models are presented as reductions inHbA1c. The evidence for Type 1 diabetes is morecompelling than that for Type 2 diabetes or formixed patient groups (types 1 and 2), wherefindings are unclear. Given these findings, theeconomic analysis considered in this report isprimarily based on Type 1 diabetes.
The majority of trials of patient education areshort term, not extending beyond a 1–2-yearfollow-up, so data on long-term outcomes are notwidely available. In order to assess the long-termimpact of health technologies in diabetestreatment, and to consider the cost-effectiveness oftechnologies, economic models have extrapolatedavailable data. We review below the economicmodelling literature as it relates to diabetes.
General literature on modelling of cost-effectiveness in treatment of diabetesOnly a limited number of model-based approacheshave assessed economic outcomes and cost-effectiveness for Type 1 or Type 2 diabetes. Table 29provides summary detail on the modellingapproaches identified. Models are described inoutline below in order to consider if they offer anopportunity to assess the cost-effectiveness ofpatient education, where differences in treatmentgroups are primarily based on HbA1c. A detailedreview of these models can be found in Appendix13, which presents a summary of a critical appraisalof these studies by Chilcott and colleagues.71
DCCT Research GroupThe DCCT10 was a multi-centre RCT comparingthe effects of intensive diabetes therapy with thoseof conventional diabetes therapy on thedevelopment and/or long-term progression ofdiabetes complications of Type 1 diabetes (IDDM).The intensive therapy was designed to achieve BGvalues as close to the normal range as possiblewith three or more daily insulin injections ortreatment with an insulin pump. Conventionaltherapy consisted of one or two insulin injectionsper day (note that conventional insulin therapywas probably less than in the UK, where mostpatients would receive two injections of mixturesper day). The DCCT has been discussed in detailelsewhere.10 Most economics assessments in thefield of diabetes have been undertaken usinglargely homogeneous modelling methods, usingthe data from the DCCT.
Economic analysis
56
The model used by the DCCT Research Group72
for Type 1 diabetes compares the lifetime benefitsand costs of conventional and intensive therapy asimplemented in the DCCT.10 The model is aMonte Carlo simulation, used to predict theincidence of microvascular and neurologicalcomplications in a hypothetical sample of 10,000persons with Type 1 diabetes. The modelrandomly selects from the hypothetical population(either a primary prevention cohort or secondaryprevention) and assigns characteristics (e.g. age,disease characteristics). It uses 12 health states tocapture disease characteristics, grouped accordingto the three major complications studied in theDCCT (retinopathy, neuropathy, nephropathy),and simulates the course of the patient’s diseaseover his or her expected lifetime. The model uses1-year cycles and at each cycle an individual is inone of five retinopathy health states, one of fournephropathy health states and one of threeneuropathy health states. The probability that apatient will advance to a more severe stage ofdisease in a given year depends on the patient’scurrent state of health, treatment regime (i.e.intensive versus conventional insulin therapy) and treatment duration. The model cycles through time at a patient level, until the patientexits the model (due to death), and then the next patient is selected from the hypotheticalsample. This process is repeated in the DCCTanalysis for a sample of 10,000 individuals. At the end of the modelling process (the simulation),the time spent in each of the treatments andhealth states and the time spent alive arecalculated, costs are assigned and mean statisticsare calculated by treatment group (conventionalversus intensive). The DCCT model does notconsider hypoglycaemic events.
The DCCT model uses empirical data on diseaseprogression, over 9 years, from the DCCT, and aseries of statistical models (Weibull models) topredict the probability of patients advancing to
differing stages of disease progression such asbackground retinopathy, and/or neuropathy [e.g.Weibull model, α × β × t(α – 1), where α and β arestatistical parameters determined by the study andt is the parameter for duration of treatment;different α and β parameters were determined toreflect conventional and intensive treatmentprobabilities of progression of disease]. Thesemethods are not transferable to the assessment ofpatient education models for diabetes, usingHbA1c, as they do not use HbA1c directly to modelthe effect of treatment.
Palmer and colleaguesThe diabetes disease model developed by Palmerand colleagues73 considers the cost-effectiveness ofa range of intensive interventions for Type 1diabetes compared with conventional therapy, toconsider optimal lifetime treatment patterns. Avariant of this model has been used in an earlierNICE submission on pioglitazone in Type 2diabetes,71 but the Type 2 model has not beenpublished to date. The Type 1 model from Palmerand colleagues is a micro-simulation model,simulating the experiences of individual patients(similar to the DCCT model). The modelcomprises a series of Markov sub-models,representing the development and consequencesof renal disease, retinopathy, amputation,myocardial infarction, stroke, majorhypoglycaemic events and ketoacidosis. The dataare generally drawn from the DCCT andWisconsin Epidemiological Study of DiabeticRetinopathy (WESDR)76 studies, largely reflectingtransit probabilities to defined health states, andthese transit probabilities are not dependent onHbA1c to differentiate between patient groups.
Tomar and colleaguesThe model by Tomar and colleagues74 is notdescribed in this report as it is based on theapproach documented by the DCCT Research
Health Technology Assessment 2003; Vol. 7: No. 22
57
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
TABLE 29 Approaches to modelling the cost-effectiveness of Type 1 and Type 2 diabetes
Study Study design Approach Intervention Diabetes
DCCT72 Modelling Cost-effectiveness Conventional versus intensive therapy Type 1
Palmer et al., 200073 Modelling Cost-effectiveness Conventional versus intensive therapy Type 1(various treatment options)
Tomar et al., 199874 Modelling Cost-effectiveness Conventional versus intensive therapy Type 1(based on DCCT (plus costing study)model above)
Eastman et al., 199775 Modelling Cost-effectiveness Conventional versus intensive therapy Type 2
Group (as above), and does not offer additionaldata to inform on the modelling of diabetes forthe assessment of patient education models. Seethe cited reference for further detail.
Eastman and colleaguesEastman and colleagues75 present a diabetes modeland subsequent cost-effectiveness findings77 forType 2 diabetes. The model predicts rates ofmicrovascular complications, cardiovascular disease(CVD) and mortality that are consistent with theknown epidemiology of Type 2 disease in the USA.
A large proportion of the authors on these paperswere part of the DCCT Research Group, and thestructure of the model is very similar to that of theDCCT model discussed above, although there aresome differences in scope (i.e. CVD) and in thedefinitions across disease health states. The modeluses a structure similar to the DCCT Type 1 model(i.e. Monte Carlo simulation techniques, 1-yearcycles, US population, sub-models for specificcomplications plus a mortality sub-model), withthe addition of a heart-disease sub-model, and theauthors apply it to Type 2 disease.
The effect of glycaemic control on microvascularcomplications is simulated by adjusting theincidence rates for complications under standardcare (data on HbA1c for standard care are fromWESDR) using HbA1c data for comprehensive care(i.e. a comparison of different modalities oftreatment, standard versus comprehensive). Inassessing microvascular complications in this waythe level of HbA1c is a direct input to the transitprobabilities used in the model (i.e. ratio of theaverage HbA1c in standard care to the averageHbA1c in comprehensive care); however, the HbA1c
inputs are transformed using a power function (toreflect risk gradients) and then multiplied usinghazard rates for standard care. [The effect ofglycaemic control is simulated by adjustingincidence rates for complications under standardcare, where relative hazard rates = cβ × hazardrates under standard care, where c = HbA1c
(comprehensive care)/HbA1c (standard care), and βis a parameter value determined from DCCTretinopathy research in IDDM patients.] Themodel presented uses hazard rates from theWESDR, where rates are the average of those forpatients taking insulin and those not takinginsulin, with rates categorised by duration ofdiabetes (e.g. 1–4, 5–9, 10–14 years). Base analysisassumes glycaemic control has no effect on CVD.Data on risk gradients are drawn from theDCCT78 and the model assumes that the DCCTrisk gradients in Type 1 apply to Type 2 disease.
Applying these modelling approaches tothe assessment of the cost-effectivenessof patient education modelsThe evaluation of patient education modelsrequires a mechanism for modifying risk of long-term complications according to HbA1c, withinpatient groups who are generally maintainingtheir mode of treatment (i.e. conventional orintensive). The published modelling approachesavailable for Type 1 diabetes do not offer anopportunity to undertake such modelling. TheType 2 model by Eastman and colleagues75,77 doesuse HbA1c, but it requires additional parameterinputs to establish transit probabilities (data onrelative risks between patient education andcontrol groups are not available).
Critical appraisal of the cost-effectiveness analysis presented inthe submission from the DAFNEStudy Group to NICEThe DAFNE programme for Type 1 diabetes (seethe section ‘How effectively is diabetes educationbeing provided at present?’, p. 7) is a form ofstructured patient education. An evaluation ofDAFNE is ongoing, soon to be published, and theDAFNE Study Group has submitted a report toNICE on the clinical and cost-effectiveness of theDAFNE intervention. The clinical effectivenessdata from DAFNE have not been included as partof this review as the design of the study does notinclude a comparison with a concurrent controlgroup for a period of ≥12 months. However, giventhe absence of literature on the cost-effectivenessof patient education models for diabetes, and theobvious interest in the DAFNE intervention, wereview the economic component of the submissionfrom the DAFNE Study Group.
In order to outline the cost-effectiveness analysisand the economic model presented by the DAFNEStudy Group, we use a structured proforma for thecritical appraisal of economic submissions.20 Weprovide an outline review of the differentcomponent parts (e.g. structure, data, analysis) ofthe model presented by the DAFNE Study Group.
Statement of the problemThe DAFNE submission contains a clear statementthat the economic analysis is assessing the cost-effectiveness of DAFNE to the NHS over a 10-yearperiod. The economic evaluation states that cost-effectiveness analysis is based on modelling thecosts and outcomes of DAFNE relative to baseline
Economic analysis
58
(current standard practice of two or three pre-specified insulin dose injections a day), and theevaluation considers Type 1 diabetes only (as thiswas the focus of the DAFNE study). Thesubmission does indicate that DAFNE has thepotential to be adapted to Type 2 diabetes, butdoes not offer any detail within the economicevaluation. The form of evaluation is cost–utilityanalysis, with results in terms of costs and quality-adjusted life-years (QALYs) presented separately.The submission reports that DAFNE is dominant(i.e. offers greater benefits than usual care and anet cost saving over time) in terms of the cost-effectiveness analysis, hence summary cost-perQALY statistics are not appropriate. However, wediscuss these findings further below, in the contextof the data and the assumptions used to considerthe cost-effectiveness of the intervention.
The comparatorThe evaluation uses current standard practice oftwo or three pre-specified insulin dose injectionsper day as the comparator. There is no discussionover the rationale behind the comparator, butearlier discussions within the NICE appraisalprocess have indicated that ‘usual care’ is difficult todefine in the context of patient education. The basecase cohort used in the DAFNE economic modelare stated as having 3.6 insulin injections per day.
Intervention: patient education modelThe DAFNE intervention is presented in detailwithin the submission, and briefly defined asDAFNE with dietary freedom and insulin doseadjustment, in the overview of the economicevaluation (see further detail in the sections ‘Howeffectively is diabetes education being provided atpresent?’, p. 7, ‘DAFNE trials’, p. 51, andAppendix 4).
Summary of the cost-effectivenessmodelThe cost-effectiveness model considers the long-term cost and benefit implications of delaying theonset of microvascular complications of diabetes.The DAFNE clinical findings show a reduction inHbA1c compared with control groups at 6 months,and differences between HbA1c are used to assessthe cost-effectiveness of the intervention overtime, when compared with current standardpractice. However, it must be noted that the dataon HbA1c used in the base analysis are not thatreported in the DAFNE study, but are based onAustrian, German and DAFNE trial data.
The model consists of a series of sub-models thatsimulate the progression of microvascular
complications (nephropathy, neuropathy,retinopathy, erectile dysfunction), plus severehypoglycaemia and ketoacidosis. Macrovascularcomplications are not addressed in the submission,which is reasonable given the uncertaintysurrounding the relationship between HbA1c andmacrovascular disease.
Sub-models for nephropathy, retinopathy andneuropathy are similar to those we see in themodels documented above.72,73,75 The modelintroduces a sub-model to describe patientexperience of erectile dysfunction, for severehypoglycaemia (although the model assumes nodifference between intervention groups withrespect to severe hypoglycaemia) and ketoacidosis.Mortality within the model is tied to thenephropathy sub-model.
Cohort informationThe economic evaluation is based on a cohortanalysis of 100 intervention and 100 controlpatients. The cohort is defined using the DAFNEtrial patient characteristics.
One important aspect of the base characteristics isthe mode of insulin treatment, that is, conventionalor intensive. The model assumes a base case of overthree insulin injections per day, which according tothe protocol used in the DCCT (DCCT is the sourcefor much of the data on hazard rates) wouldconstitute intensive therapy; however, the base caseprobabilities (hazard rates) for development ofcomplications (discussed below) are generally basedon conventional insulin therapy (fewer than threeinsulin injections per day).
Assessment of the impact of theintervention (glycaemic control)The model uses baseline data from Austrian,German and DAFNE trial data to inform on thereduction in HbA1c. The base analysis assumes areduction of –0.9% in HbA1c, with benefit assumedto remain over a 4-year period, and thereafter abenefit of –0.26% is assumed. Data from theclinical review detailed in Chapter 4 indicate thatpatient education may reduce HbA1c, although thefindings are variable. The only true test of patienteducation in Type 1 did not find a statisticallysignificant difference in HbA1c
22 (although thisstudy was a small underpowered RCT), while oneslightly larger RCT (SDIS21) and a larger CCTstudy from Mühlhauser and colleagues23 indicatethat patient education may have lasting effects onHbA1c. The DAFNE Study Group report a 0.53%reduction in HbA1c in the DAFNE interventiongroup over 12 months compared with baseline;
Health Technology Assessment 2003; Vol. 7: No. 22
59
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
data at 6 months showed a 1% reduction in HbA1c
compared with a control group (however, as stated,these results must be viewed with caution owing tothe design of the DAFNE trial).
The DAFNE submission does not discuss the useof HbA1c to predict differences in long-termcomplications. Although data from long-termtrials and epidemiological studies have providedevidence that ‘good’ metabolic control reduceschronic complications (e.g. DCCT ResearchGroup,10 WESDR76 and UKPDS79), studies do notprovide a definitive assessment of the causalrelationship between specific levels of glycaemicexposure (HbA1c) and the risk of complications,78 asconfounding is possible from a number of sources.
Clinical outcomesLong-term complicationsLong-term complications (retinopathy, neuropathyand nephropathy) are modelled based on (a)probabilities of disease (hazard rates) from DCCTfindings and from unpublished data (Eastman RCand colleagues: personal communication) and (b)a method of modifying the probability of diseaseaccording to differences in HbA1c betweenintervention and control groups. This riskmodification methodology is published byEastman and colleagues75,77 in relation to theassessment of standard care versus comprehensivecare in Type 2 patients. The equation used toprovide a relative hazard rate for long-termcomplications (using differences in HbA1c)employs values reported from the DCCT.78
The erectile dysfunction sub-model is based on astudy by Klein and colleagues,80 where probabilityof disease is determined via statistical modelling,using HbA1c as a risk-modifying variable. Risks forerectile dysfunction are dependent on theneuropathy sub-model.
Adverse eventsThe model structure includes sub-models forsevere hypoglycaemia and ketoacidosis. Severehypoglycaemia is assumed to be the same for bothintervention and control groups within the model,although there is a small effect due to differencesin mortality. Given this assumption, it may havebeen appropriate to exclude the sub-model forsevere hypoglycaemia from the model presented(although it does offer an opportunity to considerhypoglycaemia in any sensitivity analyses). Thesub-model for ketoacidosis uses effectiveness datafrom an Austrian study, which assumes a reductionin ketoacidosis events, not DAFNE trial data(where no significant differences are reported).
DAFNE trial data, and data from other aspects ofthe clinical data reviewed as part of this report(detailed in Chapters 4–6), have not identified anyconclusive difference between intervention andcontrol groups for ketoacidosis. Ketoacidosis doeshave a major impact on the DAFNE estimate ofnet costs associated with patient education models(presenting as a cost offset due to the assumedreduction in events for the education group), andsome consideration should be given to the basecase assumption in the context of the DAFNEreview.
CostsThe DAFNE submission uses only direct NHScosts for medical interventions, categorised aseither diabetic treatment or microvascularcomplications.
The assessment of the cost for DAFNE iscomprehensive, resulting in an estimated DAFNEcost per person attending of £545 (this is the costused in the economic model). This estimateincludes costs associated with delivering theDAFNE programme and the training andeducation required, together with ongoing qualityassessment, with these estimated average costs percentre spread across an expected 120 attendeesper centre per year.
Costs associated with microvascular complicationsare presented in outline, with appropriate unit costdata sources. Nephropathy and neuropathy arethe two complications with the greatest potentialcost impact, and this is borne out by the summarycost data presented in the model. Retinopathy is asignificant complication of diabetes but it is not ascostly to treat, and it should be borne in mind thatall patients should have regular screening andearly laser treatment if necessary, which willreduce visual loss considerably. Assumptionssurrounding treatment patterns for patients withnephropathy are from expert opinion. Withinneuropathy, treatment for foot ulcers is asignificant potential cost item. Given the largecosts associated with the treatment of nephropathyand neuropathy, relatively small differences inpatient experiences (intervention versus control)will produce substantial cost differences.
Costs for ketoacidosis and severe hypoglycaemiaare based on data from the NHS reference costlistings.81 The cost for ketoacidosis appearsreasonable given that the condition requireshospitalisation on each occasion. Given theassumption in the model of equal patientexperience with respect to severe hypoglycaemia
Economic analysis
60
(reflected in clinical trial data), the costs associatedwith severe hypoglycaemia should have no impacton base case analysis (other than through themortality effects across groups). The assumptionssurrounding the incidence of ketoacidosis shouldbe viewed with caution as such assumptions arenot supported by the clinical trial data in theSouthampton review of the clinical effectiveness ofthe interventions (the section ‘Assessment ofeffectiveness’, p. 16, reports that there is limitedevidence, two trials report conflicting evidence,with one RCT reporting no significant differenceand one CCT reporting a significant reduction inevents).
Benefits/utilitiesUtility data for the cost-effectiveness model arederived using data from a survey of Type 2patients and the modelling of results inconjunction with patient characteristics,complications and health state values/utilities froma direct visual analogue scale (VAS) response andan indirect score from the EQ-5D (EuroQol healthstate classification questionnaire) tariff values. Thedata applied to yield estimates of QALYsexperienced by patients are not yet published andare supported by an abstract from one of theauthors of the submission, Bagust and colleagues.82
Caution must be exercised over the interpretationand use of the data, for several reasons.
First, values from Type 2 patients, who willgenerally have developed diabetes later in life,may not be generalisable to Type 1 patients, whowill often have had the disease for most of theirlives. There is a growing literature on the contextof health state values/utilities and the importanceof adaptation effects (where patients adapt overtime to morbid conditions), and also the impact ofpatient/respondent experience of illness andduration of disease, on health values/utilities.83–85
Second, the data used to estimate QALY valueswithin the model are not clearly stated. The datapresented as an appendix to the DAFNE modelare not the data applied within the modelcalculations. The model applies utility data thatare derived using a multiplicative model andevidence to support the validity of themultiplicative model has not been provided(abstract only, as above).
The literature on health state values associatedwith diabetes is not large, but a few studiesindicate that the differences in scores betweenthose with complications and those without maynot be as large as indicated in the study described
in the DAFNE submission. Below we discuss someof the available health state valuation studies inorder to offer context and background for thepresent review, aware of the fact that some of thesestudies are small experimental studies andfindings may not be generalisable.
A recent publication from Redekop andcolleagues86 reports data from the Dutch samplestudied in the same health utility survey cited bythe authors of the DAFNE analysis. The datareported are based on health state values derivedusing health state descriptions from patients withType 2 diabetes and the EQ-5D tariff values,87 andalso direct VAS scores from the sample. Theauthors report a derived EQ-5D utility score (n = 1136) of 0.81 for patients with nocomplications and 0.72 for patients withmicrovascular complications. The VAS score was0.72 for no complications and 0.67 formicrovascular complications (n = 1224).
Wu and colleagues88 present analyses on healthstate values for diabetes derived via a mappingprocess, from SF-36 (Short-Form 36 Health StatusQuestionnaire) responses to the values availablefrom the QWB. The findings are from anexperimental study based on analysis from asample of 89 respondents completing the SF-36.The paper presents estimates of QWB scoresassociated with a move from ‘general populationhealth state values’ to a condition in whichpatients are ‘Type 1 diabetics, with nocomplications’, and from the ‘no complications’diabetic state to a state involving ‘diabeticretinopathy’, these estimates may be helpful toadd context to the present review. Table 30presents outline findings from the study by Wuand colleagues. Caution must be exercised whenconsidering the data presented in theirexperimental study.
Data from Wu and colleagues88 indicate thathealth state values associated with different statesshow only small differences in valuations, forexample for a move from ‘no complications’ to‘other’ (i.e. neuropathy or nephropathy alone) wesee a change of 0.03, 0.02 and 0.09 for the agegroups in Table 30. However, as indicated by thedata, there are some inconsistencies with thefindings from the study.
With regard to diabetic retinopathy, Brown andcolleagues89 report utility values associated withvarying degrees of visual loss from diabeticretinopathy. Utility values were elicited usingstandard gamble (SG) and time trade-off (TTO)
Health Technology Assessment 2003; Vol. 7: No. 22
61
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
techniques, across five sub-groups with varyingdegrees of visual loss, ranging from 0.85 to 0.59for TTO and from 0.70 to 0.90 for SG scores.Overall, in the sample of 95 respondents the TTOvalues were 0.77 and the SG scores were 0.88 (withvisual acuity ranging from 20/20 vision to handmotion visual acuity in the best eye).
Kiberd and Jindal,90 in a study on screening toprevent renal failure in insulin-dependent patientswith diabetes, estimate the health state utility forpatients with diabetes to be 0.838; utilities varybetween 1.0 (perfect health) and 0 (death). Theauthors determined these values using a TTOformat in a sample of 17 healthcare workers notassociated with their study (this sample consistedof nephrologists, clinical house staff, nurses andone social worker). The sample of healthcareworkers estimated values for six health states, oneof which was “insulin-dependent diabetes alone”.The authors do not report any further detail onthe health state valuation exercise.
Studies on the QoL related to diabetes indicatethat complications have a significant impact onpatient’s health-related QoL (WESDR,91 DCCT92).However, the literature on health state values fordiabetes and diabetic complications is notextensive and it is not possible to say withconfidence what the impact may be in terms of thedisutility associated with diabetic complications.Therefore, we suggest that caution should beexercised when applying the data from Bagust andcolleagues,82 which involves substantial reductionsin QALY values for some of the health states usedin the model.
MortalityMortality enters the model via the sub-model fornephropathy. Data on mortality are drawn from a10-year observational follow-up study on a sampleof 939 insulin-dependent diabetic patients.93
Incremental cost-effectivenessThe results from the DAFNE model analysispresented (base case) offer incremental costs that reflect a cost saving over time andincremental benefits over time which are positive, therefore the submission reports that the DAFNE intervention is dominant over the current standard practice. Table 31presents the base case results in the DAFNE Study Group submission (their Table 3.5).
Sensitivity analysisOne-way sensitivity analysis has been undertaken and reported in the submission. Given the large number of data points andassumptions applied in the model and the manner in which many of the assumptions mayinteract, it would have been useful to have haddetails on multivariate sensitivity analyses.Sensitivity analysis does not report impact ofvariations in the assumptions surrounding QALY values.
Economic analysis
62
TABLE 30 Age- and health-specific QWB scoresa
Age (years) General populationb Type 1 diabetes with Type 1 diabetes with Otherc
no complications retinopathy only
<45 0.82 0.73 ± 0.07 0.76 ± 0.05 0.70 ± 0.0845–64 0.75 0.68 ± 0.09 0.72 ± 0.09 0.66 ± 0.07≥ 65 0.70 0.64 ± 0.08 0.62 ± 0.07 0.55 ± 0.05
a QWB scores range on a continuum of health from 0 (death) to 1 [asymptomatic function (perfect health)].b Data on general population are from previous studies; see Wu and colleagues88 for details.c Individuals with Type 1 diabetes with diabetic neuropathy or nephropathy alone, or with other complications.From Table 3 in Wu S, Sainfort F, Tomer R. et al. Development and application of a model to estimate the impact of Type 1diabetes on health-related quality of life. Diabetes Care 1998;21:725–31.
TABLE 31 DAFNE study group base case cost-effectivenessresults
Per patient
Incremental cost Undiscounted –£3012Discounted –£2679
Incremental EQ-5D QALY Undiscounted 0.12Discounted 0.11
Incremental VAS QALY Undiscounted 0.10Discounted 0.09
Incremental life-years Undiscounted 0.05Discounted 0.05
Reproduced from Table 3.5 in the NICE submission bythe DAFNE Study Group.
Southampton assessment of cost-effectivenessGeneralAs discussed above, we have not identified suitablemodelling methodology to consider the cost-effectiveness of patient education models versususual care. The submission to NICE from theDAFNE Study Group uses data from a number ofsources, together with modelling methodspublished for Type 2 and data from unpublishedsources to estimate the cost-effectiveness of theDAFNE intervention.
In order to make some judgement as to thepotential cost-effectiveness of patient educationmodels, we use some data from the DAFNEsubmission together with other assumptions,described below.
CostsThe intervention costs estimated for the DAFNEintervention provide a good basis on which toconsider the costs for patient education models forType 1 diabetes. As with the DAFNE approach,two other clinical trials94,95 for Type 1 diabetes arebased on the methods developed by Mühlhauserand colleagues in Düsseldorf.23 The DAFNEsubmission estimates the cost for the structurededucation programme to be approximately £545per patient attending, with the programmedelivered on an outpatient basis. Should theDAFNE intervention be applied to Type 2diabetes, we would expect it to have similarresource and cost implications to those for Type 1.
We present in Appendix 14 estimates of UK staffcosts for the educational interventions describedin the four trials included in the clinical reviewcovering Type 1 diabetes, together with estimatesrelated to the DAFNE educational intervention.We estimate that the SDIS21 intervention wouldinvolve a minimum staff cost of £506 in year 1,with an ongoing staff input at approximately £145per year. The minimum staff costs for educationdescribed in studies by Terent and colleagues22
and Starostina and colleagues24 are estimated at£567 and £578, respectively, and the studydescribed by Mühlhauser and colleagues23 has anestimate of minimum staff costs of between £130and £163. All of these studies will have additionalcosts associated with educational materials,training, capital set-up costs and ongoing qualityassessment costs.
The submission to NICE from the Association ofClinical Diabetologists (ACD), detailing experience
of education at Poole Hospital, documents aprogramme of education for newly diagnosedType 2 diabetes, consisting of three diabeteseducation sessions (DESs) spread over a period of8–10 weeks, with an outpatient appointment witha consultant at 4 months following diagnosis. Thecosts associated with the diabetes educationprogramme in Poole are estimated to beapproximately £33,000 per year for the centre;approximately £66 per patient based on anestimated 500 new patients per year. This is acrude estimate of direct input resource, with someallowance for overhead costs. Other ongoing costswill need to be considered (e.g. training, audit,facility space), but it offers an indication of therelatively low intervention costs of the patienteducation developed in Poole. Using the costestimates from the ACD submission, and applyingthe estimate from the DAFNE Study Group of 120Type 1 patients trained per centre per year, wouldoffer a cost estimate of £275 per patient attendingthe programme. The submission from the ACDoffers an indication of the benefits from theprogramme; however, the methods are notdetailed and the study appears to be a pragmaticobservational study, with variations in methodsover time.
Effects/complicationsThe review of the clinical effectiveness of patienteducation models indicates that there is asignificant difference in HbA1c in relation toeducation, although the presence of othertreatment aspects in the package of care maycreate some uncertainty over the actual cause ofthe difference in HbA1c. However, assuming areduction of around 0.5% in HbA1c, as a result ofpatient education models, there are difficulties inassessing the actual clinical impact of such aneffect with respect to patients’ health outcomes.Methods for the modelling of disease progressionand cost-effectiveness using HbA1c, as a means ofdifferentiating between patient groups, are notcommon and we were unable to identify methodsrelating to Type 1 diabetes. For Type 2 disease,one approach is that of Eastman and colleagues.75
As with the DCCT analysis of Type 1 diabetes (i.e.conventional versus intensive insulin therapy), it isnot possible to establish whether HbA1c isresponsible for the reduction in incidence ofdiabetic complications, as differences with respectto changes in diabetic treatment are present. TheDCCT data do not provide a definitive assessmentof the causal relationship between specific levels ofglycaemic exposure (HbA1c) and the risk ofcomplications,78 as confounding is possible from anumber of sources. Therefore, it is difficult to
Health Technology Assessment 2003; Vol. 7: No. 22
63
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
assess the specific impact that a reduction inHbA1c will have on long-term outcomes (althoughthere is broad acceptance that a reduction inHbA1c is associated with a reduction in theincidence of long-term complications).
The DAFNE Study Group have submitted a modelthat uses the methods published by Eastman andcolleagues for Type 2 diabetes, and have been ableto apply the model to Type 1 diabetes, given dataavailable to them from further analyses byEastman and colleagues (personalcommunication). Structurally, the model reflects adisease progression model for patients withdiabetes across a number of different complicationareas. The probabilities used to transit patientsbetween states are partly from the DCCT andpartly from unpublished sources (for nephropathy,neuropathy and retinopathy), and the means ofadjusting the probability of experiencingcomplications as a result of a reduced HbA1c
measure may be reliant on the effects of changingmode of treatment as well as the effect ofimproved HbA1c. However, given the absence ofdata to inform on disease progression otherwise,the model offers some indications as toprogression of disease in an intervention versuscontrol cohort analysis.
The base analysis of the cost-effectiveness modelsubmitted to NICE also incorporates a number ofother uncertain parameter inputs. For example,we are unsure of the estimated base case clinicaleffect (HbA1c) and the estimated impact of healthoutcomes and complications on health-related QoLand QALY values. We have re-run some analysesusing the structured model provided by the DAFNE Study Group and present the findingsbelow.
Southampton changes to DAFNEmodel assumptions� Assume no effect on ketoacidosis – data from
DAFNE [British Medical Journal (BMJ)submission] does not report a significantdifference.
� Assume no difference in outpatient reviews –data from DAFNE (BMJ submission) does notreport a significant difference.
� Assume a reduction in HbA1c of 0.53% – whichis the reported difference (DAFNE submission)between the DAFNE intervention group at 12months and baseline.
� Assume annual probability of progression toESRD is 0.05, data from DCCT.72
� Assume annual probability of first amputationat 0.01, data from DCCT.72
When these alterations are used in the DAFNEmodel structure, the prediction remains one of anet cost saving, although at £668 per patient(£536 when discounted) this is not as dramatic asfound by the DAFNE base case analysis (based onthe same cohort specifications as the submittedmodel); see Table 32.
Given the changes to the input assumptions above,the DAFNE model predicts an improvement in lifeyears of 0.034 per patient (discounted incrementaleffect) and an improvement in QALYs of between0.06 (VAS) and 0.08 (EQ-5D tariff) per patient(discounted incremental effect) – smaller benefitsthan those shown by the DAFNE base caseanalysis. While the cost-effectiveness prediction isone of cost saving, together with positive benefits,the emphasis on the predicted benefits is lessimportant. However, should the DAFNEintervention result in additional costs, the benefitsestimated within the model would need to bescrutinised further.
Given the structure of the DAFNE model and themethods used to derive the QALY values appliedin the model, it is not easy to alter the inputvalues which drive the QALY calculations. We haveundertaken some sensitivity analysis on the QALYalgorithm used to estimate the QALY values (VASand EQ-5D tariff values) associated with incidenceof complications. Together with the aboveparameter inputs and the baseline patientcharacteristics, we reduced the QALY decrementsassociated with nephropathy and neuropathycomplications by 50% (of DAFNE base caseinputs). The results, shown in Table 32, were areduction in discounted QALYs saved within themodel from 0.0609 QALYs on the VAS to 0.054QALYs, and from 0.0776 QALYs per patient using the EQ-5D tariff to 0.063 QALYs perpatient.
Economic analysis
64
TABLE 32 Cost-effectiveness results based on Southamptonadjustments to the DAFNE input parameter values
Per patient
Incremental cost Undiscounted –£668Discounted –£536
Incremental EQ-5D QALY Undiscounted 0.066Discounted 0.063
Incremental VAS QALY Undiscounted 0.057Discounted 0.054
Incremental life-years Undiscounted 0.036Discounted 0.035
Given the relatively small costs associated with theDAFNE intervention, and given the 10-year timehorizon for analysis, only small improvements interms of mortality and/or health-related QoL (e.g.QALY gains) are required to enable the DAFNEintervention (and patient education generally) toappear cost-effective. For example, an additionalintervention cost of £545 together with thepredicted increase in insulin treatment costs ofapproximately £450 per patient (discounted over 10 years) would require an improvement over the same period of 0.05 QALYS to give a cost per QALY of just under £20,000, or animprovement of 0.10 QALYs to offer a cost perQALY estimate of just under £10,000. However, it
may be that we are not concerned with theadditional insulin costs in such a simplistic ‘backcalculation’, given that the comparison of intensive versus conventional insulin therapy isgenerally regarded as a cost-effective treatmentoption.10,72
Overall, given the relatively low costs and theexpectation of reduced longer term complications,the cost-effectiveness profile for the DAFNEpatient education model and similar models ofpatient education appears to be potentiallyfavourable. However, this is dependent on theclinical effectiveness of patient education models(i.e. improvements in HbA1c).
Health Technology Assessment 2003; Vol. 7: No. 22
65
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Implications for other partiesIf patient education were effective in improvingdiabetic control and reducing long-termcomplications of diabetes there would be an impacton patients, their families and other parties. QoLmay be affected in both positive and negativeways. If people with diabetes gain confidence inmanaging their condition, reduce their anxietiesand have better outcomes, then QoL should besignificantly improved. In contrast, this could beoffset if, despite increased knowledge brought aboutby the education, they feel that they are unable tomanage the disease successfully. Inability to adhereto the change in diet might be the commonestexample of failure of self-management.
Factors relevant to NHS policyThere is anecdotal evidence that patients receiveconflicting information from different healthcareproviders. Education requires a consistentapproach from all professional staff. It is thereforeimportant that any shift of diabetes care, forexample from hospital to primary care settings,should be accompanied by consistent advice (thismay be covered by the forthcoming NICEguidelines on diabetes care, and is not addressedin this report). In order to implement any onecommon learning curriculum, it is likely that therewill be a need for interprofessional education andalso a need for an organisational culture thatsupports empowerment.
Spending more time on education will requirechanges in working practices for all professionalsinvolved. Similarly, patients who have becomemore effective self-managers as a result ofsuccessful education may require healthcaredelivery of a different style from that experiencednow. One barrier to implementation may be thatsome practitioners may already feel that they areproviding adequate ‘education’ for patients withdiabetes. Furthermore, consideration needs to begiven to the current problems of staffing in certaindisciplines within the NHS (e.g. DSNs, dietitians).It is likely that most education is provided byDSNs and dietitians. Anecdotal evidenceencountered in the course of this review suggeststhat there is a shortage of both disciplines andthat funding is only part of the problem – even if
funds were available, recruitment in some areas isdifficult. Anecdotal evidence also suggests thatthere are considerable time pressures in diabetesclinics, partly due to the increases in prevalence ofboth types of diabetes, and that physician timemay also be a constraint.
The educational ‘models’ that have been reviewedin this report are mainly additional to traditionalinformal education within clinics. If staff shortagesmean that it is difficult to provide education inclinics, then that creates a significant barrier toimplementation of newer models. There may needto be a hierarchy of educational needs until suchtime as recruitment difficulties have been overcome.
ConclusionsStatement of principal findingsThe main findings of this review of patienteducation models for diabetes are summarisedbelow.
EfficacyInterventions for Type 1 diabetesThe results from studies of education for patientswith Type 1 diabetes suggest thateducation/intensified treatment programmes canproduce significant effects in terms of diabeticcontrol. These results also indicate that theseeffects may be relatively long-lasting. In addition,the results of one trial with a long-term follow-uphave demonstrated significant effects of theintervention on diabetic complications, such asretinopathy. However, it should be noted that thistrial also provided educational support throughoutthe trial. Two studies reported greater knowledgeand better diabetic control in educated groups(although one study did not test these differencesstatistically).
The benefits of diabetes control cannot beattributed solely to the education that is offered tothe patients, as in all but one study patients wereintensifying their treatment regimens. Theeducational component is part of theintensification of insulin therapy.
Interventions for Type 2 diabetesThe effects on diabetic control (e.g. HbA1c, BMI,cholesterol) were limited in studies of interventions
Health Technology Assessment 2003; Vol. 7: No. 22
67
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Chapter 11
Discussion and conclusions
teaching multiple topics of self-management forType 2 diabetes. Modest effects were demonstratedin studies focusing on diet and exercise alone.These effects were not large, but those that werepresent did appear to be relatively long-lasting.Little evidence has been put forward for the effectsof education on diabetic end-points or cognitiveoutcomes, although some positive effect on patientknowledge was demonstrated. In two studiesreporting increased knowledge, HbA1c decreasedin one and use of OHAs was reduced in the other.
These inconclusive findings are unfortunate asmost patients with diabetes have Type 2 diabetesand incidence is increasing. It would be impossibleat this point to say definitively what characteristicsof an educational programme (if any) aimed atpatients with Type 2 diabetes might produce long-term positive effects.
Education for patients is already provided,although in varying amounts, and should continueas there is likely to be little negative effect(although those patients who find themselvesunable to act on advice may have increasedanxiety due to education). However, there is littleevidence to suggest whether and how educationalprogrammes might currently be directed to achievemaximal benefit for patients with Type 2 diabetes.
Interventions for patients with either Type 1 orType 2 diabetesTwo CCTs that included patients with either Type1 or Type 2 diabetes suggested that educationcould reduce HbA1c levels. However, results fromtwo RCTs did not demonstrate any clear effects ofthe educational interventions on outcomes. Twostudies reported increased knowledge in educatedgroups, but there was no clear correspondencebetween increased knowledge and diabetic control.
Cost-effectivenessThis report is concerned with the cost-effectiveness of patient education models fordiabetes, not the cost-effectiveness of intensivediabetic therapy. The findings from the literaturereview of economic evaluations do not offer anyindications as to the cost-effectiveness of patienteducation models for diabetes. Although there arepotential benefits from education models in termsof improved glycaemic control (i.e. HbA1c), thereare difficulties in considering the cost-effectivenessof interventions in diabetes based only onimprovements in HbA1c. Trials of patienteducation are mostly short term and importantoutcomes such as diabetic complications areobserved in the longer term. Trials such as the
SDIS provide a combination of education andtreatment intensification and it is not possible toisolate the benefits of patient education.Therefore, an assessment of the cost-effectivenessof the intervention is difficult.
Intervention costs are largely direct costs ofeducation programmes, constituting NHS stafftime and subsequent capital and trainingrequirements. Costs for the intervention arerelatively small, with submissions fromsponsors/consultees estimating intervention costsat £545 per patient for a 5-day DTTP in Type 1patients to £66 per patient for an educationprogramme aimed at newly diagnosed Type 2patients. The upper cost estimate is acomprehensive assessment of resource use andNHS costs. Improvements in HbA1c are expectedto offer long-term benefits in terms of a reducedincidence of diabetic complications.
The DAFNE Study Group presents an economicevaluation that finds the DAFNE intervention costsaving over a 10-year period, with added healthbenefits (i.e. life-years saved, QALYs gained).Although there is uncertainty over some aspects ofthe economic model used to assess the cost-effectiveness of DAFNE, we would support theintervention as potentially cost-effective in Type 1patients where the benefits in terms of improvedHbA1c are significant and are considered over a10-year time horizon.
Other issues and methodologicalconcernsComplexity of the interventionsPatient education is an example of a complexintervention as it is a package of care that hasseveral interconnecting components. This presentsa number of problems for evaluation and also forthe interpretation of any demonstrated effects. Itis difficult to establish with any precision what the‘active ingredient’ causing any such effect is. It maybe, for example, that knowledge of one key topicis responsible for the effect; on the other hand itmay be that it is a subtle combination of factorsthat may thereafter be difficult to reproduce,outwith the setting in which the education wasundertaken or with the providers of the education.
Not only are educational interventions complex inthemselves, but they exist in a complexenvironment of management of a chronic disease.Educational interventions will interact with factorssuch as the medical management of diabetes, the
Discussion and conclusions
68
overall healthcare setting in which patients areroutinely seen and patient lifestyles. These factorsmay affect the effectiveness of an intervention ormay have indirect impacts through other factorssuch as compliance. Ideally these complexitieswould be considered in modelling exercises andpilot studies prior to conducting an RCT asrecommended by the Medical Research Council(MRC) framework for development and evaluationof RCTs for complex interventions.96 Few of theinterventions seem to have been developed in away such that the crucial components ofinterventions can be teased apart from the aspectsof the intervention that may be less important.
ConfoundingThere is likely to be confounding in some studiesof this nature, for instance between intensifyinginsulin treatments and the education provided inthose trials for Type 1 patients. Other confoundsmay include personal factors such as thepersonality types of participants who volunteer fora research trial and who are able to remainthroughout the duration of the trial. In somestudies the patients were to greater or lesserextents self-selected. When patients volunteer toparticipate in programmes it is always a concernthat they may be more motivated or otherwisediffer from those who have not volunteered.Similarly, results on self-report measures may becompromised as some participants may try toanticipate the desired effect or to give sociallydesirable answers; these are reasons for ensuringthat self-report measures are validated instrumentswhich may reduce some of these effects.
Quality of study designMany of the studies were of poor design. A fewthat claimed to be randomised were onlyrandomised in the broadest sense, for instancerandomly choosing the order in whichinterventions would be implemented inconsecutive groups of patients. These studies havebeen labelled CCTs in this report. Such designissues were often poorly reported.
Many studies were also fairly small and thereforelikely to be underpowered, particularly whenmultiple interventions were tested. Very fewstudies mentioned performing prior powercalculations in order to determine an appropriatesize for the study.
Quality of reportingThe quality of reporting of important designissues was poor in most studies. The method ofrandomisation was not described in most studies.
In addition, most studies made no mention of anyefforts to conceal the allocation of patients totreatment groups. This is a major shortcomingthat can produce significant bias.
Most of the included studies do not include thelevel of detail about the intervention that wouldallow for replication of the study, a basicrequirement of placing a study into the scientificliterature. This shortcoming is important, not onlyscientifically but also practically. If studies haveshown that an intervention has been effective,then sufficient detail should have been providedto allow that intervention to be implemented inother settings.
Another problem that relates to the poor qualityof reporting is an uncertainty about the nature ofthe control group in many of the studies. It hasbeen assumed in most cases that the control groupwas receiving ‘usual care’. However, in many caseswhat this consists of is unclear. The extent towhich the interventions actually differed from thecontrols is sometimes unclear. This can obscurethe determination of what in the intervention maybe effective and it may influence the size of effectthat is shown for an intervention (either an over-or an underestimate). This can also affect thegeneralisability of studies if it is not clear to whatextent a study resembles usual practice where theintervention might be implemented.
Length of follow-upBecause diabetes is a chronic disease with anatural history of worsening metabolic control andthe development of very serious long-termcomplications, it is critical to demonstrate thatinterventions can have lasting effects. Ideally, trialswould report on interventions that were conductedand then evaluated after a reasonably long follow-up in which no further intervention wasconducted. However, there are very few suchstudies in the diabetes education literature.
Clearly, studies that report results immediatelyfollowing an intervention or with very brief follow-up are not useful in this context. Such studies wereexcluded. However, studies that evaluatedoutcomes at least 12 months following theintroduction of an intervention were included. Afew of these studies involved relatively shortinterventions with long follow-ups, but many usedrelatively lengthy interventions with additionaleducational sessions at intervals perhaps lastingfor the entire year or more. With such a mix ofdesigns it is difficult to draw any conclusions aboutwhether there are time-limited interventions in
Health Technology Assessment 2003; Vol. 7: No. 22
69
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
diabetes education that are effective. It is thereforedifficult to draw any conclusions as to theoptimum length of an intervention.
AttritionMany included studies had fairly high levels ofdrop-out between initial recruitment andreporting of results. This is of concern for anumber of reasons. Most studies did not reportthat they performed an ITT analysis, insteadtesting for differences between intervention andcontrol groups on the basis of patients whoremained in each group at the time of evaluation.When there is considerable attrition this canproduce misleading results, particularly if there isdifferential attrition between groups. If, forinstance, the most motivated patients remain in anintervention while those who are less motivateddrop out, then the estimate of effectiveness for anunselected group of patients would beoverestimated. Even testing for (or statisticallyadjusting for) differences in baselinecharacteristics will not adjust for effects such asmotivational differences that are not captured inbaseline evaluations. If attrition is greater in thecontrol group than the intervention group, thiscan also affect the results. The most likely effect isto reduce the estimate of the effectiveness of theintervention as the patients who are leastmotivated toward self-management and who aremost ill are the mostly likely to leave the study.
High attrition rates affect the validity of studyresults, but they are also a practical concern. If anintervention results in very high attrition rates,then it is questionable as to whether largenumbers of patients would attend such anintervention as a component of usual care.
Theoretical underpinning to educationGiven the poor quality of reporting, it is unclearwhether certain characteristics of studies havesimply not been reported or whether they werenot incorporated into the studies. Primary amongthese is a theoretical foundation. Although healthpsychology is well established and a great numberof findings suggest that there are particularmethods of health promotion that are moreeffective than others, very little of this researchseems to have been incorporated into studies ofdiabetes education. This is a disappointing findingas an integrated theoretically motivated approachwould be more likely to make swifter progress.
TransferabilityIt is unclear to what extent educationalinterventions delivered in other countries are
transferable to the UK and it is important toconsider this within the context of theseinterventions. Cultural issues, not only of ethnicitybut also traditions and customs, may have animpact on outcomes. Patient health beliefs andattitudes may also be different from one country toanother, and finally, the healthcare context(private/state provision) may also affect outcomes.
Strengths and limitations of thereviewThis review has a number of strengths which leadto a minimisation of bias. The review isindependent of any vested interest and it bringstogether the evidence for the effectiveness ofpatient education models for diabetes by theapplication of consistent methods of criticalappraisal. It was guided by the principles forundertaking a systematic review and prior toundertaking the rapid review, the methods of thereview were set out in a research protocol(Appendix 1). This protocol defined the researchquestion, inclusion criteria, quality criteria, dataextraction process and methods employed toundertake the different stages of the review.Finally, an advisory group has informed the rapidreview from its initiation, through thedevelopment of the research protocol andcompletion of the report.
There were certain limitations placed on thisreview. Owing to differences in the design,duration, outcome measures and reporting ofstudies, synthesis of the included studies wasthrough narrative review with no formal meta-analysis. Despite being guided by the principlesfor undertaking a systematic review, owing to timerestrictions placed on the review the authors ofreferences were not contacted for further details oftheir trials where data were lacking. As publishedpapers are usually limited to 2500–5000 words, itmay be that some details of the trials are notpublished.
Implications for further researchThis report has served to highlight a shortage ofhigh-quality information regarding the efficacy ofeducation in diabetes. While the nature of thechronic disease demands that patients managediabetes themselves and obviously this cannot beachieved without education, there is little goodevidence to suggest exactly how patients should beeducated and trained in order to facilitate good
Discussion and conclusions
70
metabolic control and high QoL. If the goal offurther research is to evaluate patient educationper se, then RCTs with the following characteristicsare needed:
� long-term follow-up� explicit tests of time-limited interventions with
long-term follow-up� designs and statistical tests appropriate to test
single aspects of interventions� detailed reporting of interventions and
comparators� careful consideration of study attrition and
appropriate analysis� explicit comparisons between study and control
groups rather than within-group, before andafter measures
� inclusion of validated measures of QoL andother psychological outcomes such as stress andanxiety.
If it is acknowledged that patient education is onlya part of the care of patients with diabetes, thentrying artificially to isolate the effects of educationmay not be appropriate. In this case, the MRCframework provides useful recommendations fordeveloping evaluations of complex interventions.96
Diabetes education should be considered in thecontext of overall diabetes management includingeducation, support and behavioural change, drugtreatment and surveillance and treatment ofcomplications. These evaluations should perhapsbe considered in the broader context of
understanding theory, testing interventioninteractions and long-term surveillance of theprogramme after testing effectiveness. A broaderrange of outcome measures may be appropriate,for instance including behavioural outcomes thatmay be measured qualitatively. Qualitativeresearch which focuses on process is particularlyrelevant to practice to allow a betterunderstanding of quantitative evidence, and thereis a need for research to focus on both outcomesand processes.
The goals of treatment differ for differentpatients. In patients with Type 2 diabetes whoseBG is at a desirable level, it may be a goal toreduce or eliminate the use of oral agents or tomaintain BG within a range rather than to reduceit. Trials should make such treatment goals clearand to report separately on the basis of treatmentgoals. Newly diagnosed patients are likely to reactdifferently to patients who have been dealing withdiabetes for some time. The natural history ofType 2 diabetes will mean that treatment goalsand options are likely to change over time.Therefore, rather than reporting on mixed groupsof patients who differ in these characteristics, itwould be useful to determine what kinds oftreatment packages are most effective for differentpatient subgroups.
Research should also address the problem ofperforming systematic reviews of complexinterventions, such as diabetes treatment andteaching programmes.
Health Technology Assessment 2003; Vol. 7: No. 22
71
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
We are very grateful to the advisory panelbelow who provided expert advice and
comments on the protocol and/or draft of thisreport, but we absolve them from anyshortcomings in the final report, which remainsthe responsibility of SHTAC alone. NormanWaugh is guarantor.
We thank the following: Dr Alison Avenell,Research Fellow in Health Services Research,University of Aberdeen; Dr Helen Cooper,Lecturer in Health Care Education, University ofLiverpool; Professor Jack Dowie, Professor ofHealth Impact Analysis, London School ofHygiene and Tropical Medicine; Dr Simon Heller,Reader in Medicine, University of Sheffield; MariaLeveridge, Senior Dietitian, Peterborough DistrictHospital NHS Trust; Professor Ingrid Mühlhauser,University of Düsseldorf; Joumana Naqib, Project
Coordinator, Care Developments, Diabetes UK;Professor Robert Peveler, Professor of LiaisonPsychiatry, University of Southampton; Dr SusanRoberts, Consultant Physician, North TynesideGeneral Hospital.
We would also like to thank Mr Simon O’Neill atDiabetes UK for providing information and Ms Liz Hodson at the Information Service,Southampton Health Technology Centre.
NoteA small amount of information was submitted toNICE in confidence and references to thisinformation have been removed from this version.However, it should be noted that the Institute’sAppraisal Committee had access to the full reportto draw up their guidance.
Health Technology Assessment 2003; Vol. 7: No. 22
73
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Acknowledgements
1. Bell J, Hockaday T. Diabetes mellitus. InLedingham J, Warrell D, editors. Concise Oxfordtextbook of medicine. Oxford: Oxford UniversityPress; 2000. pp. 734–70.
2. British Diabetic Association (now Diabetes UK).Diabetes in the United Kingdom – 1996. London:British Diabetic Association; 1995.
3. Drake AJ, Smith A, Betts PR, Crowne EC, Shield J-PH. Type 2 diabetes in obese whitechildren. Arch Dis Child 2002;86:207–8.
4. Ehtisham S, Kirk J, McEvilly A, Shaw N, Jones S,Rose S, et al. Prevalence of type 2 diabetes inchildren in Birmingham. BMJ 2001;322:1428.
5. Fuller JH, Elford J, Goldblatt P, Adelstein AM.Diabetes mortality: new light on anunderestimated public health problem. Diabetologia1983;24:336–41.
6. Wong JS, Pearson DW, Murchison LE, WilliamsMJ, Narayan V. Mortality in diabetes mellitus:experience of a geographically definedpopulation. Diabet Med 1991;8:135–9.
7. Waugh NR, Dallas JH, Jung RT, Newton RW.Mortality in a cohort of diabetic patients. Causesand relative risks. Diabetologia 1989;32:103–4.
8. Gatling W, Williams Z, Houston AC, Walters D,Campbell M, Hill RD. Ten year follow-up of acommunity based diabetic population reveals anexcess mortality in middle-aged female diabeticpatients. Diabet Med 1990;6:11a.
9. UK Prospective Diabetes Study (UKPDS) Group.Intensive blood-glucose control withsulphonylureas or insulin compared withconventional treatment and risk of complicationsin patients with type 2 diabetes (UKPDS 33).Lancet 1998;352:837–53.
10. The Diabetes Control and Complications TrialResearch Group. The effect of intensive treatmentof diabetes on the development and progression oflong-term complications in insulin-dependentdiabetes mellitus. N Engl J Med 1993;329:977–86.
11. Diabetes UK. Fact sheet No 2 – Diabetes: thefigures. URL: http://www.diabetes.org.uk/infocentre/fact/fact2.htm (accessed 11 February2002).
12. Davies M, Day J. Screening for non-insulin-dependent diabetes mellitus (NIDDM): how oftenshould it be performed? J Med Screen 1994;1:78–81.
13. Audit Commission. Testing times: a review ofdiabetes services in England and Wales. London:Audit Commission; 2000.
14. Diabetes UK. Patient education for effectivediabetes self-management. Diabetes UK: London;2002.
15. Department of Health. National ServiceFramework for Diabetes: Standards. London:Department of Health; 2002.
16. Mensing C, Boucher J, Cypress M, Weinger K,Mulcahy K, Barta P, et al. National standards fordiabetes self-management education. Diabetes Care2002;25 Suppl 1:S140–7.
17. American Association of Diabetes Educators. The1999 Scope of Practice for Diabetes Educators andthe Standards of Practice for Diabetes Educators.Diabetes Educ 2000;26:519–25.
18. European Diabetes Policy Group. A guide to type 2diabetes mellitus 1998–1999. URL:http://www.diabetesguidelines.com/health/dwk/pro/guidelines/type2/3.1htm (accessed 22 January2002).
19. CRD. Undertaking Systematic Reviews of Researchon Effectiveness. York: Centre for Reviews andDissemination; 2001. No. 4.
20. Drummond MF, Jefferson TO. Guidelines forauthors and peer reviewers of economicsubmissions to the BMJ. The BMJ EconomicEvaluation Working Party. BMJ 1996;313:275-83.
21. Reichard P, Britz A, Cars I, Nilsson BY, Sobocinsky-Olsson B, Rosenqvist U. The Stockholm DiabetesIntervention Study (SDIS): 18 months’ results. ActaMed Scand 1988;224:115–22.
22. Terent A, Hagfall O, Cederholm U. The effect ofeducation and self-monitoring of blood glucose onglycosylated hemoglobin in type I diabetes. Acontrolled 18-month trial in a representativepopulation. Acta Med Scand 1985;217:47–53.
23. Mühlhauser I, Bruckner I, Berger M, Cheta D,Jorgens V, Ionescu-Tirgoviste C, et al. Evaluation ofan intensified insulin treatment and teachingprogramme as routine management of type 1(insulin-dependent) diabetes. TheBucharest–Düsseldorf Study. Diabetologia1987;30:681–90.
24. Starostina EG, Antsiferov M, Galstyan GR,Trautner C, Jorgens V, Bott U, et al. Effectivenessand cost–benefit analysis of intensive treatmentand teaching programmes for type 1 (insulin-
Health Technology Assessment 2003; Vol. 7: No. 22
75
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
References
dependent) diabetes mellitus in Moscow – bloodglucose versus urine glucose self-monitoring.Diabetologia 1994;37:170–6.
25. Greenfield S, Kaplan SH, Ware J-EJ, Yano EM,Frank HJ. Patients’ participation in medical care:effects on blood sugar control and quality of life indiabetes. J Gen Intern Med 1988;3:448–57.
26. The Diabetes Control and Complications TrialResearch Group. Adverse events and theirassociation with treatment regimens in theDiabetes Control and Complications Trial. DiabetesCare 1995;18:1415–27.
27. Glasgow RE, Osteen VL. Evaluating diabeteseducation. Are we measuring the most importantoutcomes? Diabetes Care 1992;15:1423–32.
28. Brown SA, Kouzekanani K, Garcia AA, Hanis CL.Culturally competent diabetes self-managementeducation for Mexican Americans – The StarrCounty Border Health Initiative. Diabetes Care2002;25:259–68.
29. Campbell EM, Redman S, Moffitt PS, Sanson-Fisher RW. The relative effectiveness ofeducational and behavioral instruction programsfor patients with NIDDM: a randomized trial.Diabetes Educator 1996;22:379–86.
30. Trento M, Passera P, Tomalino M, Bajardi M,Pomero F, Allione A, et al. Group visits improvemetabolic control in type 2 diabetes: a 2-yearfollow-up. Diabetes Care 2001;24:995–1000.
31. Heller SR, Clarke P, Daly H, Davis I, McCullochDK, Allison SP, et al. Group education for obesepatients with type 2 diabetes: greater success atless cost. Diabet Med 1988;5:552–6.
32. Raz I, Soskolne V, Stein P. Influence of small-groupeducation sessions on glucose homeostasis inNIDDM. Diabetes Care 1988;11:67–71.
33. Domenech MI, Assad D, Mazzei ME, Kronsbein P,Gagliardino JJ. Evaluation of the effectiveness ofan ambulatory teaching/treatment programme fornon-insulin dependent (type 2) diabetic patients.Acta Diabetol 1995;32:143–7.
34. Kronsbein P, Jorgens V, Muhlhauser I, Scholz V,Venhaus A, Berger M. Evaluation of a structuredtreatment and teaching programme on non-insulin-dependent diabetes. Lancet1988;2:1407–11.
35. Kaplan RM, Hartwell SL, Wilson DK, Wallace JP.Effects of diet and exercise interventions oncontrol and quality of life in non-insulin-dependent diabetes mellitus. J Gen Intern Med1987;2:220–8.
36. Uusitupa M, Laitinen J, Siitonen O, Vanninen E,Pyorala K. The maintenance of improvedmetabolic control after intensified diet therapy inrecent type 2 diabetes. Diabetes Res Clin Pract1993;19:227–38.
37. Ridgeway NA, Harvill DR, Harvill LM, Falin TM,Forester GM, Gose OD. Improved control of type2 diabetes mellitus: a practical education/behaviormodification program in a primary care clinic.South Med J 1999;92:667–72.
38. Wing RR, Epstein LH, Nowalk MP, Koeske R,Hagg S. Behavior change, weight loss, andphysiological improvements in type II diabeticpatients. J Consult Clin Psychol 1985;53:111–22.
39. Wing RR, Epstein LH, Nowalk MP, Scott N,Koeske R, Hagg S. Does self-monitoring of bloodglucose levels improve dietary compliance for obesepatients with type II diabetes? Am J Med1986;81:830–6.
40. Samaras K, Ashwell S, Mackintosh AM, Fleury AC,Campbell LV, Chisholm DJ. Will older sedentarypeople with non-insulin-dependent diabetesmellitus start exercising? A health promotionmodel. Diabetes Res Clin Pract 1997;37:121–8.
41. Wing RR, Epstein LH, Nowalk MP, Scott N. Self-regulation in the treatment of type II diabetes.Behav Ther 1988;19:11–23.
42. Gilliland S, Perez G, Azen S, Carter J. Strong inbody and spirit: lifestyle intervention for NativeAmerican adults with diabetes in New Mexico.Diabetes Care 2002;25:78–83.
43. Green LW, Kreuter MW. Health PromotionPlanning: an educational and ecological approach.Mountain View, CA: Mayfield; 1999.
44. Bloomgarden ZT, Karmally W, Metzger J, BrothersM, Nechemias C, Bookman J, et al. Randomisedcontrolled trial of diabetic patient education:improved knowledge without improved metabolicstatus. Diabetes Care 1987;10:263–72.
45. Glasgow RE, La Chance PA, Toobert DJ, Brown J,Hampson SE, Riddle MC. Long-term effects andcosts of brief behavioural dietary intervention forpatients with diabetes delivered from the medicaloffice. Patient Educ Couns 1997;32:175–84.
46. Raji A, Gomes H, Beard J, MacDonald P, ConlinPR. A randomised trial comparing intensive andpassive education in patients with diabetesmellitus. Arch Intern Med 2002;162:1301–4.
47. Gilden JL, Hendryx MS, Clar S, Casia C, Singh SP.Diabetes support groups improve health care ofolder diabetic patients. J Am Geriatr Soc 1992;40:147–50.
48. Norris SL, Engelgau MM, Narayan KM.Effectiveness of self-management training in type 2 diabetes: a systematic review ofrandomized controlled trials. Diabetes Care2001;24:561–87.
49. Norris S, Lau J, Smith S, Schmid C, Engelgau M.Self-management education for adults with type 2diabetes. Diabetes Care 2002;25:1159–71.
References
76
50. Norris S, Nichols P, Caspersen C, Glasgow R,Engelgau M, Jack L, et al. Increasing diabetes self-management education in community settings: a systematic review. Am J Prev Med2002;22:39–66.
51. Corabian P, Harstall C. Patient diabetes educationin the management of adult type 2 diabetes.Alberta Heritage Foundation for MedicalResearch; 2001. HTA 23: Series A.
52. Huang ES, Meigs JB, Singer DE. The effect ofinterventions to prevent cardiovascular disease inpatients with type 2 diabetes mellitus. Am J Med2001;111:633–42.
53. Brown SA. Effects of educational interventions indiabetes care: a meta-analysis of findings. Nurs Res1988;37:223–30.
54. Brown SA. Studies of educational interventionsand outcomes in diabetic adults: a meta-analysisrevisited. Patient Educ Couns 1990;16:189–215.
55. Brown SA. Meta-analysis of diabetes patienteducation research: variations in interventioneffects across studies. Res Nurs Health 1992;15:409–19.
56. Padgett D, Mumford E, Hynes M, Carter R. Meta-analysis of the effects of educational andpsychosocial interventions on management ofdiabetes mellitus. J Clin Epidemiol 1988;41:1007–30.
57. Brown SA, Hedges LV. Predicting metaboliccontrol in diabetes: a pilot study using meta-analysis to estimate a linear model. Nurs Res1994;43:362–8.
58. Albano MG, Jacquemet SP. Patient education anddiabetes research: a failure. Going beyond theempirical approaches. Acta Diabetol 1998;35:207–14.
59. Griffin S, Kinmouth AL, Skinner C, Kelly J.Educational and psychological interventions foradults with diabetes. London: British DiabeticAssociation; 1998.
60. Montani S, Bellazzi R, Quaglini S, D’Annunzio G.Meta-analysis of the effect of the use of computer-based systems on the metabolic control of patientswith diabetes mellitus. Diabetes Technol Ther 2001;3:347–56.
61. Fain JA, Nettles A, Funnell MM, Charron D.Diabetes patient education research: an integrativeliterature review. Diabetes Educator 1999;25:7–15.
62. Whittemore R. Strategies to facilitate lifestylechange associated with diabetes mellitus. J NursScholarsh 2000;32:225–32.
63. Krishna S, Balas EA, Spencer DC, Griffin JZ,Boren SA. Clinical trials of interactivecomputerized patient education: implications forfamily practice. J Fam Pract 1997;45:25–33.
64. Berger M, Mühlhauser I. Implementation ofintensified insulin therapy – a Europeanperspective. Diabet Med 1995;12:201–8.
65. Kaplan RM, Atkins CJ, Wilson DK. The cost–utilityof diet and exercise interventions in non-insulin-dependent diabetes mellitus. Health Promot 1987;2:331–40.
66. Gagliardino JJ, Etchegoyen G. A model educationalprogram for people with type 2 diabetes: acooperative Latin American implementation study(PEDNID-LA). Diabetes Care 2001;24:1001–7.
67. Pieber TR, Holler A, Siebenhofer A, Brunner GA,Semlitsch B, Schattenberg S, et al. Evaluation of astructured teaching and treatment programme fortype 2 diabetes in general practice in a rural areaof Austria. Diabet Med 1995;12:349–54.
68. de Weerdt I, Visser AP, Kok GJ, de Weerdt O, van der Veen EA. Randomized controlledmulticentre evaluation of an educationprogramme for insulin-treated diabetic patients:effects on metabolic control, quality of life, andcosts of therapy. Diabet Med 1991;8:338–45.
69. Gruesser M, Bott U, Ellerman P, Kronsbein P,Joergens V. Evaluation of a structured treatmentand teaching program for non-insulin-treated type II diabetic outpatients in Germany after the nationwide introduction of reimbursementpolicy for physicians. Diabetes Care1993;16:1268–75.
70. UK Prospective Diabetes Study (UKPDS) Group.Effect of intensive blood-glucose control withmetformin on complications in overweightpatients with type 2 diabetes (UKPDS 34). Lancet1998;352:854–65.
71. Chilcott J, Wight J, Lloyd Jones M, Tappenden P.The clinical effectiveness and cost-effectiveness ofpioglitazone for type 2 diabetes mellitus: a rapidand systematic review. Health Technol Assess 2002;5(19).
72. The Diabetes Control and Complications TrialResearch Group. Lifetime benefits and costs ofintensive therapy as practiced in the diabetes controland complications trial. JAMA 1996;276:1409–15.
73. Palmer AJ, Weiss C, Sendi PP, Neeser K, Brandt A,Singh G, et al. The cost-effectiveness of differentmanagement strategies for type I diabetes: a Swissperspective. Diabetologia 2000;43:13–26.
74. Tomar R, Lee S, Wu S, Klein R, Moss SE, Fryback DG, et al. Disease progression and cost ofinsulin dependent diabetes mellitus: developmentand application of simulation model. J Soc HealthSyst 1998;5:24–37.
75. Eastman RC, Javitt JC, Herman WH, Dasbach EJ,Zbrozek AS, Dong F, et al. Model of complicationsof NIDDM. I. Model construction andassumptions. Diabetes Care 1997;20:725–34.
Health Technology Assessment 2003; Vol. 7: No. 22
77
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
76. Klein R, Klein BE, Moss SE, Cruickshanks KJ. TheWisconsin Epidemiologic Study of DiabeticRetinopathy: XVII. The 14-year incidence andprogression of diabetic retinopathy and associatedrisk factors in type 1 diabetes. Ophthalmology 1998;105:1801–15.
77. Eastman RC, Javitt JC, Herman WH, Dasbach EJ,Copley-Merriman C, Maier W, et al. Model ofcomplications of NIDDM. II. Analysis of thehealth benefits and cost-effectiveness of treatingNIDDM with the goal of normoglycemia. DiabetesCare 1997;20:735–44.
78. The Diabetes Control and Complications TrialResearch Group. The relationship of glycemicexposure (HbA1c) to the risk of development andprogression of retinopathy in the diabetes controland complications trial. Diabetes 1995;44:968–83.
79. UK Prospective Diabetes Study (UKPDS) Group.Tight blood pressure control and risk ofmacrovascular and microvascular complications intype 2 diabetes: UKPDS 38. BMJ 1998;317:713.
80. Klein R, Klein BE, Moss SE, Cruickshanks KJ.Prevalence of self-reported erectile dysfunction inpeople with long-term IDDM. Diabetes Care 2002;19:135–41.
81. NHS. NHS Reference Costs. London: StationeryOffice; 2000.
82. Bagust A, Wilson E, Downs KE, Perry AS, Harrison DJ. Utility and quality of life in theCODE-2 study for type 2 diabetes. DiabetesForthcoming.
83. Dolan P. The effect of experience of illness onhealth state valuations. J Clini Epidemiol1996;49:551–64.
84. Dolan P. Effect of age on health state valuations. J Health Serv Res Policy 2000;5:17–21.
85. Dolan P. Whose preferences count? Med DecisMaking 1999;19:482–6.
86. Redekop WK, Koopmanschap MA, Stolk RP,Rutten GEHM, Wolffenbuttel BHR, Niessen LW.Health-related quality of life and treatmentsatisfaction in Dutch patients with type 2 diabetes.Diabetes Care 2002;25:458–63.
87. Dolan P. Modelling valuations for EuroQol healthstates. Med Care 1997;35:1095–108.
88. Wu S, Sainfort F, Tomar R, et al. Development andapplication of a model to estimate the impact oftype 1 diabetes on health-related quality of life.Diabetes Care 1998;21:725–31.
89. Brown MM, Brown GC, Sharma S, Shah G. Utilityvalues and diabetic retinopathy. Am J Ophthalmol1999;128:324–30.
90. Kiberd BA, Jindal KK. Screening to prevent renalfailure in insulin dependent diabetic patients: aneconomic evaluation. BMJ 1995;311:1595–9.
91. Klein R, Klein BE. Relation of glycemic control todiabetic complications and health outcomes.Diabetes Care 1998;21 Suppl 3:C39–43.
92. The Diabetes Control and Complications TrialResearch Group. Influence of intensive diabetestreatment on quality-of-life outcomes in thediabetes control and complications trial. DiabetesCare 1996;19:195–203.
93. Rossing P, Hougaard P, Borch-Johnsen K, ParvingHH. Predictors of mortality in insulin dependentdiabetes: 10 year observational follow up study.BMJ 1996;313:779–84.
94. Blonde L. Management of type 2 diabetes: updateon new pharmacological options. Manag Care2000;9 (8 Suppl):11–17.
95. Tildesley HD, Mair K, Sharpe J, Piaseczny M.Diabetes teaching – outcome analysis. Patient EducCouns 1996;29:59–65.
96. Medical Research Council. A framework fordevelopment and evaluation of RCTs for complexinterventions to improve health. URL:http://www.mrc.ac.uk/; 2000.
97. Dunn SM, Bryson JM, Hoskins PL, et al.Development of the diabetes knowledge (DKN)scales: forms DKNA, DKNB, and DKNC. DiabetesCare 1984;7:36–41.
98. Reichard P, Berglund A, Britz A, Levander S,Rosenqvist U. Hypoglycaemic episodes duringintensified insulin treatment: increased frequencybut no effect on cognitive function. J Intern Med1991;229:9–16.
99. Reichard P, Britz A, Carlsson P, Cars I, Lindblad L,Nilsson BY, et al. Metabolic control andcomplications of 3 years in patients with insulindependent diabetes (IDDM): The StockholmDiabetes Intervention Study (SDIS). J Intern Med1990;228:511–17.
100. Reichard P, Rosenqvist U. Nephropathy is delayedby intensified insulin treatment in patients withinsulin-dependent diabetes mellitus andretinopathy. J Intern Med 1989;226:81–7.
101. Reichard P, Pihl M. Mortality and treatment side-effects during long-term intensified conventionalinsulin treatment in the Stockholm DiabetesIntervention Study. Diabetes 1994;43:313–17.
102. Reichard P, Britz A, Rosenqvist U. Intensifiedconventional insulin treatment andneuropsychological impairment. BMJ 1991;303:1439–42.
103. Reichard P, Berglund B, Britz A, Cars I, NilssonBY, Rosenqvist U. Intensified conventional insulintreatment retards the microvascular complicationsof insulin-dependent diabetes mellitus (IDDM):the Stockholm Diabetes Intervention Study (SDIS)after 5 years. J Intern Med 1991;230:101–18.
References
78
104. Reichard P, Toomingas B, Rosenqvist U. Changesin conceptions and attitudes during five years ofintensified conventional insulin treatment in theStockholm Diabetes Intervention Study (SDIS).Diabetes Educ 1994;20:503–8.
105. Reichard P, Nilsson BY, Rosenqvist U. The effect oflong-term intensified insulin treatment on thedevelopment of microvascular complications ofdiabetes mellitus. N Engl J Med 1993;329:304–9.
106. Reichard P, Pihl M, Rosenqvist U, Sule J.Complications in IDDM are caused by elevatedblood glucose level: the Stockholm DiabetesIntervention Study (SDIS) at 10-year follow up.Diabetologia 1996;39:1483–8.
107. Reichard P. To be a teacher, a tutor and a friend:the physician’s role according to the StockholmDiabetes Intervention Study (SDIS). Patient EducCouns 1996;29:231–5.
108. Brown SA, Hanis CL. Culturally competentdiabetes education for Mexican Americans: theStarr County Study. Diabetes Educ 1999;25:226–36.
109. Laitinen JH, Ahola IE, Sarkkinen ES, Winberg RL,Harmaakorpi-Iivonen PA, Uusitupa MI. Impact ofintensified dietary therapy on energy and nutrientintakes and fatty acid composition of serum lipidsin patients with recently diagnosed non-insulin-dependent diabetes mellitus. J Am Diet Assoc1993;93:276–83.
110. Laitinen J, Uusitupa M, Ahola I, Laakso M,Siitonen O. Metabolic and dietary variables
associated with glycaemic control in patients withrecently diagnosed type II diabetes mellitus.Diabetes Nutr Metab Clin Exp 1994;7:77–87.
111. Uusitupa MI. Early lifestyle intervention inpatients with non-insulin-dependent diabetesmellitus and impaired glucose tolerance. Ann Med1996;28:445–9.
112. Vanninen E, Uusitupa M, Lansimies E, Siitonen O,Laitinen J. Effect of metabolic control on autonomicfunction in obese patients with newly diagnosedtype 2 diabetes. Diabetic Med 1993;10:66–73.
113. Vanninen E, Uusitupa M, Siitonen O, Laitinen J,Lansimies E. Habitual physical activity, aerobiccapacity and metabolic control in patients withnewly-diagnosed type 2 (non-insulin-dependent)diabetes mellitus: effect of 1-year diet and exerciseintervention. Diabetologia 1992;35:340–6.
114. Vijan S, Hofer TP, Hayward RA. Estimated benefitsof glycemic control in microvascular complicationsin type 2 diabetes. Ann Intern Med1997;127:788–95.
115. DAFNE Study Group. DAFNE (Dose Adjustmentfor Normal Eating): its clinical and costeffectiveness for education people with type 1diabetes mellitus in diabetes self-management. 2002.
116. Netten A, Rees T, Harrison G. Unit Costs ofHealth and Social Care, 2001. Canterbury:Personal Social Services Research Unit, Universityof Kent at Canterbury; 2002.
Health Technology Assessment 2003; Vol. 7: No. 22
79
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
The methods below were approved by NICE atthe start of the review.
Research questionTo undertake a systematic review of the clinicaland cost-effectiveness of models for educatingpeople with Type 1 and Type 2 diabetes mellitusin diabetes self-management.
Clarification of research questionand scope� Self-management in diabetes refers to achieving
and maintaining BG control through diet,exercise, oral medications and insulins.
� The primary questions for this review arewhether current models of diabetes self-management education are sufficiently effectivein terms of clinical indices (see outcomes below)and in terms of costs and benefits; and if not,what other models might be introduced.
� The educational interventions to be consideredin this review will be defined as available modelsfor educating people with diabetes in diabetesself-management with the likelihood that thesewill include those passively transferringknowledge, those based on principles ofempowerment, group and individualprogrammes and combinations thereof.
� The main comparator for this review will beusual care in clinics or primary care. This willvary among clinics and general practices, butwill include informal education andunevaluated, locally developed educationpackages. In many existing hospital serviceseducation will be provided by DSNs or othersspecifically trained in diabetes education. Inother cases providers may have little or noformal training. An anticipated lack of data oncurrent education provision will mean thatresearch results may not be directly comparableto particular existing programmes or to an‘average’ existing programme. Instead, it islikely that conclusions will be limited tocomparisons that exist within trials.
� Early appraisal of some literature in this areasuggests that self-management interventions are
generally complex, often including education aswell as changes in the intensity of medicaltreatment. It should be noted that there may bea low likelihood of locating trials that will beinformative about educational interventions per se (without confounding with intensity oftreatment). It may be necessary to assesspackages of care which combine, for example,more intensive insulin regimens with theeducation required to use those.
� The potential clinical benefit of an effectiveprogramme of education would be better self-management. This may be measured in thelong term by a reduced level of diabetes-relatedcomplications and in the short term bymaintenance of recommended levels of BGcontrol, as reflected by GHb levels andhypoglycaemic episodes. Other potentialbenefits would be greater flexibility of lifestyle,and hence better QoL.
� Potential economic benefits include reducedcosts associated with the treatment of diabetes-related complications.
Search strategy� We will search the following databases:
Cochrane Systematic Reviews Database,Cochrane Controlled Trials Register, NHS CRD(University of York) databases (including DARE,NHS EED and HTA database), MEDLINE(Silverplatter), PubMed (previous 6 months –for latest publications), EMBASE, PsychLit,ERIC, National Research Register, ScienceCitation Index, Social Science Citation Index,EconLit, MRC Trials database, Early WarningSystem and Current Controlled Trials.
� Searches will include RCTs, CCTs, systematicreviews and meta-analyses for evidence ofefficacy. Searches will also include terms relatingto learning mechanisms, so as to exclude trialsthat appraise the effectiveness of self-management alone, since the focus of thereview is on how to facilitate self-management,rather than whether self-management in itself isvaluable.
� Because the type of diabetes may not always beaddressed in trials and some trials may includepatients with both types of diabetes, diabetes
Health Technology Assessment 2003; Vol. 7: No. 22
81
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Appendix 1
Rapid review methods from the research protocol
types will not be searched for individually. Abroad search strategy will be used and all trialswill be collated and filtered on retrieval of theabstracts and full papers.
� Searches will be limited to the years 1980 to thepresent. Older publications will not be soughtbecause there are existing reviews that havecaptured the relevant publications prior to1980; these reviews and their included trials willbe assessed for inclusion according to theinclusion criteria (see below). Searches will alsobe limited to English language. Reportspublished only as meeting abstracts will beexcluded. Unpublished master’s dissertationsand theses will be excluded.
� Bibliographies of included studies and otherrelevant papers will be assessed for relevantstudies.
� Expert advisers will be asked to comment onthe comprehensiveness of our searches.
� The Cochrane Metabolic and EndocrineDiseases Group will be consulted.
Inclusion and exclusion criteriaSystematic reviews and meta-analyses of RCTs andCCTs (see below) and also individual RCTs andCCTs will be included.
Design� RCTs and CCTs that compare a specific
educational programme with usual care or withanother educational programme will beincluded. Because diabetes care is constantlyevolving, CCTs must have some concurrentcontrol group.
� RCTs or CCTs that compare models of groupeducation with individual education will beincluded.
Intervention� The review will be limited to educational
interventions, that is, the dissemination ofknowledge and skills brought about using anumber of approaches, which can be carriedout with the normal range of personnelavailable in diabetes care. Trials that evaluatespecific, specialised psychological interventions,such as cognitive/behavioural or psychoanalytictherapy, or counselling alone will be excluded.Educational interventions that include apsychological component will be included.
� Studies of education solely about specificcomplications (e.g. foot care) will not beincluded.
� Studies of case management interventions willnot be included.
ReportingIn order potentially to inform practice, includedstudies must be reported with sufficient detail tobe reproducible. They must describe the maincomponents of the educational programme, such as:
� what the intervention is with some descriptionof the topics covered
� who provides instruction (e.g. post andqualification)
� how education is delivered (e.g. in person, bycomputer)
� group or individual� length of intervention (length and number of
sessions)� target audience (e.g. Type 1, Type 2 or both;
newly diagnosed)� didactic or interactive instruction� training for the educators.
Educational interventions that are not describedin sufficient detail to replicate will not beincluded.
Participants� Participants should be diagnosed with Type 1 or
Type 2 diabetes using the standard diagnosticcriteria in effect at the inception of the study.Both newly diagnosed and patients withestablished diabetes will be included. In somecases the type of diabetes may not be clearlydefined in trials, in which case these will betreated as a separate subgroup of trials.
� Participants should be described as ‘adults’ or aminimum of 80% of participants should be 18years of age or older.
Outcomes� Diabetes is a chronic condition and
complications may not appear for years afterdiagnosis. Many ‘lifestyle’ interventions do nothave lasting effects. Therefore, included studiesmust report results from a minimum of 1 yearafter the beginning of the intervention.
Appendix 1
82
� To be included, studies must report at least oneof the primary outcomes: long-term BG levels(HbA1c), severe hypoglycaemic episodes,diabetes-related complications or QoL (asassessed by validated measures, e.g. SF-36).
� Additional outcomes that will be reported ifavailable within trials that meet the otherinclusion criteria will include BP, hospitaladmissions, relief of distress or anxiety, uptakeof screening (e.g. eye screening or BP checks),patient knowledge, patient satisfaction,achievement of individual treatment goals andresource use/costs. Any psychological measuresmust be evaluated with validated psychometricinstruments.
� Results that address individual preferredlearning styles or meeting the needs of ethnicminorities or others with specific needs will beincluded if they are reported in studies thatmeet the inclusion criteria set out above.
� Inclusion and exclusion criteria will be appliedby one reviewer and checked by a second. Anydisagreement will be resolved by discussion.
Inclusion and exclusion criteriafor papers on the cost-effectiveness of models ofdiabetic educationAll papers that present findings on the cost-effectiveness of educational interventions (asdefined above) when compared with usual care inclinics or primary care (as defined above), will bereviewed in detail, comprising a narrative reviewwith a tabulation of results where appropriate.
Methods of analysis/synthesis� Clinical effectiveness will be synthesised
through a narrative review with tabulation ofresults of included studies.
� Data will be combined statistically if of sufficient quantity and quality and if sufficientlysimilar by meta-analysis using Review Managersoftware.
Health Technology Assessment 2003; Vol. 7: No. 22
83
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
The databases were searched for publishedstudies and recently completed and ongoing
research. All searches were limited to Englishlanguage only. A flowchart outlining theidentification of studies is shown in Figure 1.
Clinical effectiveness searchstrategies� Cochrane Library (Issue 2, 2002) and
#1 DIABETES-MELLITUS*:ME#2 (DIABET*:TI or IDDM:TI) or NIDDM:TI)#3 #1:TI or #2:TI)#4 PATIENT-EDUCATION*:ME#5 MODELS-EDUCATIONAL*:ME#6 (#1 or #2)#7 (#4 or #5)#8 ((((((EDUCAT* or LEARN*) or TEACH*) orTRAIN*) or MODEL*) or PROGRAM*) orINTERVENTION*)#9 (#7 or #8)#10 SELF-CARE*:ME#11 SELF-MANAGE*#12 (SELF next MANAGE*)#13 (SELF-CARE or (SELF next CARE))#14 (PATIENT near (((EMPOWER* orCONTROL*) or MANAGE*) or REGULAT*))#15 ((((#10 or #11) or #12) or #13) or #14)#16 (#6 and (#9 or #15)).
� National Research Register (Issue 2, 2002).As for the Cochrane Library (above).
� MEDLINE (WebSPIRS), 1980–2002/06(((((explode ‘Diabetes-Mellitus’/all subheadingsin MIME,MJME) or ((diabet* or IDDM orNIDDM) in TI)) and ((‘Patient-Education’/allsubheadings in MIME,MJME) or (explode‘Learning-’/all subheadings in MIME,MJME) or(‘Models-Educational’/all subheadings inMIME,MJME) or (educat* or learn* or teach*or train* or model* or program* orintervention*)) and (((pt=randomized-controlled-trial) or (pt=controlled-clinical-trial))or (random* or (control* near (study or groupor trial or usual care))))) or (((explode ‘Diabetes-Mellitus’/all subheadings in MIME,MJME) or((diabet* or IDDM or NIDDM) in TI)) and((explode ‘Self-Care’/all subheadings inMIME,MJME) or (self regulat* or self manage*or self care or self monitor*) or (BG near4(monitor* or regulat* or manage* or control*))or (patient* near3 (empower* or control* ormanage* or regulat*))) and (((pt=randomized-controlled-trial) or (pt=controlled-clinical-trial))or (random* or (control* near (study or group
Health Technology Assessment 2003; Vol. 7: No. 22
85
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Appendix 2
Sources of information, including databases searched and search terms
Identified on searchingn = 2232
Excludedn = 704
Abstracts inspectedn = 928
Excludedn = 184
Full papers inspectedn = 224
Papers for appraisal anddata extraction
n = 40(40 papers reported24 included studies)
FIGURE 1 Flowchart of identification of studies (RCTs, CCTsand systematic reviews) for clinical effectiveness systematicreview. The number of references identified on initial searchingincludes duplicates from searches across multiple databases andalso references that were obviously inappropriate. These couldinclude studies considering conditions other than diabetes,studies in vitro, studies with non-educational interventions orstudies in inappropriate patient populations. When duplicatesand obviously inappropriate references were removed, 928abstracts remained for further consideration. These included afew references that were located for background informationoutside the formal effectiveness search. On the basis ofinspecting the abstracts, 704 references were excluded. Fullpapers for 224 references were retrieved and inspected. A few ofthese were retrieved for general background information ratherthan as potential clinical trials. From the full papers inspected,184 were excluded. The worksheet detailing the inclusion criteriacan be found in Appendix 3. A substantial number of papers thatwere retrieved were not reports of clinical trials, being, forinstance, descriptions of educational programmes or non-systematic reviews. Those references which were reports ofclinical studies of educational programmes, but which wereexcluded, are listed in Appendix 4 along with the reasons fortheir exclusion. Forty papers were included for full dataextraction and inclusion in the report. These 40 papers described24 RCTs or CCTs of education for patients with diabetes.
Appendix 2
86
or trial or usual care)))))) and (English in la)) or((explode ‘Diabetes-Mellitus’/all subheadings inMIME,MJME) and ((MUHLHAUSER-I inAUI:MEDS) or (BERGER-M in AUI:MEDS))).
� PubMED (Internet version), records added from21/08/01 to 19/7/021. (Diabetes Mellitus”[MESH OR diabetes OR
diabetic*) AND (educational OR educate*OR intervention*)
2. (diabetes OR diabetic*) AND (self-manage*OR self-care)
3. (diabetes OR diabetic*) AND (education*AND model*)
4. (diabetes OR diabetic*) AND (patienteducation) AND trial.
� EMBASE (WebSPIRS), 1980-2002/06(((((explode ‘diabetes-mellitus’/all subheadings)or ((diabet* or IDDM or NIDDM) in TI)) and((‘patient-education’/all subheadings) or(explode ‘learning-’/all subheadings) or(‘education-program’/all subheadings) or(‘teaching-’/all subheadings) or ((educat* orlearn* or teach* or train* or model* orprogram* or intervention* ) in TI))) or(((explode ‘diabetes-mellitus’/all subheadings)or ((diabet* or IDDM or NIDDM) in TI)) and((explode ‘self-care’/all subheadings) or (selfmanag* or self care) or (patient near3(empower* or control* or manage* orregulat*)))) or (((explode ‘diabetes-mellitus’/allsubheadings) or ((diabet* or IDDM or NIDDM)in TI)) and ((muhlhauser or berger) in AU)))and ((explode ‘clinical-trial’/all subheadings) or(meta-analy* or metaanaly* or systematic reviewor systematic overview))) and (English in la).
� Science Citation Index, 1980–18/07/2002diabet* and (trial* or random*) and (self-manage* or self-care or patient same educationor model* same education*).
� Web of Science Proceedings, 1990 to 18/07/2002diabet* and (trial* or random*) and (self-manage* or self-care or patient same educationor model* same education*).
� PsycINFO 1980–2002/07((explode ‘Diabetes-Mellitus’ in DE) or (diabet*and (PY=1980-2002) and (English in la) and(LA=ENGLISH))) and (((patient* neareducation*) and (PY=1980-2002) and (Englishin la) and (LA=ENGLISH)) or ((model* neareducation*) and (PY=1980-2002) and(LA=ENGLISH)) or (self care and (PY=1980-2002) and (LA=ENGLISH)) or (self manage*and (PY=1980-2002) and (LA=ENGLISH)))and ((trial* or random*) and (PY=1980-2002)and (LA=ENGLISH)).
� CINAHL 1982–2002/05((explode ‘Diabetes-Mellitus’/all topicalsubheadings/all age subheadings in DE) or(diabet* in ti,ab)) and ((((model* or patient*)near education*) or (self care) or (selfmanage*)) in ti,ab,sh) and (((clinical near trial)or (random*)) in ti,ab,sh).
� ERIC 1980–June 2002diabet$ and (model$ or self-care or self care orself manage$ or self-manage$ or patienteducation$) and (trial$ or random$).
� BEI (British Education Index), 1986–May 2002diabet$ and (model$ or self-care or self care orself manage$ or self-manage$ or patienteducation$) and (trial$ or random$).
� DARE and HTA Database (web version),searched on 18/7/021. diabet$ AND education2. diabet$ AND self manage$3. diabet$ AND self care$.
� BIOSIS 1985–18 July 20021. (((al: (diabet*)) and al: (self care)) and al:
(random*)) or (((al: (diabet*)) and al: (selfmanage*)) and al: (random*))
2. ((al: (diabet*)) and al: (education* wmodel*)) and al: (random*)
3. (al: diabet* n patient education) and al:random*.
Cost-effectiveness and QoL� MEDLINE (WebSPIRS), 1980–2002/07
((explode ‘Economics-’/all subheadings inMIME,MJME) or ((explode ‘Quality-Adjusted-Life-Years’/all subheadings in MIME,MJME) or(explode ‘Quality-of-Life’/all subheadings inMIME,MJME)) or (cost* or economic*) or((quality near2 life) or QALY) or (wellbeing orwell-being)) and ((((random* or (control* neartrial) or (clinical near trial)) or((PT=CONTROLLED-CLINICAL-TRIAL) or(PT=RANDOMIZED-CONTROLLED-TRIAL))or (pt=clinical-trial) or (metaanaly* or meta-analy* or (systematic* near review) or(systematic* near overview) or (pt=meta-analysis))) and ((((explode ‘Diabetes-Mellitus’/allsubheadings in MIME,MJME) or ((diabet* orIDDM or NIDDM) in TI)) and ((‘Patient-Education’/all subheadings in MIME,MJME) or(explode ‘Learning-’/all subheadings inMIME,MJME) or (‘Models-Educational’/allsubheadings in MIME,MJME) or (educat* orlearn* or teach* or train* or model* orprogram* or intervention*))) or (((explode
Health Technology Assessment 2003; Vol. 7: No. 22
87
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
‘Diabetes-Mellitus’/all subheadings inMIME,MJME) or ((diabet* or IDDM orNIDDM) in TI)) and ((explode ‘Self-Care’/allsubheadings in MIME,MJME) or (self regulat*or self manage* or self care or self monitor*) or(blood glucose near4 (monitor* or regulat* ormanage* or control*)) or (patient* near3(empower* or control* or manage* orregulat*)))) or (((explode ‘Diabetes-Mellitus’/allsubheadings in MIME,MJME) or ((diabet* orIDDM or NIDDM) in TI)) and((MUHLHAUSER-I in AUI:MEDS) or(BERGER-M in AUI:MEDS))))) and (English inla)).
� EMBASE (WebSPIRS), 1980–2002/06((explode ‘quality-of-life’/all subheadings) or(‘quality-adjusted-life-year’/all subheadings) or(explode ‘health-economics’/all subheadings) or(explode ‘economics-’/all subheadings) or (cost*or economic*) or ((quality near3 life) or qaly orwellbeing or well-being)) and ((((((explode‘diabetes-mellitus’/all subheadings) or ((diabet*or IDDM or NIDDM) in TI)) and ((‘patient-education’/all subheadings) or (explode‘learning-’/all subheadings) or (‘education-program’/all subheadings) or (‘teaching-’/allsubheadings) or ((educat* or learn* or teach* ortrain* or model* or program* or intervention*)in TI))) or (((explode ‘diabetes-mellitus’/allsubheadings) or ((diabet* or IDDM or NIDDM)in TI)) and ((explode ‘self-care’/all subheadings)or (self manag* or self care) or (patient near3
(empower* or control* or manage* orregulat*)))) or (((explode ‘diabetes-mellitus’/allsubheadings) or ((diabet* or IDDM or NIDDM)in TI)) and ((muhlhauser or berger) in AU)))and ((explode ‘clinical-trial’/all subheadings) or(meta-analy* or metaanaly* or systematic reviewor systematic overview))) and (English in la)).
� PubMED (Internet version, records added from24/12/01 to 18/07/02)diabetes AND (cost OR costs OR economic OReconomics).
� NHS EED (web version), searched on 18/07/02diabetes and (teaching or training or learningor management or education).
Additional searchingBibliographiesAll references to articles for which full papers wereretrieved were checked to ensure that no eligiblestudies had been missed.
ExpertsExperts were contacted for advice and peer reviewand to identify additional published andunpublished references and any currently ongoingstudies.
Web sitesDiabetes UK website: http://www.diabetes-uk.org.uk/home.htm.
Health Technology Assessment 2003; Vol. 7: No. 22
89
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Appendix 3
Inclusion criteria worksheet
Trial name or number:
Patients with Type 1 or Type 2 diabetes? Yes Unclear No Type:↓ ↓ →
NB exclude gestational diabetes next question next question EXCLUDE
Patients described as ‘adults’ or <20% Yes Unclear Nounder 18 years old? ↓ ↓ →
next question next question EXCLUDE
RCT or CCT or systematic review Yes Unclear NoNB CCT must have concurrent control ↓ ↓ →
next question next question EXCLUDE
Education programme? Yes Unclear NoNB exclude purely psychological/counselling ↓ ↓ →interventions next question next question EXCLUDE
Education for self-management of diabetes? Yes Unclear NoNB exclude education for prevention/ ↓ ↓ →treatment of specific complications (e.g. next question next question EXCLUDEfoot ulcer)
Comparator: educational programme vs Yes Unclear Nousual care OR another ed. programme? OR ↓ ↓ →Group programme vs individual programme? next question next question EXCLUDE
Is description of intervention sufficient to Yes Unclear Noreproduce? ↓ ↓ →NB must include topics (or content obtainable). next question next question EXCLUDEOther characteristics: provider, length and no. of sessions, target audience, mode of delivery (in person or distance), group or individual, didactic/interactive, changes in treatment
Follow-up from inception ≥ 1 year? Yes Unclear No Length of↓ ↓ → follow-up?
next question next question EXCLUDE
Report one or more of primary outcomes: Yes Unclear No CostsHbA1c OR severe hypoglycaemic episodes OR↓ ↓ → reported?diabetic complications OR QoL? next question next question EXCLUDENB other outcomes will also be included if primary outcomes reported
Final decision INCLUDE UNCLEAR EXCLUDE Results of (Discuss) Discussion:
DAFNEThe objective of the DAFNE (Dose Adjustment forNormal Eating) trial was to evaluate whether aflexible intensive insulin regimen, combiningdietary freedom with insulin adjustment training,can improve both metabolic control and QoL.Eligible patients were adults with established Type1 diabetes with moderate or poor glycaemiccontrol. The setting was secondary care diabetesclinics in three English health districts.
Participants were randomised into a waiting-listcontrolled trial. [DAFNE patients had a mean age40 years; a long duration of diagnosis (average 17years); poor glucose control at baseline. The studyhad low recruitment, with only 136 of 1000 invitedjoining. There had been one death, possiblyrelated to active treatment.] The interventiongroup, ‘immediate DAFNE’, attended a trainingcourse within 1–4 months of randomisation. Thecontrol group, ‘delayed DAFNE’, acted as waiting-list controls. They continued to receive their usualcare for 6 months, and then attended a ‘delayedDAFNE’ training course 6 months later. Thegroups were compared at baseline and 6 and 12months. The post-course follow-up was 12 monthsfor the immediate DAFNE group and 6 monthsfor the delayed DAFNE group.
The primary outcome measures were HbA1c, rateof severe hypoglycaemia and the ADDQoL (Auditof Diabetes-Dependent Quality of Life). Otherend-points included weight, lipids, satisfactionwith treatment (DTSQ) and psychological well-being (W-BQ12). HbA1c levels in the immediateDAFNE group fell by 1% for the first 6 monthsafter training. At 12 months, there had been someincrease, but levels still remained significantlylower than baseline by 0.5% (95% CI 0.1 to 0.9, p = 0.004). One-quarter (16/67) maintained a fallin HbA1c of >1.5% and four (6%) showed a rise of>1.5%. The levels in the delayed DAFNE groupremained constant for the first 6 months whilewaiting for training, and fell 0.7% 6 months afterthe training. ADDQoL scores improved and werefully maintained in immediate DAFNE. In delayedDAFNE, they remained constant and thenimproved after training. Similar patterns of
improvement to the ADDQoL were shown for theDTSQ and W-BQ12.
It was concluded that skills training was effectivein promoting dietary freedom, improved QoL andglycaemic control in people with Type 1 diabetes,without worsening severe hypoglycaemia orcardiovascular risk.
This trial does not meet the reviews inclusioncriteria for length of follow-up as there was noconcurrent control group for the 12-month follow-up period.
The Diabetes Control andComplications Trial (DCCT)The DCCT was a multicentre, RCT that comparedintensive therapy with conventional therapy andassessed their effects on the development andprogression of early vascular and neurologicalcomplications of Type 1 diabetes. All patients hadan educational component at the start of the trial,and the intensive treatment group continued tovisit their study centre each month and werecontacted even more frequently by telephone toreview and adjust their regimens. The trial doesnot meet this review’s reproducibility inclusioncriterion. The educational packages were locallydeveloped and, therefore, differed betweencentres.
United Kingdom ProspectiveDiabetes Study (UKPDS)The UKPDS was designed to establish whetherintensive BG control in patients with Type 2diabetes reduced the risk of macrovascular ormicrovascular complications. The cut-off for BGcontrol was 14 mmol/litre in the control group and6 mmol/litre in the intervention group. When BGexceeded the cut-off, treatment was altered to tryto reduce it. All patients had a 3-month dietaryrun-in period where they were seen by a physicianand dietitian. All patients also continued to receivedietary advice from a dietitian throughout thestudy period.
Health Technology Assessment 2003; Vol. 7: No. 22
91
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Appendix 4
Details of excluded studies
Although education was given to the participants,this was similar in both groups and therefore doesnot meet the inclusion criteria.
Other trials that were excludedfrom the review:Trials excluded owing to study design(i.e. not RCT or CCT or wrongcomparator)Bajaj S, Mehrotra R, Singh K, Kumar D. Assessment of
knowledge regarding metabolic control in diabetics. J Assoc Physicians India 2001;49:296–7.
Berger M. Evaluation of a teaching and treatmentprogramme for type I diabetic patients. Diabetes Educ1984;10(Spec no):36–8.
Brown SA, Hanis CL. A community-based, culturallysensitive education and group-support interventionfor Mexican Americans with NID. Diabetes Educ1995;21:203–10.
Coates VE, Boore JRP. Knowledge and diabetes self-management. Patient Educ Couns 1996;29:99–108.
Constable J, Buckingham C, Bean L. Evaluating theeffect of an education programme on quality of life.(Research on effectiveness of education for diabeticpatients.). J Diabetes Nurs 2000;4:104–7.
Ginsberg BH, Tan MH, Mazze R, Bergelson A. Stageddiabetes management: computerizing a disease statemanagement program. J Med Syst 1998;22:77–87.
McCulloch DK, Mitchell RD, Ambler J, Tattersall RB. Aprospective comparison of ‘conventional’ and highcarbohydrate/high fibre/low fat diets in adults withestablished type 1 (insulin-dependent) diabetes.Diabetologia 1985;28:208–12.
Mühlhauser I, Bott U, Overmann H, Wagener W,Bender R, Jorgens V, et al. Liberalized diet in patientswith type 1 diabetes. J Intern Med 1995;237:591–7.
Mühlhauser I, Overmann H, Bender R, Jorgens V,Berger M. Predictors of mortality and end-stagediabetic complications in patients with type 1 diabetesmellitus on intensified insulin therapy. Diabet Med2000;17:727–34.
Perry TL, Mann JI, Lewis-Barned NJ, Duncan AW,Waldron MA, Thompson C. Lifestyle intervention inpeople with insulin-dependent diabetes mellitus(IDDM). Eur J Clin Nutr 1997;51:757–63.
Ryle A, Boa C, Fosbury J. Identifying the causes of poorself-management in insulin dependent diabetics: theuse of cognitive-analytic therapy techniques. 1993.
Rynne A, McKenna K. Evaluation of an outpatientdiabetes education programme. (Research evaluatinga four-session multidisciplinary outpatientprogramme. 29 refs). Br J Occup Ther 1999;62:459–65.
ter Braak EW, de Valk HW, de la Bije YF, van der LaakMF, van Haeften TW, Erkelens DW. Response totraining in blood glucose awareness is related toabsence of previous hypoglycaemic coma. DiabetesCare 2000;23:1199–200.
Watson MK, McDaniel JL, Gibson MH. An innovativeapproach to home health education: the critical pathto self-care for adults with diabetes. Home Health CareManag Pract 1996;8:41–51.
Trials excluded owing to nature ofpatients (patients not type 1 or 2and/or not adults)Agewall S, Wikstrand J, Samuelsson O, Persson B,
Andersson OK, Fagerberg B. The efficacy of multiplerisk factor intervention in treated hypertensive menduring long-term follow up. Risk Factor InterventionStudy Group. J Intern Med 1994;236:651–9.
Narayan KM, Hoskin M, Kozak D, Kriska AM, Hanson RL, Pettitt DJ, et al. Randomized clinical trialof lifestyle interventions in Pima Indians: a pilotstudy. Diabet Med 1998;15:66–72.
Turnin MC, Bourgeois O, Cathelineau G, Leguerrier AM,Halimi S, Sandre-Banon D, et al. Multicenterrandomized evaluation of a nutritional educationsoftware in obese patients. Diabetes Metab 2001;27(2 Pt 1):139–47.
Ward AK. Educational feedback in the management oftype 2 diabetes in general practice. Educ Gen Pract1996;7:142–50.
Trials excluded owing to nature ofeducation (i.e. not educationprogramme, no details education ornot reproducible)Abourizk NN, O’Connor PJ, Crabtree BF, Schnatz JD.
An outpatient model of integrated diabetes treatmentand education: functional, metabolic, and knowledgeoutcomes. Diabetes Educ 1994;20:416–21.
Abraira C, Colwell J, Nuttall F, Sawin CT, Henderson W,Comstock JP, et al. Cardiovascular events andcorrelates in the Veterans Affairs Diabetes FeasibilityTrial. Veterans Affairs Cooperative Study on GlycemicControl and Complications in type II Diabetes. ArchIntern Med 1997;157:181–8.
Albisser AM, Harris RI, Sakkal S, Parson ID, Chao SC.Diabetes intervention in the information age. MedInform 1996;21:297–316.
Basch CE, Walker EA, Howard CJ, Shamoon H, ZybertP. The effect of health education on the rate ofophthalmic examinations among African Americanswith diabetes mellitus. Am J Publ Health 1999;89:1878–82.
Benjamin EM, Schneider MS, Hinchey KT.Implementing practice guidelines for diabetes careusing problem-based learning. A prospective
Appendix 4
92
controlled trial using firm systems. Diabetes Care1999;22:1672–8.
Boehm S, Schlenk EA, Raleigh E, Ronis D. Behaviouralanalysis and behavioural strategies to improve self-management of type II diabetes. Clin Nurs Res 1993;2:327–44.
Brown SA, Harrist RB, Villagomez ET, Segura M,Barton SA, Hanis CL. Gender and treatmentdifferences in knowledge, health beliefs, andmetabolic control in Mexican Americans with type 2diabetes. Diabetes Educ 2000;26:425–38.
Carlson A, Rosenqvist U. Diabetes care organization,process, and patient outcomes: effects of a diabetescontrol program. Diabetes Educ 1991;17:42–8.
Clark C-MJ, Snyder JW, Meek RL, Stutz LM, ParkinCG. A systematic approach to risk stratification andintervention within a managed care environmentimproves diabetes outcomes and patient satisfaction.Diabetes Care 2001;24:1079–86.
Clarke P, Gray A, Adler A, Stevens R, Raikou M, Cull C,et al. Cost-effectiveness analysis of intensive blood-glucose control with metformin in overweightpatients with type II diabetes (UKPDS No. 51).Diabetologia 2001;44:298–304.
Close CF, Collins A, Gregory W, Hill C, Jarrett RJ, Jones SL, et al. Intensive therapy and progression toclinical albuminuria in patients with insulindependent diabetes mellitus and microalbuminuria.BMJ 1995;311:973–7.
Colwell JA. The feasibility of intensive insulinmanagement in non-insulin-dependent diabetesmellitus. Implications of the Veterans AffairsCooperative Study on Glycemic Control andComplications in NIDDM. Ann Intern Med 1996;124(1 Pt 2):131–5.
Daniel M, Green LW, Marion SA, Gamble D, HerbertCP, Hertzman C, et al. Effectiveness of community-directed diabetes prevention and control in a ruralAboriginal population in British Columbia, Canada.Social Scie Med 1999;4:815–32.
de Sonnaville JJ, Bouma M, Colly LP, Deville W, WijkelD, Heine RJ. Sustained good glycaemic control inNIDDM patients by implementation of structuredcare in general practice: 2-year follow-up study.Diabetologia 1997;40:1334–40.
Fasching P, Derfler K, Maca T, Kurzemann S, HoworkaK, Schneider B, et al. Feasibility and efficacy ofintensive insulin therapy in type 1 diabetes mellitusin primary care. Diabet Med 1994;11:836–42.
Fosbury JA, Bosley CM, Ryle A, Sonksen PH, Judd SL. Atrial of cognitive analytic therapy in poorly controlledtype I patients. Diabetes Care 1997;20:959–64.
Gaede P, Vedel P, Parving HH, Pedersen O. Intensifiedmultifactorial intervention in patients with type 2diabetes mellitus and microalbuminuria: the Stenotype 2 randomised study. Lancet 1999;353:617–22.
Groeneveld Y, Petri H, Hermans J, Springer M. Anassessment of structured care assistance in themanagement of patients with type 2 diabetes ingeneral practice. Scand J Prim Health Care 2001;19:25–30.
Heitzmann CA, Kaplan RM, Wilson DK, Sandler J. Sexdifferences in weight loss among adults with type IIdiabetes mellitus. J Behav Med 1987;10:197–211.
Hejlesen OK, Andreassen S, Frandsen NE, SorensenTB, Sando SH, Hovorka R, et al. Using a doubleblind controlled clinical trial to evaluate the functionof a Diabetes Advisory System: a feasible approach?Comput Methods Programs Biomed 1998;56:165–73.
Hiss RG, Gillard ML, Armbruster BA, McClure LA.Comprehensive evaluation of community-baseddiabetic patients: effect of feedback to patients andtheir physicians: a randomized controlled trial.Diabetes Care 2001;24:690–4.
Julius U, Gross P, Hanefeld M. Work absenteeism intype 2 diabetes mellitus: results of the prospectiveDiabetes Intervention Study. Diabete Metab 1993;19(1 Pt 2):202–6.
Korhonen T, Uusitupa M, Aro A, Kumpulainen T,Siitonen O, Voutilainen E, et al. Efficacy of dietaryinstructions in newly diagnosed non-insulin-dependentdiabetic patients. Comparison of two different patienteducation regimens. Acta Med Scand 1987;222:323–31.
Krier BP, Parker RD, Grayson D, Byrd G. Effect ofdiabetes education on glucose control. J La State MedSoc 1999;151:86–92.
Levin SR, Coburn JW, Abraira C, Henderson WG,Colwell JA, Emanuele NV, et al. Effect of intensiveglycemic control on microalbuminuria in type 2diabetes. Veterans Affairs Cooperative Study onGlycemic Control and Complications in type 2Diabetes Feasibility Trial Investigators. Diabetes Care2000;23:1478–85.
Manning RM, Jung RT, Leese GP, Newton RW. Thecomparison of four weight reduction strategies aimedat overweight diabetic patients. Diabet Med 1995;12:409–15.
Mazzuca SA, Moorman NH, Wheeler ML, Norton JA,Fineberg NS, Vinicor F, et al. The diabetes educationstudy – a controlled trial of the effects of diabetespatient education. Diabetes Care 1986;9:1–10.
Mengham LH, Morris BF, Palmer CR, White AJS. Isintensive dietetic intervention effective for overweightpatients with diabetes mellitus? A randomisedcontrolled study in a general practice. Pract DiabetesInt 1999;16:5–8.
Morgan BS, Littell DH. A closer look at teaching andcontingency contracting with type II diabetes. PatientEduc Couns 1988;12:145–58.
Muchmore DB, Springer J, Miller M. Self-monitoring ofblood glucose in overweight type 2 diabetic patients.Acta Diabetol 1994;31:215–9.
Health Technology Assessment 2003; Vol. 7: No. 22
93
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Ohkubo Y, Kishikawa H, Araki E, Miyata T, Isami S,Motoyoshi S, et al. Intensive insulin therapy preventsthe progression of diabetic microvascular complicationsin Japanese patients with non-insulin-dependentdiabetes mellitus: a randomized prospective 6-yearstudy. Diabetes Res Clin Pract 1995;28:103–17.
Olivarius NF, Beck-Nielsen H, Andreasen AH, HorderM, Pedersen PA. Randomised controlled trial ofstructured personal care of type 2 diabetes mellitus.BMJ 2001;323:970–5.
Pascale RW, Mullen M, Wing RR, Bononi P, Butler BA.Effects of a behavioural weight loss program stressingcalorie restriction versus calorie plus fat restriction inobese individuals with NIDDM or a family history ofdiabetes. Diabetes Care 1995;18:1241–8.
Peters ALD. Application of a diabetes managed careprogram. The feasibility of using nurses and acomputer system to provide effective care. DiabetesCare 1998;21:1037–43.
Piette JD, Weinberger M, McPhee SJ. The effect ofautomated calls with telephone nurse follow-up onpatient-centered outcomes of diabetes care: arandomized, controlled trial. Med Care 2000;38:218–30.
Piette JD, Weinberger M, McPhee SJ, Mah CA, KraemerFB, Crapo LM. Do automated calls with nurse follow-up improve self-care and glycemic control amongvulnerable patients with diabetes? Am J Med 2000;108:20–7.
Piette JD, Weinberger M, Kraemer FB, McPhee SJ.Impact of automated calls with nurse follow-up ondiabetes treatment outcomes in a Department ofVeterans Affairs Health Care System: a randomizedcontrolled trial. Diabetes Care 2001;24:202–8.
Shamoon H, Duffy H, Fleischer N, Engel S, Saenger P,Strelzyn M, et al. The effect of intensive treatment ofdiabetes on the development and progression oflong-term complications in insulin-dependentdiabetes mellitus. N Engl J Med 1993;329:977–86.
Shichiri M, Kishikawa H, Ohkubo Y, Wake N. Long-termresults of the Kumamoto Study on optimal diabetescontrol in type 2 diabetic patients. Diabetes Care2000;23 Suppl 2:B21–9.
Small M, Macrury S, Boal A, Paterson KR, MacCuishAC. Comparison of conventional twice dailysubcutaneous insulin administration and a multipleinjection regimen (using the NovoPen) in insulin-dependent diabetes mellitus. Diabetes Res 1988;8:85–9.
Smith DM, Weinberger M, Katz BP. A controlled trial toincrease office visits and reduce hospitalizations ofdiabetic patients. J Gen Intern Med 1987;2:232–8.
Spiess K, Sachs G, Pietschmann P, Prager R. A programto reduce onset distress in unselected type I diabeticpatients: effects on psychological variables andmetabolic control. Eur J Endocrinol 1995;132:580–6.
Tatti P, Lehmann ED. Preliminary results from arandomised controlled clinical trial for evaluating theteaching utility of an interactive educational diabetessimulator (AIDA). ADA scientific sessions 61st ScientificSessions of the American Diabetes Association,Philadelphia, Pennsylvania, USA, June 22–26,2001;Diabetes 50[Supplement 2], A25.2001.
The Diabetes Control and Complications Trial ResearchGroup. Nutrition interventions for intensive therapyin the Diabetes Control and Complications Trial. J AmDiet Assoc 1993;93:768–72.
The Diabetes Control and Complications Trial ResearchGroup. The effect of intensive treatment of diabeteson the development and progression of long-termcomplications in insulin-dependent diabetes mellitus.N Engl J Med 1993;329:977–86.
The Diabetes Control and Complications Trial ResearchGroup. Implementation of treatment protocols in theDiabetes Control and Complications Trial. DiabetesCare 1995;18:361–76.
The Diabetes Control and ComplicationsTrial/Epidemiology of Diabetes Interventions andComplications Research Group. Retinopathy andnephropathy in patients with type 1 diabetes fouryears after a trial of intensive insulin therapy. N EnglJ Med 2001;342:381–9.
UK Prospective Diabetes Study (UKPDS) Group. Tightblood pressure control and risk of macrovascular andmicrovascular complications in type 2 diabetes:UKPDS 38. BMJ 1998;317:713.
UK Prospective Diabetes Study (UKPDS) Group. Effectof intensive blood-glucose control with metformin oncomplications in overweight patients with type 2diabetes (UKPDS 34). Lancet 1998;352:854–65.
UK Prospective Diabetes Study (UKPDS) Group.Intensive blood-glucose control with sulphonylureasor insulin compared with conventional treatment andrisk of complications in patients with type 2 diabetes(UKPDS 33). Lancet 1998;352:837–53.
UK Prospective Diabetes Study (UKPDS) Group.Efficacy of atenolol and captopril in reducing risk ofmacrovascular and microvascular complications intype 2 diabetes: UKPDS 39. BMJ 1998;317:720.
Vinicor F, Cohen SJ, Mazzuca SA, Moorman N, WheelerM, Kuebler T et al. DIABEDS: a randomized trial ofthe effects of physician and/or patient education ondiabetes patient outcomes. J Chronic Dis 1987;40:345–56.
Wagner EH, Grothaus LC, Sandhu N, Galvin MS,McGregor M, Artz K, et al. Chronic care clinics fordiabetes in primary care – a system-wide randomizedtrial. Diabetes Care 2001;24:695–700.
Weinberger M, Kirkman S, Samsa GP, Shortliffe EA,Landsman PB, Cowper PA, et al. A nurse-coordinatedintervention for primary-care patients with non-insulin-dependent diabetes-mellitus: impact on
Appendix 4
94
glycemic control and health-related quality of life. J Gen Intern Med 1995;10:59–66.
Weinberger M, Kirkman MS, Samsa GP, Cowper PA,Shortliffe EA, Simel DL, et al. The relationshipbetween glycemic control and health-related quality-of-life in patients with non-insulin-dependentdiabetes-mellitus. Med Care 1994;32:1173–81.
Wing RR, Epstein LH, Paternostro-Bayles M, Kriska A,Nowalk MP, Gooding W. Exercise in a behaviouralweight control programme for obese patients withtype 2 (non-insulin-dependent) diabetes. Diabetologia1988;31:902–9.
Worth R, Home PD, Johnston DG, Anderson J,Ashworth L, Burrin JM, et al. Intensive attentionimproves glycaemic control in insulin-dependentdiabetes without further advantage from home bloodglucose monitoring: results of a controlled trial. BMJ1982;285:1233–40.
Trials excluded owing to length offollow-upDiabetes education program in Bulgaria. Patient Educ
Couns 2001;43:111–14.
Campbell LV, Barth R, Gosper JK, Jupp JJ, Simons LA,Chisholm DJ. Impact of intensive educationalapproach to dietary change in NIDDM. Diabetes Care1990;13:841–7.
Cox DJ, Gonder FL, Julian D, Cryer P, Lee JH,Richards FE, et al. Intensive versus standard bloodglucose awareness training (BGAT) with insulin-dependent diabetes: mechanisms and ancillaryeffects. Psychosom Med 1991;53:453–62.
de Weerdt I, Visser AP, Kok GJ, de Weerdt O, van derVeen EA. Randomized controlled multicentreevaluation of an education programme for insulin-treated diabetic patients: effects on metabolic control,quality of life, and costs of therapy. Diabet Med 1991;8:338–45.
Estey AL, Tan MH, Mann K. Follow-up intervention: itseffect on compliance behavior to a diabetes regimen.Diabetes Educ 1990;16:291–5.
Falkenberg MG, Elwing BE, Goransson AM, HellstrandBE, Riis UM. Problem oriented participatoryeducation in the guidance of adults with non-insulin-treated type-II diabetes mellitus. Scand J Prim HealthCare 1986;4:157–64.
Feddersen E, Lockwood DH. An inpatient diabeteseducator’s impact on length of hospital stay. DiabetesEduc 1994;20:125–8.
Genev NM, McGill M, Hoskins PL, Constantino MI,Plehwe W, Yue DK, et al. Continuing diabeteseducation by telephone. Diabet Med 1990;7:920–1.
Glasgow RE, Toobert DJ, Mitchell DL, Donnelly JE,Calder D. Nutrition education and social learninginterventions for type II diabetes. Diabetes Care1989;12:150–2.
Glasgow RE, Toobert DJ, Hampson SE, Noell JW. Abrief office-based intervention to facilitate diabetesdietary self-management. Health Educ Res 1995;10:467–78.
Glasgow RE, Toobert DJ, Hampson SE. Effects of a briefoffice-based intervention to facilitate diabetes dietaryself-management. Diabetes Care 1996;19:835–42.
Greenfield S, Kaplan SH, Ware J-EJ, Yano EM, FrankHJ. Patients’ participation in medical care: effects onblood sugar control and quality of life in diabetes. J Gen Intern Med 1988;3:448–57.
Hartwell SL, Kaplan RM, Wallace JP. Comparison ofbehavioural interventions for control of type IIdiabetes mellitus. Behav Ther 1986;17:447–61.
Hassell J, Medved E. Group/audiovisual instruction forpatients with diabetes. J Am Diet Assoc 1975;66:465–70.
Horwitz BL. Cooperative learning as an approach foreducating diabetic patients and their spouses. J N YState Nurses Assoc 1993;24:15–17.
Howorka K, Pumprla J, Wagner-Nosiska D, GrillmayrH, Schlusche C, Schabmann A. Empowering diabetesout-patients with structured education: short-termand long-term effects of functional insulin treatmenton perceived control over diabetes. J Psychosom Res2000;48:37–44.
Jaber LA, Halapy H, Fernet M, Tummalapalli S,Diwakaran H. Evaluation of a pharmaceutical caremodel on diabetes management. Ann Pharmacother1996;30:238–43.
Jayasuriya R, Griffiths R, Cheung J. Outcomeassessment of a community based model of generalpractitioner care diabetes patients. Pract Diabetes Int2000;17:179–82.
Kalergis M, Pacaud D, Strychar I, Meltzer S, Jones PJ,Yale JF. Optimizing insulin delivery: assessment ofthree strategies in intensive diabetes management.Diabetes Obes Metab 2000;2:299–305.
Kaplan RM, Wilson DK, Hartwell SL, Merino KL,Wallace JP. Prospective evaluation of HDL cholesterolchanges after diet and physical conditioningprograms for patients with type II diabetes mellitus.Diabetes Care 1985;8:343–8.
Ligtenberg PC, Godaert GL, Hillenaar EF, Hoekstra JB.Influence of a physical training program onpsychological well-being in elderly type 2 diabetespatients. Psychological well-being, physical training,and type 2 diabetes. Diabetes Care 1998;21:2196–7.
Mazzuca KB, Farris NA, Mendenhall J, Stoupa RA.Demonstrating the added value of community healthnursing for clients with insulin-dependent diabetes. J Commun Health Nurs 1997;14:211–24.
Miller CK, Edwards L, Kissling G, Sanville L. Nutritioneducation improves metabolic outcomes among olderadults with diabetes mellitus: results from arandomized controlled trial. Prev Med 2002;34:252–9.
Health Technology Assessment 2003; Vol. 7: No. 22
95
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Mulrow C, Bailey S, Sonksen PH, Slavin B. Evaluationof an Audiovisual Diabetes Education Program:negative results of a randomized trial of patients withnon-insulin-dependent diabetes mellitus. J Gen InternMed 1987;2:215–9.
O’Brien S, Hardy K. Impact of a care pathway-drivendiabetes education programme. (Research evaluatingthe impact of a programme at Whiston Hospital,Prescot. 18 refs). J Diabetes Nurs 2000;4:147–9.
Philis-Tsimikas A, Walker C. Improved care for diabetesin underserved populations. J Ambulatory Care Manage2001;24:39–43.
Pratt C, Wilson W, Leklem J, Kingsley L. Peer supportand nutrition education for older adults withdiabetes. J Nutr Elder 1987;6:31–43.
Rabkin SW, Boyko E, Wilson A, Streja DA. Arandomized clinical trial comparing behaviormodification and individual counseling in thenutritional therapy of non-insulin-dependent diabetesmellitus: comparison of the effect on blood sugar,body weight, and serum lipids. Diabetes Care 1983;6:50–6.
Smith L, Weinert C. Telecommunication support forrural women with diabetes. Diabetes Educ 2000;26:645–55.
Trento M, Passera P, Tomalino M, et al. Therapeuticgroup education in the follow-up of patients withnon-insulin treated, non-insulin dependent diabetesmellitus. Diabetes Nutr Metab Clin Exp 1998;11:212–16.
Turnin MC, Beddok RH, Clottes JP, Martini PF, AbadieRG, Buisson JC, et al. Telematic expert systemDiabeto. New tool for diet self-monitoring fordiabetic patients. Diabetes Care 1992;15:204–12.
Wdowik MJ, Kendall PA, Harris MA, Keim KS.Development and evaluation of an interventionprogram: “Control on Campus”. Diabetes Educ2000;26:95–104.
Werdier D, Jesdinsky HJ, Helmich P. A randomized,controlled study on the effect of diabetes counselingin the offices of 12 general practitioners. RevEpidemiol Sante Publique 1984;32:225–9.
Wilson W, Pratt C. The impact of diabetes educationand peer support upon weight and glycemic controlof elderly persons with noninsulin dependentdiabetes mellitus (NIDDM). Am J Public Health1987;77:634–5.
Trials excluded owing to outcomes (i.e.no report of diabetic control, QoL orend-points)Agewall S, Wikstrand J, Dahlof C, Fagerberg B. A
randomized study of quality of life during multiplerisk factor intervention in treated hypertensive menat high cardiovascular risk. J Hypertens 1995;13:1471–7.
Calle-Pascual AL, Rodriguez C, Camacho F, Sanchez R,Martin-Alvarez PJ, Yuste E, et al. Behaviourmodification in obese subjects with type 2 diabetesmellitus. Diabetes Res Clin Pract 1992;15:157–62.
Funnell MM, Arnold MS, Fogler J, Merritt JH,Anderson LA. Participation in a diabetes educationand care program: experience from the diabetes carefor older adults project. Diabetes Educ 1998;24:163–7.
Hanefeld M, Fischer S, Schmechel H, Rothe G, SchulzeJ, Dude H, et al. Diabetes Intervention Study. Multi-intervention trial in newly diagnosed NIDDM.Diabetes Care 1991;14:308–17.
Keyserling TC, Ammerman AS, Samuel-Hodge CD,Ingram AF, Skelly AH, Elasy TA, et al. A diabetesmanagement program for African American womenwith type 2 diabetes. Diabetes Educ 2000;26:796–805.
Korhonen T, Huttunen JK, Aro A, Hentinen M,Ihalainen O, Majander H, et al. A controlled trial onthe effects of patient education in the treatment ofinsulin-dependent diabetes. Diabetes Care 1983;6:256–61.
McNabb WL, Quinn MT, Rosing L. Weight loss programfor inner-city black women with non-insulin-dependent diabetes mellitus: PATHWAYS. J Am DietAssoc 1993;93:75–7.
Power L. Group approach to diabetes care. Apreliminary note. Postgrad Med 1983;73:211–16.
Racette SB, Weiss EP, Obert KA, Kohrt WM, Holloszy JO.Modest lifestyle intervention and glucose tolerance inobese African Americans. Obes Res 2001;9:348–55.
Rettig BA, Shrauger DG, Recker RR, Gallagher TF,Wiltse H. A randomized study of the effects of ahome diabetes education program. Diabetes Care1986;9:173–8.
Wheeler LA, Wheeler ML, Ours P, Swider C. Evaluationof computer-based diet education in persons withdiabetes mellitus and limited educationalbackground. Diabetes Care 1985;8:537–44.
Appendix 4
96
Quality criteria for RCTs – CRD Report 4Quality criteria for assessment of experimental studies
Some instructions for using a checklist for RCTs
Health Technology Assessment 2003; Vol. 7: No. 22
97
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Appendix 5
Quality assessment scales for RCTs and CCTs
1. Was the assignment to the treatment groups really random?
2. Was the treatment allocation concealed?
3. Were the groups similar at baseline in terms of prognostic factors?
4. Were the eligibility criteria specified?
5. Were outcome assessors blinded to the treatment allocation?
6. Was the care provider blinded? Not applicable
7. Was the patient blinded? Not applicable
8. Were the point estimates and measure of variability presented for the primary outcome measure?
9. Did the analyses include an intention to treat analysis?
10. Were withdrawals and dropouts completely described?
Quality item Coding Explanation
Random sequence generation AdequatePartialInadequateUnknown
Adequate: random numbers table or computer andcentral office or coded packagesPartial: (sealed) envelopes without furtherdescription or serially numbered opaque, sealedenvelopesInadequate: alternation, case record number, birthdate, or similar proceduresUnknown: just the term ‘randomised’ or ‘randomlyallocated’, etc.
1. Was the assignment to the treatment groups really random?
continued
Appendix 5
98
Quality item Coding Explanation
Concealment of randomisation
The person(s) who decide on eligibility should notbe able to know or be able to predict withreasonable accuracy to which treatment group apatient will be allocated. In trials that use goodplacebos this should normally be the case; however,different modes or timing of drug administration incombination with the use of small block sizes ofknown size may present opportunities for clinicianswho are also involved in the inclusion procedure tomake accurate guesses and selectively excludeeligible patients in the light of their most likelytreatment allocation; in centres with very lowinclusion frequencies combined with very brieffollow-up times this may also present a potentialproblem because the outcome of the previouspatient may serve as a predictor of the next likelyallocation
AdequateInadequateUnknown
Adequate: when a paper convinces you thatallocation cannot be predicted [separate persons,placebo really indistinguishable, clever use of blocksizes (large or variable)]. Adequate approachesmight include centralised or pharmacy-controlledrandomisation, serially numbered identicalcontainers, on-site computer-based system with arandomisation sequence that is not readable untilallocation and other approaches with robustmethods to prevent foreknowledge of the allocationsequence to clinicians and patients.
Inadequate: this option is often difficult. You have tovisualise the procedure and think how people mightbe able to circumvent it. Inadequate approachesmight include use of alternation, case recordnumbers, birth dates or week days, open randomnumbers lists, serially numbered envelopes (evensealed opaque envelopes can be subject tomanipulation) and any other measures that cannotprevent foreknowledge of group allocation.
Unknown: no details in text. Disagreements or lackof clarity should be discussed in the review team
2. Was the treatment allocation concealed?
Baseline characteristics
Main aim is to enable the reviewer to see whichpatients were actually recruited. It enables one toget a rough idea on prognostic comparability. A realcheck on comparability requires multivariablestratification (seldom shown)
ReportedUnknown
Consult the list of prognostic factors or baselinecharacteristics (not included in this appendix).Reviewer decides
3. Were the groups similar at baseline regarding the prognostic factors?
Prestratification
Consult the list of prognostic factors or baselinecharacteristics (not included in this appendix)
AdequatePartialInadequateUnknown
Single-centre studyAdequate: prestratification on at least one factorfrom the list or no prestratification if the number ofpatients exceeds a prespecified number
Partial: leave judgement to reviewer
Inadequate: stratification on a factor(s) not on ourlist or no stratification whereas the number ofpatients is less than the prespecified number
Unknown: no details in text and no way to deducethe procedure from the tables
Multicentre studyAdequate: must prestratify on centre. Within eachcentre the criteria for single centre studies alsoapply
Partial: impossible option
Inadequate: no prestratification on centre orviolating the criteria for single centre studies (seeabove)
Unknown: no details in text and no way to deducethe procedure from the tables
4. Were the eligibility criteria specified?
continued
Health Technology Assessment 2003; Vol. 7: No. 22
99
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Quality item Coding Explanation
Blinding of assessors
The assessor may be the patient (self-report), theclinician (clinical scale, blood pressure, etc.) or,ideally, a third person or a panel. Very important injudgement of cause of death but unimportant injudgement of death
AdequateInadequateUnknown
Adequate: independent person or panel or (self-)assessments in watertight double-blind conditions
Inadequate: clinician is assessor in trial on drugs withclear side effects or a different influence on lab.results, ECGs, etc.
Unknown: no statements on procedures and notdeducible
5. Were outcome assessors blinded to the treatment allocation?
continued
Blinding of patientsThis item is hard to define. Just the statement‘double blind’ in the paper is really insufficient if theprocedure to accomplish this is not described orreasonably deducible by the reviewer. Goodplacebos (see, hear, taste, feel, smell), trickyunmasking side-effects accounting for thesubjectivity of the outcome measurements and theaccessibility of co-interventions by the patient arerequired
AdequatePartialInadequateUnknown
Adequate: placebo described as ‘indistinguishable’and procedures watertight
Partial: just ‘double blind’ in text and no furtherdescription of procedures or nature of the placebo
Inadequate: wrong placebo
Unknown: no details in text
7. Was the patient blinded?
ComplianceDosing errors and timing errors
AdequatePartialInadequateUnknown
Unknown
Adequate: Medication Event Monitoring System(MEMS or eDEM)
Partial: blood samples, urine samples (use ofindicator substances)
Inadequate: pill count or self-report
Unknown: not mentioned
Check on blindingQuestionnaire for patients, care givers, assessorsand analysis of the results; the (early) timing iscritical because the treatment effect may be thecause of unblinding, in which case it may be used asan outcome measure
ReportedUnknown
Reviewer decides
Blinding of care givers
Look out for good placebos (see, hear, taste, feel,smell), tricky unmasking side-effects accounting forthe subjectivity of the outcome measurements andthe accessibility of co-interventions by thecaregivers
AdequatePartialInadequateUnknown
Adequate: placebo described as ‘indistinguishable’and procedures watertight (use your imaginationwith the ‘cheat’ in mind; e.g. statement thatsensitive/unmasking lab. results were kept separatefrom ward personnel)
Partial: just ‘double blind’ in text and no furtherdescription of procedures or nature of the placebo
Inadequate: wrong placebo (e.g. fructose in trial onascorbic acid)
Unknown: no details in text
6. Was the care provider blinded?
Register when they may have an impact on any ofthe outcome phenomena. Consult the list of co-interventions (not included in this appendix)
AdequatePartialInadequateUnknown
Adequate: percentages of all relevant interventionsin all groups
Partial: one or more interventions omitted oromission of percentages in each group
Inadequate: not deducible
Unknown: no statements
Co-interventions
Quality criteria for assessment of CCTs – CRD Report 4
Appendix 5
100
Quality item Coding Explanation
Results for the primary outcome measure AdequatePartialInadequateUnknown
Adequate: mean outcome in each group togetherwith mean difference and its standard error (SE) orstandard deviation (SD) or any CI around it or thepossibility to calculate those from the paper. Survivalcurve with log-rank test and patient numbers atlater time points
Partial: partially reported
Inadequate: no SE or SD, or SD without N (SE = SD/N)
Unknown: very unlikely
8. Were the point estimates and measure of variability presented for the primary outcome measure?
Intention-to-treat (ITT) analysis
Early drop-out can make this very difficult. Strictestrequirement is sensitivity analysis including earlydrop-outs
AdequateInadequate
Reviewers should not just look for the term ITT butassure themselves that the calculations wereaccording to the ITT principle.
9. Did the analysis include an intention to treat analysis?
Dealing with missing values
The percentage missing values on potentialconfounders and outcome measurements (seldomgiven) is a rough estimate of a trial’s quality. Onecan carry them forward, perform sensitivity analysisassuming the worst and best case scenarios, usestatistical imputation techniques, etc. Note that thedefault option (deletion) assumes that the value israndomly missing, which seems seldom justified
AdequatePartialInadequateUnknown
Adequate: percentage of missing values anddistribution over the groups and procedure ofhandling this stated
Partial: some statement on numbers or percentages
Inadequate: wrong procedure (a matter of greatdebate)
Unknown: no mentioning at all of missing and notdeducible from tables
Loss to follow-up
This item examines both numbers and reasons;typically an item that needs checking in the methodssection and the marginal totals in the tables. Notethat it may differ for different outcome phenomenaor time points. Some reasons may be reasons givenby the patient when asked and may not be the truereason. There is no satisfactory solution for this
AdequatePartialInadequateUnknown
Adequate: number randomised must be stated.Number(s) lost to follow-up (dropped out) statedor deducible (from tables) for each group andreasons summarised for each group.
Partial: numbers, but not the reasons (or vice versa)
Inadequate: numbers randomised not stated or notspecified for each group
Unknown: no details in text
Were the groups similar at baseline in terms of prognostic factors?
Were the eligibility criteria specified?
Were outcome assessors blinded to the treatment allocation?
Were the point estimates and measure of variability presented for the primary outcome measure?
Did the analyses include an intention to treat analysis?
Were withdrawals and dropouts completely described?
Were participants likely to be representative of the intended population?
Psychometric instrumentsMany different measures of psychologicalconstructs were used to quantify knowledge,attitudes, QoL and other psychological variables.Only the results using instruments known to be orreported to be validated were data extracted. A fewstudies used measures that were constructed for thepurposes of the study about which no validationinformation was provided. Unfortunately, thefailure to use validated instruments or to validatetheir own instrument means that these resultscannot be clearly interpreted. The use ofunvalidated psychometric instruments represents alost opportunity to collect valuable information.
QoLThe diabetes quality of life measure (DQOL) wasused by Trento and colleagues.30 The measure wasoriginally designed for use in the DCCT. Theoriginal intent was to evaluate the burden of anintensive diabetes treatment regimen. However, itwas also designed for broader application indiabetes as the scale items cover a range of issuesrelevant to diabetes and its treatment. Theinstrument addresses satisfaction with treatment,impact of treatment, worry about the future effectsof diabetes and worry about social/vocational issuesas well as an overall well-being scale. The itemsare answered on a five-point scale. Test–re-testreliability ranges from 0.78 to 0.92. The test hasalso been shown to have good internal consistencyin patients with either Type 1 or Type 2 diabetes.
QoL was tested by Kaplan and colleagues35 usinga previously validated scale used in chronicobstructive pulmonary disease. The indexconceptualises health as two components: currentstate of health and prognosis. The measure hasthree scales: mobility, physical activity and socialactivity. Patients are also classified as having any of36 symptoms or problems that might inhibitfunction. Levels of well-being are the socialpreferences that society associated with observablelevels of functioning.
QoL was measured by Gilden and colleagues47
with questions focused on self-care skills. The self-
care skills included diet, exercise, medicationadministration, monitoring blood tests and threegeneral items. QoL was subdivided into twosubscales (QLa and QLb). QLa indicated moredemanding and intensive lifestyle changes due todiet, exercise and other general factors. QLbreflected less demanding behaviours includingmedication compliance and self-testing. It seemsthat the knowledge, QoL, stress and familyinvolvement scales used in this study may havebeen tested for internal consistency togetheryielding a Cronbach’s α of 0.93.
Other measures of psychologicalstatusGilden and colleagues47 assessed stress using nineitems adapted from another validated scale. Thenine items were answered on a three-point scalewith a higher score indicating less stress.
Depression was assessed in the Gilden study47
using Zung’s Mood Scale. The scale consists of 20items. The total index and four subscales werescored: pervasive affective disturbance,physiological disturbance, psychomotordisturbance and psychological disturbance. Scoresrange from 25 to 100 with lower scores reflectingless depression.
KnowledgeThe knowledge questionnaire used in theMühlhauser study23 was a 37-item illustratedquestionnaire. It included general aspects ofdiabetes, metabolic self-monitoring, rules forchanging insulin dose, treatment and preventionof hypoglycaemia and diet. The internal reliabilityof the questionnaire was 0.8. A Russian version ofthe same questionnaire was used in the Starostinastudy.24
The Diabetes Knowledge Scale – form A (DKNA)97
is a 15-item scale with Cronbach’s α > 0.82. Thescale was used by Campbell and colleagues.29 Themultiple-choice questions include questions on thenormal range for BG, the causes ofhypoglycaemia, insulin requirements during illness
Health Technology Assessment 2003; Vol. 7: No. 22
101
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Appendix 6
Psychological instruments used in included trials
and the status of rice as a carbohydrate food.Additional items test basic survival informationand other valid content.
Knowledge of diabetes was tested by Trento andcolleagues30 using the GISED. This questionnairewas developed by the Education Study Group ofthe Italian Society for Diabetes. The 38-itemquestionnaire was slightly modified to clarify themeaning of some terms. The internal consistencywas found to be acceptable and internal validitywas checked by cluster analysis.
Kronsbein and colleagues34 used a knowledgequestionnaire that was designed for the trial(DTTP–NIDDM). The questionnaire consisted of21 multiple-choice items. Additional informationwas not evaluated as it was in a Germanpublication.
Gilden and colleagues47 measured knowledgeusing a 24-item questionnaire including generalknowledge, nutrition and pharmacy.
Bloomgarden and colleagues44 assessed knowledgein a standardised manner with an interviewer.Eight questions were used and were assumed to bevalidated as a Centers for Disease Controlpublication on diabetes knowledge measures wascited as their source. The knowledge score wassimply the sum of correct answers.
Measures of adoption ofeducational recommendationsand satisfactionAttitudes to diabetes and its treatment wereassessed by Cooper and colleagues (see footnote toTable 12) using the Diabetes IntegrationQuestionnaire. The questionnaire measures theintegration of diabetes and its treatment into thelifestyle and personality of the patient. It is a 19-item scale. Higher summary scores are related tobetter psychological adjustment to diabetes. Thequestionnaire is reported to be reliable and valid.
Treatment effectiveness was assessed by Cooperand colleagues (see footnote to Table 12) using aquestionnaire derived from an interview tool.Patients respond to seven items (two on treatmenteffectiveness in relation to self-care, three onseriousness, and two on personal control) on afive-point scale. Patients were also asked aboutself-care treatment effectiveness for 11 areas (e.g.
physical activity, not smoking, glucose testing). Foreach of these areas patients were asked the degreeto which that area was believed important incontrolling diabetes and the degree to which thatarea will prevent future complications. An overalltreatment effectiveness score was created byaveraging scores across all the treatmenteffectiveness questions. This questionnaire wasreported to be reliable and valid.
Satisfaction was assessed by Campbell andcolleagues29 using an 18-item scale developed andvalidated by the authors. It was shown to havegood internal consistency and reliability.
Health behaviours were evaluated by Trento andcolleagues30 using the Condotte di Riferimento(CdR). The questionnaire consisted of 16 itemsthat posed hypothetical situations of the form‘what would you do if …’. The test evaluatedwhether patients were able to identify underlyinghealth problems and react correctly. Thequestionnaire was checked for internal consistencyusing Cronbach’s α and internal validity waschecked by cluster analysis.
Glasgow and colleagues45 assessed patient dietusing the Kristal Food Habits Questionnaire(FHQ). The FHQ is a 20-item scale that measuresfour dimensions of fat related dietary habits. Asummary score across the four dimensions wasused in the analyses. The FHQ has been validated.
Other validated instruments usedA number of additional instruments were used invarious studies. These instruments are not beingdescribed, because the studies in which they wereused did not report the results of these measuresat a 12-month or later evaluation.
The SF-36 was used to measure QoL in theSamaras trial.40 An apparent variation of this scalewas also used by Ridgeway.37
The Beck Depression Inventory was used byWing.39 Although this is a valid psychometricinstrument, the use of the instrument has beenquestioned in patients who are not depressed.
Ridgeway37 used the Life Skills cognitiveknowledge of diabetes test provided by theDiabetes Education Society and approved by theAmerican Diabetes Association.
Appendix 6
102
Health Technology Assessment 2003; Vol. 7: No. 22
103
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Appendix 7
Data extraction: Type 1 diabetes
Reference and design Intervention Participants Outcome measures
Surname and year: Reichardet al., 1988–9621,98–107
Source: published
Country: Sweden
Setting: outpatient clinic
Language: English
Trial design: RCT
Called the StockholmDiabetes Intervention Study(SDIS)
Treatment intervention:intensified conventionaltreatment (ICT) withstructured education.Patients attended singly or inpairs
Topics: intermediatemetabolism, especially therole of insulin, insulinsubstitution, effect of insulinsubstitution by varying foodintake and exercise,hypoglycaemia and counterregulation, microvascularcomplications, performingand interpreting BG tests andprinciples of insulinsubstitution in relation to testresults. Recommendedmultiple insulin injections andfrequent home BGmonitoring. Goals for homeBG levels individually set. Goalwas to reduce HBA1c to 7%
Tutoring: initial tutoringperformed with telephonecontacts at least every 2weeks. Patients suggestedsolutions to problems butphysician intervened ifdangerous. Patients used dailyglucose tests and wrotedown results. Initially phonecontacts every 2 weeks ormore often if needed. Ifpatients did not call, physiciancalled them. As they grewmore confident, called every3–4 weeks. Continuoustutoring on demand startedwhen metabolic control wasoptimal. Patients could reachphysician at any time of dayvia pager
Provider: physician
Eligibility exclusion criteria:born 1930 or later (in 1982);IDDM appearing at age ≤ 30,and with insulin dependencywithin 1 year from diagnosis;no known abuse of alcohol ordrugs; non-proliferativeretinopathy of any degreepresent (includingpreproliferative retinopathy),no previous photocoagulation;normal serum creatinine;unsatisfactory BG controlaccording to physician incharge of patient
How selected: 111 patientsasked to participate, 102accepted (they did notbeforehand have to acceptthe intensified programme ifrandomised to such atreatment)
Numbers involved: total N =102: Intervention N = 48;control = 54.
Nos on insulin: all
Tablets:
Diet alone:
Type of diabetes: Type 1
Duration of diabetes in years:mean (SD):Intervention=17.9 (6.4);control =16.3 (4.9)
Baseline measurements ofoutcome parameters: seeresults as recalculated overtime following drop-outs
Gender: Intervention: male N = 26, female N = 22;control: male N = 28, femaleN = 26
Primary outcomes used:HbA1c, hypoglycaemicepisodes, ketoacidoticincidents, diabetic retinopathy,neuropathy, nephropathy
Secondary outcomes used:mortality, hospital admissions,BP, well-being, BMI, footulcers, time and number ofpatient visits, risk factors forcomplications, dietary intake,cognitive function andneuropsychological function
Individual preferred learningstyle addressed?: no
Any sub-groups (e.g. ethnicgroups): no
Normal range(s) foroutcomes: HbA1c = 4–6%;mild retinopathy = level 2.2or less; UAER rates:normoalbuminuria: <20µg/min; microalbuminuria:20–200 µg/min
Nephropathy: >200 µg/min
Nerve conduction velocities= lower normal method was41 m/s
How outcomes assessed:
HbA1c: by lab. measurement(altered over trial period buthigh correlation betweenmethods)
Retinopathy: grading systemas used in ETDRS (EarlyTreatment DiabeticRetinopathy Study), mean of2 ophthalmologists grading
Nephropathy: UAER,analysed in 24-h urinesamples
Neurophysiologicalassessment: conduction
continued
Appendix 7
104
Reference and design Intervention Participants Outcome measures
Length and no. of sessions: 2education sessions, 3 and 2 hlong, respectively. Seen in theclinic every second month.Had frequent phone contactwith the physician –reachable at any time of dayvia a pager. After 7.5 yearsICT patients returned toroutine diabetes care
Mode:
Treatment changes: yes
Training of trainers:
Theory:
Control intervention:advised to monitor their BG.Visited the clinic every fourthmonth. Given instructions onhow to use home BG testingand insulin doses wereadjusted to achieve lower BG.Test results discussed at clinicvisits. Treatment goal was toreduce BG without giving riseto serious hypoglycaemia.Many patients had frequentcontact with physician after7.5-year period.
Protocol changes to bothgroups: protocol changedtwice: after 3 years, in orderto achieve lower BG levels inthe control group and after7.5 years, in order to letintensively treated patientsreturn to routine diabetes care
Duration of intervention: 7.5years
Age, mean (SD): Intervention= 30.0 (7.5); control = 31.7(7.3) ethnic groups: notreported
Losses to follow-up: At 3years, 97 patients remainedIntervention N = 44, controlN = 53. At 5 years 96remained Intervention N =44 and control N = 52, At7.5 years 89 remained, At 10years 43 remained.
Compliance: no dataavailable
velocities determined in theperoneal, tibial and suralnerves
Hypoglycaemia: patientreports. Serioushypoglycaemic episodesdefined as requiring helpfrom someone else orresulting in a coma
Risk factors for microvascularcomplications: patients withHbA1c during study ≥ 9%were compared with patientswith levels below this
Neuropsychological andcognitive tests: a battery ofcomputerised tests from theAutomated PsychologicalTest (APT) system, notreported here
Well-being: not a validatedmeasure, not reported here
Dietary intake: analysed by adietitian. A non-judgemental48-h recall used with patientunprepared
Validated: yes
Timing of outcomes same forboth groups: assume yes
Length of follow-up: for 7.5years during the educationprogramme, then returnedto normal care – followed upfor 2.5 more years
Results: Values given for outcomes are the mean of all the values measured at approximately 4-monthintervals over the specified time period. Mean (SEM) given, unless stated otherwise.
HbA1c (%):
Intervention group Control group
Time Baseline Follow-up Baseline Follow-up Difference between groups
18 months 9.5 (0.2) 7.5 (from graph) 9.4 (0.2) 9.0 (from graph) p = 0.00053 years 9.5 (0.2) 7.4 (0.1) 9.4 (0.2) 9.0 (0.2) p = 0.000015 years 9.5 (0.2) 7.2 (0.1) 9.4 (0.2) 8.7 (0.1) p < 0.0017.5 years 9.5 (1.3) 7.1 (0.7) 9.4 (1.4) 8.5 (0.7) p = 0.00110 years 9.5 (1.4) 7.2 (0.6) 9.4 (1.2) 8.3 (1.0) p < 0.001
After 3 years: the number of patients with mean HbA1c levels above the initial mean of 9.5% was reduced from 20 to 0 inICT group and from 27 to 10 in RT group.
Health Technology Assessment 2003; Vol. 7: No. 22
105
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Retinopathy:Number of patients demonstrating mild retinopathy at 18 months
Changes in mean retinopathy level: number of patients at 18 months
Sum of patients with preproliferative or proliferative changes in at least one eye (level 5, <5 or worse)
Percentage of patients demonstrating serious retinopathy
Mean retinopathy level (12 grade scale 0.5–6.0)
Visual acuity (percentage of patients)
Visual deterioration (percentage of patients)
Intervention group Control group
Time Baseline Follow-up Baseline Follow-up Difference between groups
18 months 27 28 26 18 p = 0.011
Intervention group Control group
Time Baseline Follow-up Baseline Follow-up Difference between groups
18 months 3 7 4 15
Intervention group Control group
Time Baseline Follow-up Baseline Follow-up Difference between groups
7.5 years NA 27 NA 52 p = 0.0110 years 0 33 0 63 p = 0.003
Intervention group Control group
Time Baseline Follow-up Baseline Follow-up Difference between groups
18 months 2.4 (0.1) 2.8 (0.2) 2.6 (0.1) 3.2 (0.2)3 years 2.4 (0.1) 3.2 (0.2) 2.6 (0.1) 3.6 (0.2) NS5 years 2.4 (0.1) 3.5 (0.2) 2.6 (0.1) 4.1 (0.2) p < 0.05
After 5 years: proliferative retinopathy appeared in at least one eye in 10 ICT patients and 15 RT patients (NS).
Intervention group Control group Difference between groups
Better 6 5Unchanged 26 19Worse 16 30 0.024
Time Intervention group Control group Difference between groups
7.5 years 14 35 p = 0.02
Time Intervention group Control group Difference between groups
10 years 18 37 p = 0.04
Appendix 7
106
Normoalbuminuria: number of patients
Microalbuminuria: number of patients
Nephropathy: number of patients
Nephropathy: percentage of patients
Mean UAER levels (µµg/minute)
GFR: glomerular filtration rate (ml/minute)
Intervention group Control group
Time Baseline Follow-up Baseline Follow-up Difference between groups
18 months 34 35 36 273 years 34 35 35 307.5 years 34 33 33 26
Intervention group Control group
Time Baseline Follow-up Baseline Follow-up Difference between groups
18 months 11 9 13 193 years 8 6 13 137.5 years 8 8 13 11
Intervention group Control group
Time Baseline Follow-up Baseline Follow-up Difference between groups
18 months 3 4 3 63 years 2 3 3 87.5 years 2 3 3 12 p = 0.01
Intervention group Control group
Time Baseline Follow-up Baseline Follow-up Difference between groups
3 years 122 (3) 115 (3) 126 (3) 119 (3)5 years 122 (3) 112 (3) 126 (3) 115 (4)7.5 years 122 (19) 109 (19) 126 (21) 110 (27) NS10 years 123 (19) 110 (18) 127 (22) 109 (25) NS
Intervention group Control group
Time Baseline Follow-up Baseline Follow-up Difference between groups
3 years (diff. 1.3 (0.1) 1.3 (0.1) 1.4 (0.1) 1.6 (0.1) p = 0.031from below)5 years 55.7 (26.7) 46.0 (26.1) 74.3 (31.0) 239.9 (129.7) p < 0.057.5 years 56 (175) 45 (110) 63 (206) 119 (219) p = 0.04
Intervention group Control group
Time Baseline Follow-up Baseline Follow-up Difference between groups
10 years 5 7 7 26 p = 0.012
Health Technology Assessment 2003; Vol. 7: No. 22
107
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Neuropathy: number (percentage) of patients who exhibited neuropathy
Neurophysiology: Nerve Conduction Velocities: peroneal nerve
Tibial nerve
Sural nerve
Hypoglycaemia:Percentage of patients experiencing at least one serious hypoglycaemic episode in the time period
Intervention group Control group
Time Baseline Follow-up Baseline Follow-up Difference between groups
5 years 13 16 17 34 p < 0.017.5 years 5 (12%) 6 (14%) 8 (17%) 13 (28%) NS10 years 12% 14% 16% 32% p = 0.041
Intervention group Control group
Time Baseline Follow-up Baseline Follow-up Difference between groups
18 months 42.5 (0.7) 42.3 (0.6) 42.1 (0.7) 40.5 (0.7)3 years 43.0 43.4 42.1 40.8 NS5 years 43.0 (0.7) 42.8 (0.6) 42.1 (0.7) 39.3 (0.7) p < 0.017.5 years 43.2 43.0 42.0 38.5(from graph)10 years 42.9 (4.4) 41.3 (3.8) 41.9 (4.7) 36.2 (11.6) p = 0.007
Intervention group Control group
Time Baseline Follow-up Baseline Follow-up Difference between groups
18 months 41.2 (0.7) 41.6 (0.6) 40.2 (0.7) 39.1 (0.8)3 years 41.3 42.7 40.4 40.5 NS5 years 41.3 (0.8) 42.1 (0.6) 40.4 (0.7) 37.7 (0.8) p < 0.0017.5 years 41.3 42.5 40.4 37.8(from graph)10 years 41.3 (5.4) 41.1 (4.2) 40.4 (5.0) 35.1 (11.8) p = 0.002
Intervention group Control group
Time Baseline Follow-up Baseline Follow-up Difference between groups
18 months 45.1 (0.7) 44.1 (0.8) 45.3 (0.8) 43.1 (0.8)3 years 44.3 44.0 42.8 37.9 NS5 years 44.2 (1.3) 40.3 (1.8) 42.6 (1.7) 36.5 (2.0) p < 0.057.5 years 44.5 42.5 43.0 34.2(from graph)10 years 44.2 (8.6) 39.7 (12.0) 42.5 (12.3) 30.8 (18.4) p = 0.008
Time Intervention group Control group Difference between groups
18 months 48 22 p = 0.0033 years 57 23 p = 0.0015 years 77 56 p < 0.057.5 years 80 58 p < 0.0510 years 86 73 NS
Appendix 7
108
Total number of serious hypoglycaemic episodes in the time period
Number of patients requiring emergency room visits
Mean total number of serious hypoglycaemic episodes per patient per year
Ketoacidosis: number of patients experiencing an episode
Blood pressure: systolic (mmHg)
Blood pressure: diastolic (mmHg)
Number of patients receiving treatment for hypertension
Time Intervention group Control group Difference between groups
18 months 41 283 years 102 285 years 242 98
Time Intervention group Control group Difference between groups
5 years 1.1 0.47.5 years 1.1 0.410 years 1.06 0.47 p = 0.003
After 5 years: patients unconscious at least once: ICT= 41% (18) RT= 19%, (10) p < 0.05
Time Intervention group Control group Difference between groups
7.5 years 1 210 years 1 4
Time Intervention group Control group Difference between groups
5 years 7 117.5 years 11 17
Time Intervention group Control group Difference between groups
18 months 8 83 years 11 3During last 2.5 years 8 8 NS(from 7.5 to 10 years)
Intervention group Control group
Time Baseline Follow-up Baseline Follow-up Difference between groups
3 years 129.2 (2.0) 127.0 (2.3) 133.2 (2.0) 131.8 (2.1)5 years 129 (2) 126 (2) 133 (2) 133 (2)10 years 129.3 (13.5) 124.9 (15.4) 133.2 (15.8) 132.2 (15.7) p = 0.029
Intervention group Control group
Time Baseline Follow-up Baseline Follow-up Difference between groups
3 years 77.5 (1.4) 78.0 (1.2) 78.5 (1.0) 81.2 (1.2)5 years 77 (1) 77 (1) 79 (1) 78 (1)10 years 79.4 (9.4) 74.1 (8.6) 78.4 (8.4) 77.3 (8.7) p = 0.085
Health Technology Assessment 2003; Vol. 7: No. 22
109
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Body mass index
Energy intake (kcal/day)
Mortality:Number of patients who had died
Time and number of patient visitsAfter18 months: patients in the ICT group required a mean of 45 minutes per patient per month foreducation, visits and telephone contacts, compared with 10 minutes per patient per month for patients inthe RT group. Between 3 and 5 years after the start of the study there were no longer any differencesbetween the groups
Neuropathic foot ulcersAfter 7.5 years: number of patients who developed neuropathic foot ulcers: ICT = 0, RT = 3
Risk factors for complicationsAfter 3 years: patients with HbA1c ≥ 9% (the mean value for RT group) were compared with those withlower values. There was significantly more deterioration in the former (retinopathy p = 0.028;nephropathy, p = 0.025, neuropathy, p = 0.018)22 ICT patients (50%, 95% CI 34 to 66%) and 27 RT patients (73%, 95% CI 61 to 84%) deteriorated withrespect to one complication or more (p = 0.024)
Methodological commentsAllocation to treatment groups: partialBlinding of outcome assessors: all investigators (ophthalmologist, neurophysiologist, laboratorypersonnel) except the physician in charge of the study were unaware of the treatment group of theindividual patientsAllocation concealment: randomisation performed with closed identical envelopesAnalysis by ITT: no
Intervention group Control group
Time Baseline Follow-up Baseline Follow-up Difference between groups
18 months 22.6 (0.3) 22.9 (0.3) 22.8 (0.3) 22.9 (0.3)3 years 22.6 (0.3) 23.4 (0.4) 22.8 (0.3) 23.0 (0.3)5 years 22.5 (0.3) 23.8 (0.4) 22.8 (0.3) 22.8 (0.3)7.5 years 22.5 (0.3) 23.9 (0.5) 22.8 (0.4) 23.3 (0.4) NS10 years 22.5 (2.0) 24.2 (3.4) 22.8 (2.5) 23.9 (2.9) NS
Intervention group Control group
Time Baseline Follow-up Baseline Follow-up Difference between groups
3 years 1812 (82) 1768 (99) 1829 (77) 1758 (63) NS
Time Intervention group Control group Difference between groups
3 years 4 05 years 4 17.5 years 4 310 years 4 3
After 3 years: 4 patients in intervention group had died. After 5 years: one control patient had died. After 7.5 years: 4 patients in intervention group and 3 in control group had died. After 10 years: 4 patients in intervention group and 3 in control group had died.
Appendix 7
110
Comparability of treatment groups: yesMethod of data analysis: hypothesis tests (t-tests, Wilcoxon tests and Mann–Whitney U-tests). Contingencytables analysed by chi-squared test. Linear regression used when appropriate. For multivariate analysesused logistic regression. Some results expressed as means with 95% CI, majority mean and SEM.Sample size/power calculation: noAttrition/drop-out: numbers and reasons given
General commentsGeneralisability: inclusion criteria defined. Do not know what proportion of eligible patients inpopulation participatedConflict of interests: Swedish Division of Novo-Nordisk, Boehringer Mannheim Scandinavica, SwedishMedical Research Council, Groschinsky FoundationOther: values given are mean values over the whole study period at 1.5, 3, 5, 7.5 and 10 years. States thatafter 3 years an effort was made to reduce Hba1c below 9% in all control patients, ? how
Quality criteria (CRD Report 4) RCTs
1. Was the assignment to the treatment groups really random? Partial2. Was the treatment allocation concealed? Inadequate3. Were the groups similar at baseline in terms of prognostic factors? Reported4. Were the eligibility criteria specified? Yes5. Were outcome assessors blinded to the treatment allocation? Adequate6. Were the point estimates and measure of variability presented for the primary outcome measure? Partial7. Did the analyses include an ITT analysis? Inadequate8. Were withdrawals and drop-outs completely described? Adequate
Health Technology Assessment 2003; Vol. 7: No. 22
111
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Reference and design Intervention Participants Outcome measures
Surname and year: Terent et al., 198522
Source: published
Country: Sweden
Setting: community
Language: English
Trial design: RCT
4 Groups:A = Education + SMBGB = SMBGC = EducationD = Control
Treatment intervention:education (for Groups A &C) individual
Provider: physician anddietitian
Topics: special modelconstructed to explaininterplay between foodconsumption, BG levels,insulin and urinary glucoseexcretion. Also taught abouthypoglycaemic episodes,hyperglycaemic episodes,foot care, injections andurine testing
Sessions: six hourly lessonsduring 1 month
Treatment changes: self-monitoring, see below
Training of trainers:
Theory:
Mode: Given questionnaire at1 and 6 months after end ofthe course to test knowledgeof diabetes and related issues
SMBG: Groups A & B hadextra visit at outpatient deptat start of phase II. SMBGdemonstrated by physician.SMBG groups “encouragedto change their insulin doseto achieve preprandial values<7 mmol/litre andpostprandial values <10 mmol/litre”
Control intervention:Standard therapy. Groups B(phase I) and D (phases I–III)continued their pre-trialchecking habits. Fasting BGand 24-h urinary glucosevalues measured every 3rdmonth at outpatients.Physical exam. 6 monthly. Allpatients had device formonitoring urinary glucose
Duration of intervention:1 month
Eligibility. All adult patients(aged ≥ 17) with Type 1adiabetes in municipality,diagnosed ≤ 20 years.
How selected: from surveyof diabetes in area. N = 37,first randomised into 2groups: formal educationgroup N = 19, standardtherapy N = 18. After 6months of education (phaseI) a second randomisationperformed. Teaching ofSMBG completed in 6months (phase II) andpatients followed for further6 months (phase III)
Numbers involved: N = 37in 4 groupsGroup A N = 10 education+ SMBGGroup B N = 8 usual care +SMBGGroup C N = 9 education +educationGroup D N = 10 usual care+ usual care
Nos on insulin: all
Type of diabetes: type I
Duration of diabetes (years)(mean ± SD):Group A 11.6 ± 6.2, GroupB 13.0 ± 3.8, Group C 5.0± 3.9, Group D 12.5 ± 5.1
Baseline measurements ofoutcome parameter: HbA1(mean ± SD):Group A 12.3 ± 3.2, GroupB 11.8 ± 1.4, Group C 11.2± 2.0, Group D 11.1 ± 2.3
Gender (M/F):Group A 6/4, Group B 3/5,Group C 4/5, Group D 8/2
Ages (mean ± SD):Group A 28.5 ± 6.2, GroupB 27.6 ± 6.8, Group C 25.7± 5.4, Group D 25.0 ± 4.6
Ethnic groups: not given
Losses to follow-up: none
Compliance: all attendededucation sessions. Thenumber of urinary glucosetesters in education groups A
Primary outcomes used:HbA1, hypoglycaemicepisodes, ketoacidoticincidents
Secondary outcomes used:diabetes knowledge
Individual preferred learningstyle addressed: no
Any sub-groups: no
Normal range(s) foroutcomes: 95% CI for HbA14.7 to 8.0%
How outcomes assessed:HbA1 by lab. (columnchromatography), knowledgeby questionnaire,hypoglycaemic episodes bymedical record
Validated: yes for HbA1, nofor knowledge
Timing of outcomes same forboth groups:
Length of follow-up: 18months from inception
continued
Appendix 7
112
Knowledge about diabetes, insulin, oral hypoglycaemics, testing and physical exercise: not validatedmeasure. Knowledge about food exchange and good distribution over the daytime: not validated measure
Methodological commentsAllocation to treatment groups: not statedBlinding of outcome assessors: yes (HbA1 values not accessible to investigators or patients until end of study)Allocation concealment: not statedAnalysis by ITT: no drop-outsComparability of treatment groups: duration of diabetes significantly shorter in Group CMethod of data analysis: within-group comparisons, no analysis between groupsSample size/power calculation: noAttrition/drop-out: none
General commentsGeneralisability: good – only 4 eligible patients in the community excluded – reasons givenConflict of interests: funding support not mentionedOther: very small number of patients in each group
Quality criteria (CRD Report 4) RCTs
1. Was the assignment to the treatment groups really random? Unknown2. Was the treatment allocation concealed? Unknown3. Were the groups similar at baseline in terms of prognostic factors? Reported4. Were the eligibility criteria specified? Yes5. Were outcome assessors blinded to the treatment allocation? Adequate6. Were the point estimates and measure of variability presented for the primary outcome measure? Partial7. Did the analyses include an ITT analysis? No drop-outs8. Were withdrawals and drop-outs completely described? No drop-outs
Reference and design Intervention Participants Outcome measures
Outcome Group A Group B (SMBG) Group C (education) Group D (control)(education + SMBG)
& C increased from 9 (47%)to 15 (79%). For SMBG ingroups A & B, proportion ofweekly testers was 89% inphase II and 78% in phase III.Adherence to SMBG equallygood in Groups A and B. ForSMBG patients, averagenumber of visits toOutpatient dept was 6 inphase II and 5 in phase III
HbA1 levels (mean ±SD): no significancetesting between groupsonly within
12 months = 11.0 ±2.618 months = 10.2 ±1.9
12 months =10.8 ±1.018 months = 9.8 ±3.0
12 months =9.9 ± 2.518 months = 10.2 ±2.1
12 months = 9.5 ±3.218 months = 10.4 ±2.1
Hypoglycaemic episodes:(no statistical analysis)
7 in groups A + B 14 in groups C + D
Ketoacidosis: (nostatistical analysis)
2 3
Health Technology Assessment 2003; Vol. 7: No. 22
113
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Reference and design Intervention Participants Outcome measures
Surname and year:Mühlhauser et al., 198723
Source: published
Country: Romania
Setting: hospital-based
Language: English
Trial design: prospectivecontrolled trial (3 groups)CCT
Treatment interventions:
IDTTP (intensive treatmentand teaching programme):Düsseldorf model
Provider: 2 nurses trained inDüsseldorf
Topics: BG as normal aspossible; metabolic self-monitoring (blood or urine);self-adaptation of insulindose; recording self-monitoring, doses andhypoglycaemic episodes;liberalised diet
Sessions: 5 days in groups ofabout 10 patients
Treatment changes: IDTTPused different insulins as wellas different therapy
Training trainers:
Theory:
Mode:
BDTTP (basic treatment andteaching programme):adaptation of IDTTP
Provider: 2 teaching nurses
Topics: aglucosuria withoutsignificant hypoglycaemicreactions; simple rules forself-adjustment of insulin;matching diet to insulinpreparation used
Sessions: 4 days
Treatment changes:
Training trainers:
Theory:
Mode:
Control intervention:standard treatment ofhospital (no self-adjustmentor self-monitoring, rigid diet,individual diseasemanagement instruction byphysician in charge)
Duration of intervention:Initially 1 year for all groups.For second year the controlentered IDTTP and IDTTPfollowed for 2 years. BDTTPonly 1 year
Eligibility: ketosis-prone,insulin-dependent diabeticpatients, aged 15–40 years.Excluded if: admissionprimarily for severe acute orchronic disorder unrelated todiabetes, mental retardationor psychiatric diseases thatwould interfere withparticipation in groupteaching programme,clinically overt diabeticnephropathy, proliferativeretinopathy or blindness,severe foot complications
How selected: consecutiveadmissions to hospital fordiabetic metabolicdecompensation or newlydiagnosed diabetes
Numbers involved: 300(IDTTP 100, BDTTP 100,Control 100)
Type of diabetes: 1
Duration diabetes (medianyears): IDTTP 6, BDTTP 5,control 5
Baseline measurements ofoutcome parameter (mean± SEM):
HbA1: IDTTP 12.3 ± 0.2,BDTTP 11.7 ± 0.2, Control12.5 ± 0.2
BMI (kg/m2): IDTTP 21.8 ±0.2, BDTTP: 21.5 ± 0.2,control: 21.7 ± 0.3
Knowledge: IDTTP 16 ± 1,BDTTP: 17 ± 1, control: 16± 1
No. hospitalised in yearbefore study: IDTTP 5,BDTTP 46, control 53
Gender (M/F, %): IDTTP:57/43, BDTTP 54/46, control60/40
Age (years) (mean ± SEM):26 ± 1
Ethnic groups: not reported
Losses to follow-up (year 1):IDTTP 2%, BDTTP 8%,control 7%
Primary outcomes used:HbA1, hypoglycaemicepisodes, ketoacidoticincidents
Secondary outcomes used:knowledge, hospitaladmissions, BMI, insulin dose(U/kg weight), insulininjections, frequency of self-monitoring
Individual preferred learningstyle addressed?: no
Any sub-groups (e.g. ethnicgroups): no
Normal range(s) foroutcomes: HbA1c (mean ±2SD in 50 healthy subjects):5.4–7.6%
How outcomes assessed:HbA1c: lab. Hypoglycaemicand ketoacidotic episodes:interview and record review
Knowledge: questionnaire
Hospital admissions: baseline= self-report; record review
Validated: no info. on validity,adequate reliability
Timing of outcomes same forboth groups: yes
Length of follow-up: 1 yearfrom inception reportedhere. IDTTP followed for 2years
continued
Appendix 7
114
Methodological comments6-month HbA1c data reportedAllocation to treatment groups: consecutive patients to each group. Reports that the order of conditionswas chosen randomly, but patient groups not recruited concurrently rather consecutively. This may haveresulted in more ill patients entering control group because they were recruited first and might be morelikely to be hospitalisedBlinding of outcome assessors: not reportedAllocation concealment: no
Outcome IDTTP (n = 98 BDTTP (n = 92 Control (n = 93 Difference (1 year) unless stated unless stated) unless stated) between groups
otherwise) otherwise) otherwise)
HbA1 (estimated fromgraph; mean)
9.3%**b 11.2% 12.8% **sig. to control p < 0.01bsig. to BDDT, p <0.01
Severe hypoglycaemia(total no. of patients withat least one episode)
12 5 6 (n = 97)
(total number ofepisodes)
27 5 9 ( n= 97)
Ketoacidosis(no of patients with atleast one episode)
2** 3* 13 (n = 97) *sig. to control, p < 0.05**sig. to control p < 0.01
(total no. of episodes) 2** 4* 16 (n = 97) *sig. to control, p < 0.05**sig. to control p < 0.01
Knowledge (mean ±SEM)
32 ± 1**a 26 ± 1** 24 ± 1 **sig. to control p < 0.01asig. to BDTTP p < 0.05
Hospitalisations(no. of patientshospitalised)
42**a 57** 84 **sig. to control p < 0.01asig. to BDTTP p < 0.01
(total no. of hospitaladmissions and days)
67**a;630 days**a
100**;967 days**
173;1447 days
**sig. to control p < 0.01asig. to BDTTP p < 0.01
Number of daily insulindoses (1/2/>3)
0/44/56**b 9/76/15 19/71/9 **sig. to control p < 0.01bsig. to BDDT, p < 0.01
Daily insulin dose (U/kgweight) (mean ± SEM)
0.70 ± 0.02 (n = 85) 0.67 ± 0.02 (n = 83) 0.65 ± 0.03 (n = 80)
BMI (mean ± SEM) 23.3 ± 0.3*a 22.6 ± 0.2 22.4 ± 0.3 *sig to control, p < 0.05asig to BDTTP p < 0.05
Frequency of self-monitoring (data notpresented)
Health Technology Assessment 2003; Vol. 7: No. 22
115
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Analysis by ITT: deaths in control group accounted for in hypoglycaemia and ketoacidosis analyses,otherwise not reportedComparability of treatment groups: BDTTP group significantly lower in HbA1 at baselineMethod of data analysis: hypothesis tests, confidence intervals not provided. Procedure adopted to adjustfor baseline differences in HbA1c
Sample size/power calculation: noneAttrition/drop-out: numbers and reasons for drop-outs reported
General commentsGeneralisability: patient population seems appropriateConflict of interests: partial support from Boehringer Mannheim, Novo-Industri and Becton-DickinsonOther: none
Quality criteria (CRD Report 4) CCTs
Were the groups similar at baseline in terms of prognostic factors? ReportedWere the eligibility criteria specified? YesWere outcome assessors blinded to the treatment allocation? UnknownWere the point estimates and measure of variability presented for the primary outcome measure? PartiallyDid the analyses include an ITT analysis? UnknownWere withdrawals and drop-outs completely described? PartialWere participants likely to be representative of the intended population? Yes
Appendix 7
116
Reference and design Intervention Participants Outcome measures
Surname and year: Starostina et al., 199424
Source: published
Country: Russia
Setting: National ResearchCentre for Endocrinology(Moscow)
Language: English
Trial design: CCT(prospective controlled trial)
Treatment intervention:DTTP
Topics: (based on Düsseldorfmethod) Two programmesused – one based on BGSMand one on UGSM. Patientsadvised to monitor blood orurine glucose 3–4 times dailybefore main meals and atbedtime. If insulin treatmentis intensified, patients canliberalise their diet. As moreliberalised, more frequentself-monitoring and injectionsof insulin and adaptation ofdosage
Provider: DTTP performedby 2 physicians
Sessions: 5-day inpatienttreatment and teachingprogramme
Delivery:
Treatment changes: patientsadapt insulin dosagethemselves
Training of trainers:
Theory:
Control intervention: usualcare: no structurededucation, no metabolic self-monitoring, no rules for self-adjustment of insulindosages, but withconventional strict dietaryprescriptions
Duration of intervention: 5days
Eligibility: 121 consecutivetype I diabetic patients, aged15–45, admitted to theNational Research Centre forEndocrinology for inpatienttreatment. Excluded if:significant loss of vision, renalinsufficiency, severeconcomitant ordersunrelated to diabetes
How selected: following agroup randomisationprotocol, first consecutive 61to UGSM and next 60 toBGSM. Additional 60 patientsfulfilling the inclusion criteriarecruited to control group
Numbers involved: N = 181,N = 61 UGSM (urineglucose self-monitoring), N = 60 BGSM (BG self-monitoring), N = 60 control
Nos on insulin: all
Type of diabetes: type I
Duration of diabetes (years± SE):UGSM 11 ± 0.9, BGSM 10.9± 0.8, control 10.9 ± 0.9
Baseline measurements ofoutcomes:
HbA1 (mean ± SE): UGSM12.5 ± 0.2, BGSM 12.6 ±0.2, control 12.2 ± 0.2
Severe hypoglycaemia:UGSM 2, BGSM 6, control 6Ketoacidosis: UGSM 9,BGSM 10, control 17
BMI: UGSM 23.6, ± 0.5,BGSM:= 22.4 ± 0.3, control22.3 ± 0.3
Knowledge (mean ± SE):UGSM 11 ± 0.1, BGSM 11± 0.1, control 11 ± 1
Hospitalisation (diabetesrelated year up tointervention) (meandays/patient ± SE): UGSM9.8 ± 2.6, BGSM 9.0 ± 3.4,control 11.6 ± 2.6
Sick leave (diabetes related)(mean ± SE): UGSM: 7.8 ±
Primary outcomes used:HbA1
Secondary outcomes used:costs, hypoglycaemia,ketoacidosis, diabetes-relatedhospitalisation days, diabetes-related sick leave days,knowledge
Individual preferred learningstyle addressed: no
Any sub-groups (e.g. ethnicgroups): no
Normal range(s) foroutcomes; HbA1 5–8%
How outcomes assessed:HbA1b lab. test, knowledgeby questionnaire, others notstated
Validated: knowledge testRussian version of astandardised questionnaire.Unclear if re-validated
Timing of outcomes same forboth groups: no – longer forintervention than control
Length of follow–up:intervention groups 24 months, control group 12 months
continued
Health Technology Assessment 2003; Vol. 7: No. 22
117
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Reference and design Intervention Participants Outcome measures
Outcome (mean ± UGSM (n = 55) BGSM (n = 52) Control (n = 58) Difference SE unless noted between groupsa
otherwise)
a No comparisons between intervention and control groups reported.
3.2, BGSM 11.1 ± 4.2,control 10.6 ± 2.3
No daily insulin injections:UGSM 1.9 ± 0.1, BGSM 2.3± 0.1, control: 2.2 ± 0.1
Daily insulin dose: UGSM0.67 ± 0.03, BGSM 0.73 ±0.04, control 0.68 ± 0.03
Gender (M/F): UGSM 31/30,BGSM 29/31, control 26/34
Age ranges (years ± SE):UGSM 28.7± 1.1, BGSM29.1 ± 1.1, control 29 ± 1.2
Ethnic groups: not given
Losses to follow-up: 16 (9%)(6 from UGSM, 8 fromBGSM and 2 control)(reasons given)
Compliance: not mentioned
HbA1 1 year: 9.4 ± 0.22 year: 9.2 ± .0.2
1 year: 9.3 ± 0.22 year: 9.2 ± 0.2
1 year: 12.3 ± 0.2
Hypoglycaemia (cases) 1 year: 22 year: 8
1 year: 62 year: 4
1 year: 8
Ketoacidosis (cases) 1 year: 12 year: 0
1 year: 02 year: 0
1 year: 16
BMI 1 year: 24.4 ± 0.52 year: 24.4 ± 0.5
1 year: 23.3 ± 0.32 year: 23.2 ± 0.3
1 year: 22.6 ± 0.3
Knowledge 1 year: 25 ± 12 year: 25 ± 1
1 year: 26± 12 year: 26 ± 1
1 year 11 ± 1 Increase comparable inUGSM and BGSM
Hospitalisationdays/patient (diabetesrelated)
1 year: 0.8 ± 0.62 year: 1.1 ± 0.7
1 year: 0.4 ± 0.42 year: 1.7 ± 0.8
1 year: 14.3 ± 3.6 Decrease comparablein UGSM and BGSM
Sick leave/patient(diabetes related)
1 year: 0.2 ± 0.22 year: 1.0 ± 0.7
1 year: 02 year: 0.7 ± 0.5
1 year: 10.7 ± 2.0 Decrease comparablein UGSM and BGSM
No. of daily insulininjections
1 year: 2.9 ± 0.12 year: 2.9 ± 0.1
1 year 2.9 ± 0.12 year: 3.2 ± 0.1
1 year 2.2 ± 0.1 Increase comparable inUGSM and BGSM
Daily insulin dose(IU/kg)
1 year: 0.75 ± 0.032 year: 0.70 ± 0.03
1 year: 0.74 ± 0.032 year: 0.69 ± 0.02
1 year: 0.70 ± 0.03
Appendix 7
118
Methodological commentsAllocation to treatment groups: reported as group randomisation for UGSM and BGSM. Control groupunclearBlinding of outcome assessors: not statedAllocation concealment: noAnalysis by ITT: noComparability of treatment groups: yesMethod of data analysis: data expressed as means and ±SEM. Comparisons with parametric and non-parametric tests for unpaired data, analysis of variance (ANOVA) for repeated measure, other hypothesistesting methodsSample size/power calculation: noAttrition/drop-out: 9%Participants may not have been comparable to usual care in the UK – high initial HbA1 levels
General commentsWhen UGSM used the savings from discontinuing ineffective drugs outweighed the costs of test strips andproduced net savings. When BGSM was used, net costs were incurredGeneralisability: yes – consecutive patients admitted to Research Centre. Consecutive assignment mayresult in differences due to history, etc., but all recruited within 5 monthsConflict of interests: financial support from Boehringer Mannheim, GermanyOther: not sure how control group were recruited – insufficient detail given
Quality criteria (CRD Report 4) CCTs
Were the groups similar at baseline in terms of prognostic factors? ReportedWere the eligibility criteria specified? YesWere outcome assessors blinded to the treatment allocation? UnknownWere the point estimates and measure of variability presented for the primary outcome measure? PartialDid the analyses include an ITT analysis? UnknownWere withdrawals and drop-outs completely described? AdequateWere participants likely to be representative of the intended population? No
Health Technology Assessment 2003; Vol. 7: No. 22
119
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Interventions of multifaceted self-management education
Appendix 8
Data extraction: Type 2 diabetes
Reference and design Intervention Participants Outcome measures
Surname and year: Brown et al., 200228,108
Source: published
Country: USA
Setting: community
Language: English
Trial design: RCT
Treatment intervention:Culturally referenceddiabetes self-managementgroup education interventionusing didactic and interactiveapproach, delivered inperson. 4 cohorts over 1 year
Topics: nutrition, self-monitoring, exercise,hygiene, illness days, footcare, complications (shortand long term). Promotionbehaviour changes throughproblem solving, foodpreparation demonstrationsand social support
Provider: Mexican Americannurses, dietitians andcommunity workers
Sessions: 52 contact hours (3months of weekly 2-hsessions, 6 months ofbiweekly + 3 months ofmonthly 2-h support groupsessions)
Theory: based on results offour meta-analytic reviewsand 6 years of developmentand piloting of intervention.
Delivery: groups with eachparticipant bringing a‘support’ person
Treatment changes:
Training of trainers: 4 nursesand 4 dietitians attendedseminars on diabeteseducation and participated insupervised clinical practicumwith outpatients. 8community workers withType 2 diabetes participatedin an 8-week programme ondiabetes self-management
Mode: written materialslimited owing to low literacyrates. Languagepredominantly Spanish with a
Eligibility criteria: Type 2diabetes (defined p. 260)diagnosed after 35 years ofage, aged between 35 and 70years, willing to participate.Excluded if pregnant or if hadmedical conditions for whichdiet and exercise changeswould be contraindicated
How selected: randomlyselected from rosters ofprevious research studies(none intervention studies, all blood sampling). Groupedby area of county in whichthey lived
Numbers involved: 256 (128intervention, 128 control)
Nos on insulin: intervention25, control 26
Tablets: intervention 83,control 86
Diet alone: intervention 10,control 7
Oral and insulin: intervention8, control 7
Type of diabetes?: 2
Mean duration of diabetes(years): intervention 7.6 (SD5.8), control 8.1 (SD 6.9)
Baseline measurements ofoutcome parameter (mean± SD):
HbA1c intervention 11.810% ± 3%, control11.80% ± 3.02%
BMI: intervention 32.33 ±5.97, control 32.12 ± 6.35
Cholesterol: intervention211.83 ± 45.34, control203.57 ± 48.82
Primary outcomes used:HbA1c
Secondary outcomes used:diabetes-related knowledge,fasting BG, BP, totalcholesterol, HDL and LDLcholesterol, triglycerides,health beliefs, home glucosemonitoring, BMI, costs
Individual preferred learningstyle addressed?: no
Any sub-groups: age andgender
Normal range(s) foroutcomes: none reported
How outcomes assessed?: nodetails reported
Validated?: physiologicalmeasures yes, knowledgeand health beliefs unclear
Timing of outcomes same forboth groups: yes
Length of follow-up: 12months from inception
continued
Appendix 8
120
Knowledge/beliefs not reported as not a validated measure. 3- and 6-months data reportedCosts: total for eight subjects/group = US$3070; total per person = US$384
Methodological commentsAllocation to treatment groups: reports that individuals allocated to groups and then later that groupswere randomly assigned to experimental or control conditions. In ‘data analysis’ section also states thatrandom assignment but no method describedBlinding of outcome assessors?: not reportedAllocation concealment?: not reportedAnalysis by ITT?: see method of data analysisComparability of treatment groups: reported to be no significant differences only any baseline variables
Reference and design Intervention Participants Outcome measures
Outcome (mean ± SD) Intervention group Control group Difference between groups
HbA1c (n = 112) 10.89% (2.56), 11.64% (2.85), *p < 0.05adjusted 10.87%* adjusted 11.66%
FBG (n = intervention 114, 194.95 (63.27)* 210.51 (66.55) *p < 0.05control 113)
Cholesterol (n = intervention 189.88 (36.35) 187.64 (42.66)112, control 113)
Triglycerides (n = 113) 214.43 (194.93) 198.65 (148.38)
BMI (n = intervention 113, 32.17 (6.45) 32.28 (6.52)control 114)
blend of English and eachparticipant nominated afamily member as a supportperson. Ref. 16 in trial givesmore detail of interventionplus Table 1, p. 261
Control intervention:Usual care by physicians orlocal clinics (wait-listcontrols)
Duration of intervention:12 months
Triglycerides: intervention215.35 ± 130.07, control195.58 ± 118.95
Gender (M/F): intervention51/75, control 40/86
Mean age (years):intervention 54.7 (SD 8.2),control 53.3 (SD 8.3)
Ethnic groups: all MexicanAmericans
Losses to follow-up: notreported. Baseline data on126 intervention and 126control patients, 12-monthsdata based on 112intervention and 112 controlpatients
Compliance: attendance at1st session was 79%. At endof 12 months it was 50%.Dropped to 40% at 13weeks when focus changedfrom education to supportgroup sessions
Health Technology Assessment 2003; Vol. 7: No. 22
121
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Method of data analysis: multilevel modelling (within-subjects and between-subjects analysis) whichestimates for a given subject from available data and thus does not eliminate those with missing dataStandard deviation reported, no confidence intervalsSample size/power calculation: not reportedAttrition/drop-out: not reported except numbers in results tables
General commentsGeneralisability: high HbA1c at baseline, culturally referenced to Mexican Americans, different cohortsover timeConflict of interests: funded by National Institute for Diabetes and Digestive and Kidney Diseases and theOffice of Research on Minority HealthOther:
Quality criteria (CRD Report 4) RCTs
1. Was the assignment to the treatment groups really random? Unknown2. Was the treatment allocation concealed? Unknown3. Were the groups similar at baseline in terms of prognostic factors? Reported4. Were the eligibility criteria specified? yes5. Were outcome assessors blinded to the treatment allocation? Unknown6. Were the point estimates and measure of variability presented for the primary outcome measure? Adequate7. Did the analyses include an ITT analysis? Adequate8. Were withdrawals and drop-outs completely described? Partial
Appendix 8
122
Reference and design Intervention Participants Outcome measures
Surname and year: Campbellet al., 199629
Source: published
Country: Australia
Setting: unclear
Language: English
Trial design: RCT
4 programmes: minimalinstruction (1), individualeducation (2), groupeducation (3), behaviouralprogramme (4). Allencouraged to bring asupport person
Provider: programmes 1, 2+3 were delivered by staff inthe diabetes educationservice, including 5 nurseeducators and 3 dietitians. Asingle nurse deliveredprogramme 4
Treatment intervention 1(apparently individual) =minimal education:
Sessions: two 1 h sessionswithin 2 weeks of referral
Topics: (same topics but lessdetail than others); theportion exchange dietarysystem, exercise, use ofOHAs, practical instruction inurine testing, foot care andrecommendations to consultan ophthalmologist andpodiatrist
Treatment intervention 2= individual education:
Sessions: 2 sessions for 1 hwithin 2 weeks of referral,then 30-minute sessionsapproximately monthly until12 months
Topics: same but more detailthan for intervention 1 andincluded information on thecauses, symptoms,mechanisms andcomplications of diabetes
Treatment intervention 3= group education:
Sessions: at least 2 individualsessions and a 3-day smallgroup education course.(Individual monthly sessionswere continued until acourse could be scheduled)
Mode: course involvedlectures, small groupexercises, practical sessions
Topics: same topics as theother programs. 2-h follow-
Eligibility exclusion criteria:<80 years, Type 2 for <5years, speak and writeEnglish, had received noprevious formal instruction,not taking >75% of themaximum dose OHAs, hadno terminal illness
How selected: patientsreferred by generalpractitioner.
Numbers involved: total N =238, groups 1 59, 2 57, 3 66,4 56
Nos on insulin: none
Tablets: group 1 19, 2 22, 324, 4 23.
Diet alone: groups 1 40, 235, 3 42, 4 33.
Type of diabetes?: 2
Duration of diabetes (meanyears ± SE): group 1 0.5(0.1), 2 0.9 (0.2), 3 0.4 (0.1),4 0.36 (0.1)
Baseline measurements ofoutcome parameter:
HbA1: groups 1 11.9% (SE0.6), 2 12.2% (0.5), 3 12.1%(0.6), 4 13.3% (0.6)
Knowledge: group 1 5.7 (0.4),2 5.3 (0.4), 3 5.5 (0.4), 4 4.6(0.5)
Systolic BP: groups 1 136.9(2.4), 2 135.5 (3.0), 3 137.5(2.7), 4 145.8 (3.3)
Diastolic: group 1 80.7 (1.3),2 81.6 (1.2), 3 81.7 (1.4), 491.7 (1.7)
Gender (M/F): groups 1 22/37,2 33/24, 3 35/31, 4 24/32
Mean age (years): groups 158.2 (1.3), 2 56.8 (1.5), 358.4 (1.4), 4 60.9 (1.4)
Ethnic groups: not reported
Losses to follow-up: group 240% attrition, group 3 42%,group 4 9%
Compliance:
Primary outcomes used:HbA1
Secondary outcomes used:BP, knowledge, satisfaction,uptake podiatry,ophthalmology,hospitalisations, BMI
Individual preferred learningstyle addressed?: no
Any sub-groups: no
Normal range(s) foroutcomes: HbA1 <8.5%,knowledge?
How outcomes assessed?:HbA1 lab., knowledge,satisfaction, hospitalisationsself-report, BP unclear
Validated?: HbA1, knowledge(DKNA) yes, satisfactionreported to have showngood internal consistency andreliability
Timing of outcomes same forboth groups:
Length of follow-up: 12months (minimal instructiononly 6 months) frominception
continued
Health Technology Assessment 2003; Vol. 7: No. 22
123
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Reference and design Intervention Participants Outcome measures
Outcomes (mean Group 1 Group 2 Group 3 Group 4 Difference change ± SE (minimal (individual (group (behavioural) between groupsunless noted education) education) education)otherwise)
ups were scheduled at 3 and9 months
Treatment intervention 4= behavioural:
Sessions: series of individualvisits, 3 in first month, afterwhich differed depending onpatient’s needs with aminimal schedule of 3, 6 and13 months supplementedwith phone calls
Topics: same topics as othergroups
Mode: Sessions in patient’shome
All groups:
Treatment changes: no details
Training of trainers: no details
Theory: no details except forgroup 4: based on cognitive-behavioural strategies
Participants in groups 2 and 3also had opportunity to attenda 2-h lecture on diet (group)
Duration of intervention: upto 12 months
continued
HbA1 (%): n = ?/25/19/39
No follow-up –3.3% (0.9) –3.0%(1.1) –4.8%(0.7)
Knowledge:n = ?/29/26/35
No follow-up 4.4 (0.6) 4.2(0.5) 5.6(0.6)
Systolic BP(mgHg): n = ?/16/11/37
No follow-up –6.8(5.8) –12.4(6.8) –16.9(3.8)
Diastolic BP(mgHg):n = ?/16/11/37
No follow-up –5.3(3.0)* –5.0(4.0)* –7.9(2.6) *sig. from group 4,p < 0.05
BMI n = ?/30/25/41
No follow-up –2.0 (0.4) –1.4 (0.5) –2.6 (0.5)
Cholesterol(mmol/l)n = ?/23/19/34
No follow–up 0.12 (0.20) 0.16 (0.16) –0.33(0.15)
HDL cholesterol(mmol/l)n = ?/21/16/27
No follow-up 0.02 (0.04) 0.18 (0.10) 0.06 (0.08)
Appendix 8
124
Methodological commentsAllocation to treatment groups: not describedBlinding of outcome assessors?: not describedAllocation concealment?: not describedAnalysis by ITT?: noComparability of treatment groups: significant differences in levels of education, duration since diagnosis,diastolic BP, smokingMethod of data analysis; continuous data – change scores were calculated and compared by ANCOVA(covariance analysis) with t-tests as post hoc tests, categorical data – chi-squared and pair-wise comparisons,mean and standard error givenSample size/power calculation: noAttrition/drop-out: percentages reported
General commentsGeneralisability: 94% patients asked to participate consented, high HbA1c at baselineConflict of interests: funding support not mentionedOther:
Quality criteria (CRD Report 4)
Outcomes (mean Group 1 Group 2 Group 3 Group 4 Difference change ± SE (minimal (individual (group (behavioural) between groupsunless noted education) education) education)otherwise)
3- and 6-month data reported.
Cholesterol riskratio (total/HDL)n = ?/21/15/25
No follow-up –0.25 (0.03) –0.35 (0.46) –0.59 (0.20)
Treatmentintensity:n = ?/29/27/42
No follow-up % unchanged: 75% decreased: 17% increased: 7
% unchanged: 70% decreased: 22% increased: 8
% unchanged: 74% decreased: 17% increased: 10
Satisfaction (actualscore + SE): n = ?/25/25/30
No follow-up 74.8(2.2) 77.9(2.0) 77.0(2.3)
Proportionconsultingophthalmology(%): n =?/38/37/47
No follow-up 97 95 89
Proportionconsulting podiatry(%): n =?/31/30/42
No follow-up 55 73 74
1. Was the assignment to the treatment groups really random? Unknown2. Was the treatment allocation concealed? Unknown3. Were the groups similar at baseline in terms of prognostic factors? Reported4. Were the eligibility criteria specified? Yes5. Were outcome assessors blinded to the treatment allocation? Unknown6. Were the point estimates and measure of variability presented for the primary outcome measure? Partially7. Did the analyses include an ITT analysis? Unknown8. Were withdrawals and drop-outs completely described? Reported
Health Technology Assessment 2003; Vol. 7: No. 22
125
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Reference and design Intervention Participants Outcome measures
Surname and year:Trento et al., 200130
Source: published
Country: Italy
Setting: university clinic
Language: English
Trial design: RCT
Treatment intervention:
Topics: observation phase,educational diagnosis,definition of goals anddevelopment of planincluding methods andsetting in which to deliver.Data collected on patientsbaseline education, healthbeliefs. undesirability of beingoverweight, meal planning,improving and checkingmetabolic control andpreventing complications(more detail in Table 1).Homework diaries forweight and food intake weregiven out at the end of eachmeeting, and discussed atbeginning of next
Provider: 1 or 2 physiciansand educationalist. Also GP, 2postgrad. medical students,clinical psychologist andpsychometrist helped designprogramme
Sessions: 4 sessions, over 1 heach. Sessions apparentlyrepeated every 3 monthsPatients in need/wishing tohave clinical attention wereseen on a one-to-one basis atthe end. 4-session cyclerepeated for a second year
Delivery: 6 groups of 9–10people, in person, bothdidactic and interactive(hands-on activities, groupwork, problem-solvingactivities, real-life simulationsand role play
Treatment changes: nonementioned
Training of trainers:
Theory:
Control intervention:Traditional consultationsevery 3 months in thediabetes clinic, unlessintercurrent problems. Seenby same physicians asintervention who wereunaware that patients werein the control group. Alsohad weekly diaries of bodyweight and nutrition.
Eligibility/exclusion criteria:Type 2 diabetes treated witheither diet alone or diet andOHAs, who had attendedclinic for at least 1 year
How selected: no detailsNumbers involved: total 112(56 intervention, 56 control)
Nos on insulin: none
Tablets: 50 intervention, 46control
Diet alone: 6 intervention, 10control
Type of diabetes: 2
Duration diabetes:intervention 9.4 (1–23) years,control 9.8 (1–39) years
Baseline measurements ofoutcome parameter (mean± SD):
HbA1c: intervention 7.4% ±1.4%, control 7.4% ± 1.4%,QoL (DQOL): intervention67.6 ± 19, control 66.7 ±25
Retinopathy (none/mild/moresevere): intervention 42/8/6,control 38/13/5
Knowledge: intervention14.9 ± 7.9, control 20.2 ±7.4
BMI: int: 29.7 ± 4.5, cont:27.8 ± 4.1
No. hypertensive:intervention 34, control: 25
Health conduct (CdR):intervention 11.1 ± 2.7,control: 12.0 ± 4.3
Weight (kg): intervention77.4 ± 13.1, control 78.2 ±14.6
Fasting BG (mmol/l):intervention 9.8 ± 2.6,control 10.0 ± 3.1
Cholesterol (mmol/l):intervention 5.8 ± 1.1,control 5.5 ± 0.9
HDL cholesterol (mmol/l):intervention 1.2 ± 0.3, cont1.3 ± 0.3
Primary outcomes used:HbA1c, QoL (DQOL),retinopathy
Secondary outcomes used:knowledge, BMI, healthconduct, weight, FBG,cholesterol, triglycerides,creatinine, albumin, footulcers, hypoglycaemicmedications
Individual preferred learningstyle addressed?: no
Any sub-groups: no
Normal range(s) foroutcomes: not given
How outcomes assessed?:not given
Validated?: HbA1c yes. QoLwith DQOL (slightlymodified with 6 questionsomitted from the worry,social/vocational section aspertinent to young Type 1patients) Retinopathy: unsure
Knowledge by educationstudy group of the ItalianSociety of Diabetes(reported to be valid)
Health conduct assessed byCdR: validated
Timing of outcomes same forboth groups: yes
Length of follow-up: 2 yearsfrom inception
continued
Appendix 8
126
Reference and design Intervention Participants Outcome measures
Outcome (mean ± SD) Intervention group Control group Difference between (n = 43) (n = 47) groups
HbA1c 7.5 ± 1.4% 8.3 ± 1.8% p < 0.002
DQOL 55.6 ± 15.9 80.8 ± 31.5 p < 0.001
Diabetic retinopathy 35/5/3 33/7/7 NS(none/mild/more severe)
GISED (knowledge) 24 ± 6.6 17.4 ± 8.6 p < 0.001
BMI 29.0 ± 4.4 27.6 ± 4.2 p = 0.06
No. hypertensive 26 22 NS
Health conduct (CdR) 15.8 ± 2.9 11.3 ± 4.3 p = 0.01
Weight (kg) 76.0 ± 13.4 77.1 ± 14.7 NS
Fasting BG (mmol/l) 9.9 ± 2.6 9.2 ± 2.9 NS
Total cholesterol (mmol/l) 5.7 ± 1.2 5.6 ± 1.2 NS
HDL cholesterol (mmol/l) 1.4 ± 0.4 1.3 ± 0.3 p < 0.05
Triglycerides (mmol/l) 2.1 (0.7–6.9) 1.7 (0.6–3.9) p = 0.53
Creatinine (µmol/l) 88.8 ± 16.5 87.8 ± 17.2 NS
Albuminuria (none/micro or macro) 20/21 19/22 NS
Individual education sessionsfrom same educationalist,with special reference toeating habits, homemonitoring of glucose andprevention of complications
Duration of intervention:intervention patientsaveraged 7.9 visits (7–8) andcontrol 8.2 (5–11) in 2 years
Triglyceride (mmol/l): 2.6(0.7–11.5), control 1.7(0.5–5.2)
Creatinine (µmol/l): 91.6 ±14.2, control 90.0 ± 14.0
Albuminuria (none/micro ormacro): intervention 32/24,control 37/19
Foot ulcers(never/past/active);intervention 54/0/2, control53/2/1
Hypoglycaemic treatment(intervention/control): dietonly: 6/10, sulphonylureas27/21, metformin 5/6,sulphonylureas + metformin18/19, insulin 0/0
Gender (M/F): intervention27/29, control 34/22)
Age ranges (years):intervention 62 (35–80),control 61 (43–78)
Ethnic groups: no details
Losses to follow up:intervention: 13 (3 deaths, 10moved), control: 9 (1 death,5 moved, 3 lost to follow-up)
continued
Health Technology Assessment 2003; Vol. 7: No. 22
127
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Methodological commentsAllocation to treatment groups: random number tablesBlinding of outcome assessors?: not reported (N/A for HbA1c)Allocation concealment?: not reportedAnalysis by ITT?: noComparability of treatment groups: control participants had higher levels of education and betterknowledge of diabetesMethod of data analysis: generalised linear model for repeated measures, and correlation coefficientsStandard deviation and significance levels only, no confidence intervals reportedSample size/power calculation: noAttrition/drop-out: reported as above
General commentsGeneralisability: HbA1c seems relatively low from outsetConflict of interests: Turin University research grantOther: publication of first-year results as preliminary results
Quality criteria (CRD Report 4)
Outcome (mean ± SD) Intervention group Control group Difference between (n = 43) (n = 47) groups
Foot ulcers (never/past/active) 42/1/0 45/1/1 NS
SMBG 10 14 NS
Hypoglycaemic treatment:Diet only 2 5 NSSulphonylureas 18 13 NSMetformin 3 6 NSSulphonylureas + metformin 18 25 NSInsulin 2 5 NS
1. Was the assignment to the treatment groups really random? Adequate2. Was the treatment allocation concealed? Unknown3. Were the groups similar at baseline in terms of prognostic factors? Reported4. Were the eligibility criteria specified? Yes5. Were outcome assessors blinded to the treatment allocation? Inadequate6. Were the point estimates and measure of variability presented for the primary outcome measure? Adequate7. Did the analyses include an ITT analysis? Inadequate8. Were withdrawals and drop-outs completely described? Adequate
Appendix 8
128
Reference and design Intervention Participants Outcome measures
Surname and year: Cooper et al., unpublisheda
Source: manuscriptsubmitted
Country: UK
Setting: multicentre – 2hospitals and 1 health centre
Language: English
Trial design: randomisedwait-list design
Treatment intervention:Diabetes Look After Yourself(DLAY) course:
Topics: self-management(nutrition, physical activity,relaxation, screening,management ofcomplications [foot care,sick-day rules (personalcommunication of author)]exploration of feelings, howto make best use of healthservice)
Provider: DSNs
Sessions: 8 weekly sessionsof approximately 2 h each.Delivered at staggeredintervals over 14 months
Delivery: largely interactive,small and plenary groupdiscussions, problem-basedlearning, goal setting,exercise, relaxation andpractice of skills
Treatment changes: assumenone
Training of trainers: nursetrainers trained together andprovided manual
Theory: empowerment
Ran in 3 different centres
Control intervention:Randomised but on the waitlist for 12 months
Group A (n = 30): hadoutcomes measured after 6and 12 months on DLAY
Group B (n = 23): had short-term control period for 6months and outcomesmeasured after 6 and 12months on DLAY
Group C (n = 36): long-termcontrol period for 12 monthsbefore starting DLAY
Duration of intervention: 8weeks
Eligibility criteria: Type 2diabetes diagnosed for atleast 1 year, able to givewritten consent, undergoingregular diabetes screening.Excluded if: under 21 andover 75 years old, persistentdefaulters, alcohol problem,language problem, and aphysical handicap whichprecludes them from theactivity/exercise programme(more details in Table 1)
How selected:
Numbers involved: N = 89,intervention N = 53, controlN = 36
Nos on insulin: none
Tablets: intervention 75%,control 66%
Diet alone: intervention25%, control 34%
Type of diabetes: all 2
Duration of diabetes (meanyears and range sincediagnosis): intervention 5.7(1–28); control 5.7 (1–30)
Baseline measurements ofoutcome parameters (mean± SD):
HbA1c: intervention: 7.9 ±1.7%, control: 7.0% ± 1.6%
Attitudes: intervention 73.1± 11.9, control 74.6 ± 11.0
Treatment effectiveness(median): intervention 4.4,control 4.0
BMI: intervention 32.5 ±6.7, control: 32.1 ± 6.1
Diet: intervention 71.6 ±18.2, control 69.6 ± 15.5
Exercise: intervention 50.8 ±25.5, control 48.8 ± 31.6
Self-monitoring (%):intervention 67, control 47
Gender (M/F) (%):intervention 57/43, control58/42
Age (years) (mean andrange): intervention 58.2(30–70), control 58.4 (35–73)
Primary outcomes used:HbA1c
Secondary outcomes used:Summary of Diabetes Self-Care ActivitiesQuestionnaire.Diabetes IntegrationQuestionnaire (attitudes todiabetes and its treatment).Personal Models of DiabetesQuestionnaire (treatmenteffectiveness).(qualitative outcomes notreported here)
Individual preferred learningstyle addressed: no
Any sub-groups: no
Normal range(s) foroutcomes: HbA1c: 4–6%
How outcomes assessed:HbA1c by lab., others self-report
Validated: quantitativemeasures validated
Timing of outcomes same forboth groups: yes betweenevaluations, but finalevaluation in group B 6months later
Length of follow-up: 12months from inception
continued
Health Technology Assessment 2003; Vol. 7: No. 22
129
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Methodological commentsAllocation to treatment groups: ‘blindly and randomly assigned’Blinding of outcome assessors: not reportedAllocation concealment: ‘blindly and randomly assigned’Analysis by ITT: not reportedComparability of treatment groups: higher mean HbA1c level in trial group compared with control afterattrition (7.9% vs 7.0%) – adjusted for in analysisMethod of data analysis: used both quantitative and qualitative analysis. Means, SDs and p valuesreported. Regression analysis was used in the calculation of changes in baseline HbA1c levels – to accountfor significant differences in baseline values of trial and control groupsSample size/power calculation: yes – calculated that 48 patients needed to detect a 1% change in HbA1c
This gave a power level of 95% significance at the 5% levelAttrition/drop-out: 12%
General commentsGeneralisability: only about 40% of patients asked to take part were recruited. Those refusing to take partshowed no difference in age and sex compared with those who participated. HbA1c levels were relativelygood at baseline. Patients might have been better at self-management than typical from the outset.Conflict of interests: funded by Diabetes UKOther: possible ceiling effects in treatment effectiveness evaluation
Reference and design Intervention Participants Outcome measures
Difference between Outcome (mean ± SD) Intervention group Control group groups
HbA1c 7.9 ± 2.1 7.2 ± 1.6 NS
Attitudes (scale 0–100%, 75.1 ± 11.0 70.5 ± 11.0 p = 0.01higher = better)
Treatment effectiveness 4.5 4.1 NS(median on Likert scale 0–5, higher = better)
BMI 31.3 ± 5.7 30.5 ± 3.9 NS
Diet (scale: 0–100%, 76.5 ± 12.2 68.0 ± 17.8 NShigher = better)
Exercise (scale: 0–100%, 62.5 ± 25.3 55.9 ± 25.0 NShigher = better)
Self-monitoring (% blood testing) 92 63 p = 0.002
a See footnote to Table 12.
Ethnic groups: not stated
Losses to follow up: n = 11(12%), 5 deaths (3intervention/2 control) and 6drop-outs (3 intervention/4control) (sic.)
Compliance: 76% attended 7or more sessions. (Asignificant correlationbetween attendance ratesand reductions in HbA1clevels at 12 months)
Appendix 8
130
Quality criteria (CRD Report 4) RCTs
1. Was the assignment to the treatment groups really random? Unknown2. Was the treatment allocation concealed? Unknown3. Were the groups similar at baseline in terms of prognostic factors? Reported4. Were the eligibility criteria specified? Yes5. Were outcome assessors blinded to the treatment allocation? Unknown6. Were the point estimates and measure of variability presented for the primary outcome measure? Adequate7. Did the analyses include an ITT analysis? Inadequate8. Were withdrawals and drop-outs completely described? Partial
Health Technology Assessment 2003; Vol. 7: No. 22
131
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Reference and design Intervention Participants Outcome measures
Surname and year: Heller et al., 198831
Source: published
Country: UK
Setting: hospital
Language: English
Trial design: RCT
Treatment intervention:group weight lossintervention of 4–6 patientswith a spouse or friend. Eachgiven a target weight
Topics: aim was to loseweight, what foods to eat andthose to avoid, aetiology ofdiabetes, self-monitoring, self-care, diabetic complications,the importance of eyeexaminations and foot care.Self-monitoring of urinetaught (twice a day)
Provider: one of two diabetesnurses and one dietitian
Sessions: 3 90-minutesessions at weekly intervalswith follow-up visits (90 min)at 3 and 6 months
Materials: video whichexplained foods to eat, etc., aboard for plotting weights sothe group could see progressand a book on diabetes forpatients
Delivery: group education
Treatment changes:
Training of trainers:
Theory:
Mode:
Persistent symptomsglycosuria or random BG>15 mmol/l were withdrawn
At 3 months patients visitedfor 90 minutes and lunchedwith nurse and dietitian,followed by a group discussionwith critical discussion offood choice. At 6 months visita general review undertakenand watched video again
Patients could contact nurseswithin following 6 months
Control intervention: usualclinic care, seen by doctorand then referred todietitian, seen individually.Clinic appointments asnecessary and mandatory at3, 6, 12 months. Any patientsstarted on OHAs in first yearwere withdrawn
Duration of intervention: 6 months
Eligibility criteria: all newlydiagnosed Type 2 patients(defined), overweight (BMI>27 kg/m2), 30–75 years.Excluded patients withketonuria, those in whomdiagnosis was made as aninpatient (e.g. at time ofsurgery), judged too infirmor with major languagedifficulties
How selected: from patientsreferred to clinic over 18-month period
Numbers involved: total N =87, intervention 40, control47
Nos on insulin: none
Tablets: none
Diet alone: assume all
Type of diabetes: 2
Duration diabetes: newlydiagnosed
Baseline measurements ofoutcome parameter: HbA1(mean + 95% CI):intervention 12.3% (11.4 to13.2), control 12.7% (11.9to 13.5)
Gender (M/F): intervention20/16, control 16/23
Age ranges (years) (mean +95% CI): intervention 56.6(55 to 58), control 56.4 (53to 59.9).
Ethnic groups: not reported
Losses to follow-up:intervention 4, control 8(reasons given)
Compliance: 1 control + 2intervention did not attend3-month follow-up, 1intervention did not attend6-month follow-up
Primary outcomes used:HbA1
Secondary outcomes used:knowledge, fasting BG,weight
Individual preferred learningstyle addressed? No
Any sub-groups (e.g. ethnicgroups): no
Normal range(s) foroutcomes: HbA1: 5.0–7.5%,knowledge (max. score 36)
How outcomes assessed?:knowledge self-report, lab.for HbA1
Validated?: HbA1 yes,knowledge no details ofvalidation
Timing of outcomes same forboth groups: yes
Length of follow-up: 12months from inception
continued
Appendix 8
132
Methodological commentsAllocation to treatment groups: not reportedBlinding of outcome assessors?: not reportedAllocation concealment?: not reportedAnalysis by ITT?: not reportedComparability of treatment groups: no differences reported, no statistical analysis reportedMethod of data analysis: mean or median with 95% confidence intervals; t-tests, Mann–Whitney’s and chi-squared tests usedSample size/power calculation: noAttrition/drop-out: drop-outs reported
General commentsGeneralisability: overweight population. All newly diagnosedConflict of interests: Boehringer acknowledged for donation of urine testing equipment. British DiabeticAssociation supported 2 authorsOther:
Quality criteria (CRD Report 4) RCTs
Outcome (mean ± 95% CI) Intervention group Control group Difference between (n = 36) (n = 39) groups
HbA1 9.0% (8.2 to 9.8) 9.9% (8.9 to 10.9)Proportion of patients HbA1 <7.5% 36% 28%FBG (mmol/l) 9.1 (7.9 to 10.3) 10.3 (8.8 to 11.8)Weight loss (kg) –5.5 (4 to 6.5) –3 (2 to 4) p < 0.05
Knowledge: not reported as not validated 3 and 6 months data reported.
1. Was the assignment to the treatment groups really random? Unknown2. Was the treatment allocation concealed? Unknown3. Were the groups similar at baseline in terms of prognostic factors? Reported4. Were the eligibility criteria specified? Yes5. Were outcome assessors blinded to the treatment allocation? Unknown6. Were the point estimates and measure of variability presented for the primary outcome measure? Adequate7. Did the analyses include an ITT analysis? Unknown8. Were withdrawals and drop-outs completely described? Reported
Health Technology Assessment 2003; Vol. 7: No. 22
133
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Reference and design Intervention Participants Outcome measures
Surname and year: Raz et al., 198832
Source: published
Country: Israel
Setting: hospital
Language: English
Trial design: RCT afterstratification by pre- andpost-prandial glucose andHbA1c
Treatment intervention:
Topics: explanation of thedisease, the main mode oftreatment, explanation anddemonstration of self-careand treatment techniques,the logic and practice of diet,and home exercise
Provider: physicians, a nurse,dietitian and physicaltherapist each providingdifferent topics
Sessions: three lessons within3 weeks, repeated every 4months. Patients wereencouraged to interactbetween the sessions andwere also individuallyfollowed in the diabetic clinicevery 2 months
Delivery: assume didactic,group education
Treatment changes: diet andexercise could bemanipulated, but drugtherapy unchanged
Training of trainers:
Theory:
Control intervention:Control group were followedup every 2 months
Duration of intervention: 12months
Eligibility/exclusion criteria:Type 2 diabetes, aged 30–65years, ≥ 1 year sincediagnosis, clinic record ofuncontrolled diabetes(defined) in last 12 months,no late diabetic complicationsor concurrent psychiatric orterminal illnesses
How selected: states patientswere selected from theclinic, no details.
Numbers involved: total N = 51, intervention 25,control 26
Nos on insulin: none
Tablets: 20
Diet alone: 31
Type of diabetes: 2
NB: baseline characteristicsbased on those completingstudy
Duration of diabetes (years)intervention 9.0 (SD 4.5),control 9.2 (SD 5.3)
Baseline measurements ofoutcome parameter (mean± SD):
HbA1c: intervention 10.0 ±2.7%, control 9.6 ± 2.6%
Fasting glucose: intervention200.1 ± 55.1, control 200.8± 59.9
Postprandial glucose:intervention 234.3 ± 68.6,control 238.5 ± 69.3
Cholesterol: intervention:226.1 ± 42.6, control: 220.3± 55.4
Triglyceride: intervention:232 ± 32, control: 211 ± 34
HDL cholesterol:intervention: 47.0 ± 4.2,control: 45.8 ± 4.5
Weight: 75.4 ± 11.7, 73.4 ±11.5
Gender (M/F): intervention7/16, control 10/16
Primary outcomes used:HbA1c
Secondary outcomes used:knowledge (not reportedhere), BP, weight (kg) (notreported here), pre- andpostprandial BG (notreported here), bloodcholesterol, HDL cholesterol,blood triglyceride
Individual preferred learningstyle addressed?: no
Any sub-groups (e.g. ethnicgroups): no
Normal range(s) foroutcomes: not reported
How outcomes assessed?:HbA1c lab., knowledge byself-report
Validated?: knowledge notvalidated (prepared for thisstudy)
Timing of outcomes same forboth groups: yes
Length of follow-up: 12months from inception
continued
Appendix 8
134
Methodological commentsAllocation to treatment groups: patients stratified according to mean values of pre- and postprandialglucose and HbA1c and randomised. No details of methodBlinding of outcome assessors?: labs unawareAllocation concealment?: not reportedAnalysis by ITT?: not reportedComparability of treatment groups: no differences reported in baseline characteristicsMethod of data analysis: ANOVA for repeated measures (over time) and t-tests and chi-squared betweengroups. No point estimates or confidence intervals givenSample size/power calculation: not givenAttrition/drop-out: drop-outs reported
General commentsGeneralisability:Conflict of interests: funding support not mentioned.Other:
Reference and design Intervention Participants Outcome measures
Outcomes (many Intervention group Control group Differences approximations from figure) (n = 23) (n = 26) between groups
HbA1c (%) (from Figure 3) 8.25 9.6 Interaction between intervention and time, p < 0.05
Preprandial BG (mg/dl) (from 162 210 Interaction between intervention Figure 1) and time, p < 0.01
Postprandial BG (mg/dl) (from 190 225 Interaction between intervention Figure 2) and time, p < 0.05
BP Not reported
Mean blood cholesterol (mg/dl) 213.8 ± 37.7 226.1 ± 60.8 NS
Blood triglycerides (mg/dl) 214 ± 24 204 ± 31 NS
HDL cholesterol (mg/dl) 49.6 ± 4.3 45.2 ± 4.4 NS
Weight (kg) (from Figure 4) 73 73 Interaction between intervention and time, p < 0.05
Age ranges (years):intervention 51.1 (SD 8.1),control 53.7 (SD 12.8)
Ethnic groups (Israel/Asia +Africa/Europe + America):intervention 8/7/8, control3/10/13
Losses to follow-up: 2intervention patients did notparticipate in the educationprogramme, or keepappointments
Compliance: 23 patientsparticipated in the firstmeetings, 21 in the secondand 18 in the third andfourth
Health Technology Assessment 2003; Vol. 7: No. 22
135
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Quality criteria (CRD Report 4) RCTs
1. Was the assignment to the treatment groups really random? Unknown2. Was the treatment allocation concealed? Unknown3. Were the groups similar at baseline in terms of prognostic factors? Reported4. Were the eligibility criteria specified? Yes5. Were outcome assessors blinded to the treatment allocation? Adequate6. Were the point estimates and measure of variability presented for the primary outcome measure? Inadequate7. Did the analyses include an ITT analysis? Unknown8. Were withdrawals and drop-outs completely described? Reported
Appendix 8
136
Reference and design Intervention Participants Outcome measures
Surname and year:Domenech et al.,199533
Source: published
Country: Argentina
Setting: community
Language: English
Trial design: CCT
Patients had previouslyreceived dietary advice fromtheir physicians and/or hadbeen treated with OHAs.
Treatment intervention:group intervention of up to 8patients incorporating groupdiscussion and teaching
Provider: physicians who hadpreviously participated in a 2-day instruction of theteaching programme
Sessions: 4 teaching units(90–120 min each) carriedout once per week for 1month.
Topics: normal physiologicalrange for serum glucose,symptoms of hypoglycaemia,hyperglycaemia, the renalthreshold for glucose, self-monitoring of glycosuria, theeffect of obesity, planning ofan individual meal plan, footcare, physical activity, andbasic rules to be applied onsick days
Delivery: group education
Materials: flip charts,teaching files, photographs ofdifferent food representing1000 cal, question cards toverify knowledge, anindividual log book, a patientbooklet including the maincontents, a questionnaire
Every patient wasencouraged to attendaccompanied by spouse
After session 1 a very lowcalorie diet (600 cal) wasrecommended for alternativedays for 1 week and to stopthe intake of OHA, therebygiving an opportunity to testthe effect of diet uponglucose levels. Testing forglycosuria was recommendedtwice per day, 2 h after food
Control intervention: usualcare
Duration of intervention:1 month
Eligibility/exclusion criteria:excluded if newly diagnosedType 2 diabetes, aged over60 years, presence ofadvanced microangiopathiccomplications and presenceof other severe diseases (e.g.cancer)
How selected: the first 6–7patients consulting eachphysician were selected forinclusion. In the controlgroups a larger number wereincluded as were expecting alarger drop-out and in orderto obtain a better match byage, gender and duration ofdiabetes
Numbers involved: total N = 124, intervention 53,control 71
NB: baselines based on thosecompleting study
Nos on insulin: not reported,assume nil
Tablets: intervention 29,control 32
Diet alone: assumeintervention 11, control 7
Type of diabetes: 2
Duration of diabetes (years):intervention 6.9 (±0.7),control 6.3 (±1.3)
Baseline measurements ofoutcome parameter: HbA1intervention 9% (±2.6),control 9% (±2.2)
Gender (M/F): intervention18/22, control 17/22
Age ranges (years):intervention 52.7 (SE 3.1),control 53.1 (SE 1.1)
Ethnic groups: not reported
Losses to follow up:intervention 13, control 32(details given for interventiongroup only)
Primary outcomes used:HbA1
Secondary outcomes used:knowledge weight (kg), dailyintake OHAs
Individual preferred learningstyle addressed?: no
Any sub-groups (e.g. ethnicgroups): no
Normal range(s) foroutcomes: HbA1 <7.5%
How outcomes assessed?:lab., knowledge by selfreport
Validated?: HbA1 yes,knowledge no
Timing of outcomes same forboth groups: yes
Length of follow-up: 12months from inception
continued
Health Technology Assessment 2003; Vol. 7: No. 22
137
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Knowledge not reported as not a valid measureAlso reports percentage of patients who showed an improvement of more than 0.5% which was notsignificant between groups (data in figure only)Also reports that within groups a significant correlation in those who exhibited a significant decrease inHbA1 (>0.5%) was associated with significant weight loss and a reduction in OHAs agents
Methodological commentsAllocation to treatment groups: non-randomised trialBlinding of outcome assessors?: not reportedAllocation concealment?: non-randomised trialAnalysis by ITT?: noComparability of treatment groups: reported to be comparable in socio-economic levels and matched forage, gender and duration of diabetes. Also strict criteria were adopted to standardise between the twogroups the level of dietary caloric intake and OHA prescriptionMethod of data analysis: method not reported, assume ± = SDSample size/power calculation: noAttrition/drop-out: percentages reported
General commentsGeneralisability: few baseline data reportedConflict of interests: course materials were provided by Boehringer MannheimOther: unsure of control group intervention, patients in intervention groups all had different tutors
Quality criteria (CRD Report 4) CCTs
Outcomes changes Intervention group Control group Differences (mean difference ± SD) (n = 40) (n = 39) between groups
HbA1 (%) –0.2 (0.4) +0.8 (0.4)Weight (kg) –2.4 (0.5) –0.4 (0.5) p < 0.01Daily intake of OHA (no. of tablets) –1.4 (0.2) +0.9 (0.2) p < 0.01
Were the groups similar at baseline in terms of prognostic factors? ReportedWere the eligibility criteria specified? YesWere outcome assessors blinded to the treatment allocation? UnknownWere the point estimates and measure of variability presented for the primary outcome measure? PartialDid the analyses include an ITT analysis? UnknownWere withdrawals and drop-outs completely described? AdequateWere participants likely to be representative of the intended population? Yes
Appendix 8
138
Reference and design Intervention Participants Outcome measures
Surname and year:Kronsbein et al., 198834
Source: published
Country: Germany
Setting: general practices
Language: English
Trial design: CCT, conditionsimplemented by practice
Treatment intervention:
Provider: specially trainedphysicians’ assistants
Topics: basic information,metabolic self-monitoring,reasons for raised BG levels,OHAs, diet, foot care,physical activities, sick-dayrules, late complications
Sessions: 90–120 minuteseach week for 4 weeks;groups of 4–6 patients; focuson group interaction witheach session includingexperiential, theoretical andpractical aspects
Treatment changes:
Training trainers:
Theory:
Mode:
Control intervention: usualcare within general practices;all patients before trial hadbeen given unstructureddietary advice by physiciansand/or were treated withoral sulphonylureas
Duration of intervention:4 weeks
Eligibility: WHO criteria forNIDDM
Exclusion: physical or mentalhandicaps that preventedthem from following theintervention programme
How selected: 8 GPsattending teachingprogramme volunteered tointroduce programme – 5practices immediately, 3 after1 year. Interventionparticipants: all consecutivepatients who participated infirst three courses
Numbers involved:
Starting total 127;intervention 65, control 62
Total (those completingfollow-up) 99; intervention50, control 49
Type of diabetes: 2
Duration of diabetes (years± SD): Intervention 7 ± 5,control 7 ± 6
Baseline measurements ofoutcome parameter (mean± SD):
HbA1c: intervention 7.1 ±1.6%, control 6.5 ± 1.6%
Weight (kg): intervention76.5 ± 12.6, control 75.1+12.9
Knowledge: intervention 9 ±3, control 9 ± 3
No. without glucose-lowering medications:intervention (%) 32%,control 39%
Gender (M/F) (%):intervention 42/58, control39/61
Age ranges (years) (mean ±SD): intervention 65 ± 9,control 63 ± 8
Ethnic groups: not reported
Losses to follow-up:intervention 15, control 13
Primary outcomes used:HbA1c
Secondary outcomes used:knowledge score no. on BGlowering medications,treatment with insulin,frequency self-monitoringurine, bodyweight
Individual preferred learningstyle addressed?: no
Any sub-groups (e.g. ethnicgroups): No
Normal range(s) foroutcomes: HbA1c up to 5.6%
How outcomes assessed:HbA1c by lab., knowledge byspecially designedquestionnaire, no. onmedications not reported,self-report glycosuria testing
Validated: knowledgequestionnaire assumedvalidated, reference provided
Timing of outcomes same forboth groups:
Length of follow-up: 1 yearfrom inception
continued
Health Technology Assessment 2003; Vol. 7: No. 22
139
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Methodological commentsAllocation to treatment groups: group formed by treatment within participating practices or not, all GPsreceived programme trainingBlinding of outcome assessors: not reportedAllocation concealment: not randomisedAnalysis by ITT: noComparability of treatment groups: reported that baseline characteristics of those completing and notcompleting follow-up did not differMethod of data analysis: hypothesis tests with confidence intervals for within-group and between-groupdifferencesSample size/power calculation: reported power required ~55 patients per groupAttrition/drop-out: yes
General commentsGeneralisability: both patient groups started with relatively low HbA1c and therefore may not berepresentativeConflict of interests: none reportedOther:
Quality criteria (CRD Report 4) CCTs
Outcome (mean & SD) Intervention group Control group Difference between (n = 50) (n = 49) groups (95% CI)
HbA1c 7.1 ± 1.6 6.7 ± 1.5 NS
Knowledge 13 ± 4 10 ± 4 3 (16 to 48)**
% without glucose-lowering 62 39 23 (3 to 43)*medications
Treatment with insulin 0 10 10 (2 to 18)*
Bodyweight (kg) 73.8 ± 12.6 74.8 ± 13.2 2.3 (1.0 to 3.6)**
Self-monitoring glycosuria (%) 72 2 70 (57 to 83)**
*Difference between groups, p < 0.05; **difference between groups, p < 0.0001.
Were the groups similar at baseline in terms of prognostic factors? ReportedWere the eligibility criteria specified? YesWere outcome assessors blinded to the treatment allocation? UnknownWere the point estimates and measure of variability presented for the primary outcome measure? AdequateDid the analyses include an ITT analysis? UnknownWere withdrawals and drop-outs completely described? PartiallyWere participants likely to be representative of the intended population? No
Appendix 8
140
Interventions of focused self-management education
Reference and design Intervention Participants Outcome measures
Surname and year: Kaplan et al., 198735
Source: published
Country: USA
Setting: unclear
Language: English
Trial design: RCT
Four groups: diet education(group 1), exercise education(group 2), diet and exerciseeducation (group 3) andcontrol education (control)
All given the exchange diet(1200 cal) recommended byADA and each received anexercise prescription basedon baseline exercise test. Adeposit of $40 was requestedwith return if attend andmeet predetermined goals.Treatment interventionsincorporated behaviouralmodification (stretching andwalking and target heartrate) and strategies toincrease compliance. Thecontrol did not
Sessions: groups 2 h onceper week for 10 weeks
Treatment intervention:
Group 1 (diet):
Provider: Dietitian explainedthe diet
Topics: identification of goals,used principles of modernlearning theory. Diarymonitoring of eatingbehaviour. Identification ofexternal cues that lead toover/inappropriate eating
Theory: used positivereinforcement. Alsorecorded own cognitions(positive and negative self-statements) and discussed ingroup. Also brief relaxation.Ref. 11 for more details
Treatment changes:
Training trainers:
Mode:
Group 2 (exercise):
Provider:
Topics: goal setting, planningfor exercise, self-monitoringintroduced, completion ofdiary, question answering andgroup exercise sessions.Used positive feedback, andgave suggestions formanaging problems
Eligibility/exclusion criteria:confirmed diagnosis, fastingplasma glucose >3.62mmol/l
How selected: radio +newspaper advertisementsand physicians
Numbers involved: total N = 87, unsure of groupnumbers
Nos on insulin: 19
Tablets: 29
Diet alone: 28
Type of diabetes: 2
Duration diabetes: notrecorded
Baseline measurements ofoutcome parameter: HbA1c,group 1 8.97% (SD 2.82),group 2 8.16% (3.44), group3 9.18% (2.46), control 8.21(1.54)
Gender (M/F): 32/44
Age ranges (years): group 154.87 (SD 12.32), group 253.81 (8.04), group 3 56.96(8.95), control 54.5 (8.83)
Ethnic groups: not reported
Losses to follow-up: 11(reasons given)
Compliance: averageattendance >80% for allgroups
Primary outcomes used:HbA1c, QoL
Secondary outcomes used:weight (kg)
Individual preferred learningstyle addressed?:
Any sub-groups (e.g. ethnicgroups): no
Normal range(s) foroutcomes: see appendix intext
How outcomes assessed?:HbA1c lab., QoL self-reportquestionnaire
Validated?: QoL yes
Timing of outcomes same forall groups: yes
Length of follow-up: 18months from inception
continued
Health Technology Assessment 2003; Vol. 7: No. 22
141
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Methodological commentsAllocation to treatment groups: states randomly chosen, otherwise no detailsBlinding of outcome assessors?: not reported
Reference and design Intervention Participants Outcome measures
Outcomes Group 1 (diet) Group 2 (exercise) Group 3 Group 4 (control, (18 months) (diet+ exercise) education)
HbA1c* 8.51% 9.46% 7.70%** 8.57%
QoL (change scores)* +0.03** No improvement +0.06** –0.04
Weight Data not reported, Data not reported, Data not reported, Data not reported, no changes no changes no changes no changes
*Overall marginally significant difference between groups (p < 0.10); **significant from group 4, p < 0.05.There were significant correlations between improvements in QoL and decreases in HbA1c (r = –0.22, p < 0.05).Some costs/utility analysis reported.
Treatment changes:
Training of trainers:
Theory:
Mode:
Group 3 (diet and exercise):
Provider:
Topics: modified dietaryintervention for 5 weeks,then focused on exercise,self-monitoring, foot careand stretching, then followedexercise and behaviourmodification format
Treatment changes:
Training of trainers:
Theory:
Mode:
Control intervention:(education):
Provider: exposed to healthcare specialists including anendocrinologist, podiatrist,ophthalmologist, psychologist,dietitian, official from ADA,representative fromcompany that manufactureshome glucose monitoringequipment and physiologist
Session: Each providerpresented for 1 session (2 h)in form of lecture providingdiabetes care
Treatment changes:
Training of trainers:
Theory:
Mode:
Duration of intervention: 10 weeks
Appendix 8
142
Allocation concealment?: not reportedAnalysis by ITT?: not reportedComparability of treatment groups: no significant differences reportedMethod of data analysis: change scores compared with ANOVA, no estimate of variance givenSample size/power calculation: post hoc power analysisAttrition/drop-out: percentages given
General commentsGeneralisability: minimal eligibility criteria, baseline characteristics suggest generalisableConflict of interests: funding support not mentionedOther: unsure of N in each group
Quality criteria (CRD Report 4)
1. Was the assignment to the treatment groups really random? Unknown2. Was the treatment allocation concealed? Unknown3. Were the groups similar at baseline in terms of prognostic factors? Reported4. Were the eligibility criteria specified? Yes5. Were outcome assessors blinded to the treatment allocation? Unknown6. Were the point estimates and measure of variability presented for the primary outcome measure? Inadequate7. Did the analyses include an ITT analysis? Unknown8. Were withdrawals and drop-outs completely described? Reported
Health Technology Assessment 2003; Vol. 7: No. 22
143
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Reference and design Intervention Participants Outcome measures
Surname and year: Uusitupaet al., 1992–636,109–113
Source: published
Country: Finland
Setting: hospital outpatient
Language: English
Trial design: RCT
Basic education to bothgroups: prior torandomisation for 3 months,both groups received basiceducation (basic knowledgeof NIDDM, dietary advice tolose weight, reduce intake ofsaturated fat and cholesterol,and increase the use ofunsaturated fat and unrefinedcarbohydrates)
Both groups, after the 1-yearintervention period, wereadvised to visit local healthcentres at 3-month intervalsand the research centre at 21and 27 months.
Treatment intervention:
Topics: 1. individualisedintensified dietary education(principles of the diabeticdiet, fat, carbohydrate, fibre,sweeteners, special diabeticproducts, behaviourmodification, review ofimportant things in diet, foodpreparation), recommendedan individually tailored diet,compliance measured byfood records and fatty acidsof serum lipids
2. exercise training: oral andwritten instructions –proposed walking, jogging,cycling, swimming, cross-country skiing.Recommended heart rateduring sessions 110–140beats per minute.Recommended 3–4 timesper week for 30–60 minutes
Provider: physician, DSN(s),clinical nutritionist
Length and no. of sessions:six visits to the clinic (at 2-month intervals)
Recommended frequency ofexercise training 3–4 sessionsper week of 30–60 minuteseach.
Mode: given in person at thelocal health centre
Treatment changes: no
Training of trainers:
Theory:
Eligibility criteria: obese,newly diagnosed Type 2patients aged 40–64 years,FBG levels of ≥ 6.7 mmol/l
How selected: physiciansworking in five rural and oneurban health centre inKuopio, referred all newlydiagnosed patients from1987 to 1989
Numbers involved: total N = 86, intervention 40,control 46
Nos on insulin: none
Tablets: 7 (intervention = 2,control = 5) (1 in trial 2283)
Diet alone: assume 79 (85 intrial 2283)
Type of diabetes: 2
Duration of diabetes: allnewly diagnosed
Baseline measurements ofoutcome parameters: mean(SD):
Weight (kg): intervention88.3 (14.1), control 88.8 (14)
BMI: intervention 32.0 (5.2),control 31.6 (4.8)
FBG: (mmol/l): intervention6.6 (1.9), control 7.5 (2.9)
FBG adjusted: (mmol/l):intervention 7.0, control 7.2
% patients with FBG ≤ 6.7mmol/l: intervention 37.5,control 26.1
HbA1c (%): intervention 7.1(1.8), control 7.8 (2.0)
HbA1c adjusted (%)intervention 7.4, control 7.8
% patients with HbA1c≤ 7.0%: no data reported
Total cholesterol (mmol/;):intervention 6.1 (1.2),control 6.3 (1.0)
HDL cholesterol (mmol/l):intervention 1.07 (0.25),control 1.17 (0.29)
Non-HDL cholesterol(mmol/l): intervention 5.1(1.3), control 5.1 (1.0)
Primary outcomes used:HbA1c
Secondary outcomes used:BP FBG, weight, BMI,cholesterol, HDL cholesterol,non-HDL cholesterol,triglycerides, food intake,apolipoproteins A1 and B,HDL cholesterol/totalcholesterol, drug treatment,aerobic capacity
Individual preferred learningstyle addressed: no
Any sub-groups (e.g. ethnicgroups): no
Normal range(s) foroutcomes: not reported
How outcomes assessed:bodyweight measured withelectric scale; physiologicalmeasures by lab., BP nursemeasured (mean of 3measurements), food intakeself-report
Validated: yes, except self-report measures
Timing of outcomes same forboth groups: yes
Length of follow-up: after the1-year intervention period,patients followed up for afurther 12 months
continued
Appendix 8
144
Reference and design Intervention Participants Outcome measures
Outcome (24 months: Intervention group Control group Difference between groupsintervention N = 38, control N = 44) mean ± SD
HbA1c (%)12 months 6.6 (1.6) 7.5 (1.7)24 months 7.2 (1.9) 8.0 (1.6)
HbA1c adjusted (%)12 months 6.7 7.324 months 7.4 7.9
% patients with HbA1c ≤ 7.0%12 months 74.4%** 47.8% **p = 0.00524 months 55.3%a 31.8% ap = 0.016
BMI12 months 31.4 (5.0) 31.9 (4.6)24 months 31.9 (5.0) 32.2 (4.5)
Blood pressure systolic (mmHg)12 months 137 (16) 144 (18)24 months 146 (19) 150 (22)
Blood pressure diastolic (mmHg)12 months 83 (9) 85 (9)24 months 88 (10) 87 (9)
Total cholesterol (mmol/l)12 months 6.0 (1.0) 6.4 (1.0)24 months 6.4 (1.3) 6.5 (1.1)
HDL cholesterol (mmol/l)12 months 1.20 (0.29) 1.21 (0.28)24 months 1.17 (0.24) 1.19 (0.29)
Non-HDL cholesterol (mmol/l)12 months 4.8 (1.0)24 months 5.2 (1.0)
Control intervention:
Usual education given at thelocal health centres thatoriginally referred them.They visited at 2–3-monthintervals, plus twice visitedthe outpatient clinics
Duration of intervention:12 months
Triglycerides (mmol/l):intervention 2.50 (1.44),control 2.26 (1.33)
BP systolic (mmHg):intervention 140 (16),control 137 (16)
BP diastolic (mmHg):intervention 87 (11), control83 (9)
Gender (M/F): intervention21/19, control 28/18
Age ranges (years): 40–64.Mean (SD) ages at diagnosis:intervention 52.2 (6.5),control 54.2 (6.5)
Ethnic groups: not reported
Losses to follow-up: at 2-year follow-up 2 lost in eachgroup. Reasons not given
continued
Health Technology Assessment 2003; Vol. 7: No. 22
145
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Methodological commentsAllocation to treatment groups: unclear, only reports ‘randomised’Blinding of outcome assessors: not relevantAllocation concealment: not reportedAnalysis by ITT: not reportedComparability of treatment groups: intervention group lower for FBG and HbA1c – difference not testedstatistically. Values were adjusted as covariates into MANOVA (multivariate analysis of variance)procedures and into the two-way covariance analysis (ANCOVA)Method of data analysis: MANOVA, ANCOVA, t-tests. Analysis of variance used to test differences betweengroups. p values reported. Variables expressed as mean (SD)Sample size/power calculation: noAttrition/drop-out: numbers reported, but no reasons given
General commentsGeneralisability: 108 patients were recruited and 86 randomised – 11 did not fulfil selection criteria and11 refusedConflict of interests: funding from Finnish Medical Council, Academy of Finland, Finnish Ministry ofEducation, Finnish Foundation for Diabetes ResearchOther: significant decrease for both groups for bodyweight, FBG and HbA1c during 3 months of basiceducation before randomisation
Outcome (24 months: Intervention group Control group Difference between groupsintervention N = 38, control N = 44) mean ± SD
Triglycerides (mmol/l)12 months 1.96 (0.89) 2.33 (1.19)24 months 2.34 (1.19) 2.25 (1.25)
Weight (kg)12 months 86.5 (13.7) 90.2 (14.3)24 months Men (n = 20) 91.8 (10.7); Men (n = 26) 95.1 (10.3);
Women (n = 18) 83.1 (14.2) Women (n = 18) 84.8 (18.1)
FBG (mmol/l)12 months 6.2 (1.8) 7.5 (2.2)24 months 7.1 (2.4) 8.2 (2.3)
FBG adjusted (mmol/l)12 months 6.4* 7.3 *p < 0.0224 months 7.4 8.0
% patients with FBG ≤ 6.7 mmol/l12 months 75** 52.2 **p = 0.00524 months 55.3a 31.8 ap = 0.016
Apoliprotein A112 months 1.38 (0.19) 1.41 (0.18)
Apoliprotein B12 months 1.13 (0.24)* 1.26 (0.27) *p < 0.02
HDL cholesterol/total cholesterol12 months 0.20 (0.05) 0.19 (0.05)
Drug treatment (percentage taking)24 months 12.5** 34.8 **Significant from control
p = 0.005
Most of the comparisons reported were within groups. Only comparisons between groups are reported below. Self-reportoutcomes not reported here.
Appendix 8
146
Quality criteria (CRD Report 4) RCTs
1. Was the assignment to the treatment groups really random? Unknown2. Was the treatment allocation concealed? Unknown3. Were the groups similar at baseline in terms of prognostic factors? Reported4. Were the eligibility criteria specified? Yes5. Were outcome assessors blinded to the treatment allocation? Unknown6. Were the point estimates and measure of variability presented for the primary outcome measure? Adequate7. Did the analyses include an ITT analysis? Inadequate8. Were withdrawals and drop-outs completely described? Unknown
Health Technology Assessment 2003; Vol. 7: No. 22
147
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Reference and design Intervention Participants Outcome measures
Surname and year: Ridgeway et al., 199937
Source: published
Country: USA
Setting: communityambulatory clinic
Language: English
Trial design: RCT
Treatment intervention:
Topics: dieting and exercisewere emphasised asimportant in the control ofdiabetes. Diet and exerciseprescriptions and goals setindividually. Contracts madeto emphasise patientparticipation and personalresponsibility
Provider: registered nurseand a dietitian
Sessions: 1.5 h per month ×6.
Delivery: group intervention,didactic and interactive
Treatment changes: bothgroups seen by physicians inthe usual manner.
Training of trainers: certifieddiabetes educators
Theory: didactic based on lifeskills programme
Control intervention:Assume normal care withclinic visits
Duration of intervention: 6months
Changes to treatment: OHAmedication started orincreased 1 intervention, 4control, stopped ordecreased 1 intervention, 0control, insulin increased 2intervention, 2 control; OHAreplaced by insulin 0intervention, 3 control
Eligibility/exclusion criteria:Type 2 diabetes (defined), atleast 20% over ideal weight,able to travel to clinicmonthly, judged by physicianto be able to comprehenddietary and diabetic teaching,had inadequately controlleddiabetes (FBG >150 mg/dland HbA1c above normalrange)
How selected: computerisedaudit was conducted andyielded 150 patients ofwhom 56 met inclusioncriteria
Numbers involved: total N = 56, intervention 28,control 28
Nos on insulin: intervention3, control 3
Tablets: intervention 12,control 13
Diet alone: intervention 3,control 4
Type of diabetes: 2
Duration of diabetes (years):intervention 10=, control 13
Baseline measurements ofoutcome parameter (mean± SD):
GHb: intervention 12.3 ±2.2%, control 12.3 ± 3.0%
Knowledge intervention (n = 17) 74.2, control notreported
QoL not reported
Diabetes symptoms:intervention 43.8 ± 14.7,control 44.5 ± 19
FBG: intervention 215,control 210
Total cholesterol:intervention 259, control 224
HDL cholesterol:intervention 40, control 40
Triglycerides: intervention634, control 381
LDL cholesterol: intervention133, control 119
Primary outcomes used:GHb, QoL (MOS SF-36 andDRP questionnaires),symptoms
Secondary outcomes used:knowledge (life skills test),FBG, total cholesterol, HDLcholesterol, triglycerides,LDL cholesterol
Individual preferred learningstyle addressed?: no
Any sub-groups (e.g. ethnicgroups):
Normal range(s) foroutcomes: GHb 4.8–7.8%.Knowledge scored aspercent of correct answers.No values for QoL
How outcomes assessed?:GHb by lab. Others byquestionnaire, presume self-report
Validated: GHb yes, MOS SF-36 unclear whethervalidated; unclear whetherDRP and life skills testsvalidated
Timing of outcomes same forboth groups: assume yes
Length of follow-up: 12months from inception
continued
Appendix 8
148
Methodological commentsAllocation to treatment groups: states randomly assigned in text but no details of method of anyrandomisation also states that education was recommended to patients after ‘randomisation’ which all ineducation group acceptedBlinding of outcome assessors?: not reportedAllocation concealment?: not reportedAnalysis by ITT?: noComparability of treatment groups: groups similar on baseline characteristicsMethod of data analysis: t-tests. Standard error (difference within groups) given. No other measure ofvariance reported. No confidence intervalsSample size/power calculation: not calculated, reported to be likely numbers available in a small generalinternal medicine group practiceAttrition/drop-out: yes
General commentsGeneralisability: small group, large proportion of drop outs, GHb poor at outset in both groups, patientsjudged to be able to comprehend teaching by physiciansConflict of interests: funding by dept of medicineOther: cost estimate for programme $95 for educational materials and salaries, excluding laboratory costs
Reference and design Intervention Participants Outcome measures
Outcome (12 months) Intervention group Control group Difference between groups(n = 18) (n = 20)
GHb 11.52% 11.64% NS
QoL No data presented
Knowledge 85.7 No 12-month data presented
Symptoms No data presented
Weight (lb) 186 186 NS
FBG 205 185 NS
Total cholesterol 219 234 p = 0.09
HDL cholesterol 36 37 NS
Triglycerides 485 336 NS
LDL cholesterol (in patients with triglyceride <400) 130 125 NS
Gender (M/F): intervention6/12, control 5/15
Mean age (years):intervention 62, control 65
Ethnic groups: not reported
NB: baseline characteristicsbased on those completingstudy
Losses to follow-up:intervention 10, control 8(reasons given)
Compliance: intervention atleast 5 classes
Health Technology Assessment 2003; Vol. 7: No. 22
149
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Quality criteria (CRD Report 4) RCTs
1. Was the assignment to the treatment groups really random? Unknown2. Was the treatment allocation concealed? Unknown3. Were the groups similar at baseline in terms of prognostic factors? Reported4. Were the eligibility criteria specified? Yes5. Were outcome assessors blinded to the treatment allocation? Unknown6. Were the point estimates and measure of variability presented for the primary outcome measure? Inadequate7. Did the analyses include an ITT analysis? Inadequate8. Were withdrawals and drop-outs completely described? Adequate
Appendix 8
150
Reference and design Intervention Participants Outcome measures
Surname and year: Wing et al., 198538
Source: published
Country: USA
Setting: community
Language: English
Trial design: RCT 3 groups
Treatment intervention:
Behaviour modification:
Provider: behaviouralpsychologist and nutritionist
Topics: info. on nutrition,exercise, diabetes,behavioural strategies
Self-monitor diet
Caloric goal for exercise andgroup exercise
Contingency contractrefunded $3 per lb of weightloss
Changing eating environment
Changing cognitions
Sessions: weekly for 16weeks in groups
Treatment changes:
Training of trainers
Theory:
Mode: lecture + discussionon topic related to diet andexercise.
Nutrition education:
Provider: as above
Topics: diet – followexchange list eating planclosest to caloric goal
Nutrition topics
Importance of exercise
No requirement to self-monitor either diet orexercise
No contingency contract forweight loss
Sessions: weekly for 16weeks in groups
Treatment changes:
Training of trainers
Theory:
Mode: as above
Control intervention:treatment program identicalin content with nutritioneducation except only 4monthly meetings
Duration of intervention:intervention for 16 weeksand follow-up for 1 yearafter intervention
Eligibility criteria: 30–70years of age, 20% or moreabove ideal weight for height,diabetes being treated bydiet only or by OHAmedication, Type 2 diabetesby criteria specified byNational Diabetes DataGroup
How selected: recruited vianewspaper advertisementsand articles and letters tophysicians
Numbers involved: total 53,no. in each group notreported
Nos on insulin: none
Tablets: 75%
Diet alone: 25%
Type of diabetes: 2
Duration of diabetes: 5.9years
Baseline measurements ofoutcome parameter:
HbA1: 9.3 ± 0.3 (mean ±SEM)
BMI: 34.8 ± 7
BDI: 11.2
Gender (M/F): 20/33
Age (years) (mean ± SEM):55.1 ± 1
Ethnic groups: not reported
Losses to follow-up: 3
Primary outcomes used:HbA1
Secondary outcomes used:BP, Beck depressioninventory (BDI), BMI, insulin,total cholesterol, totaltriglycerides, HDLcholesterol, FBG, activity,food frequency, eatingbehaviour inventory
Individual preferred learningstyle addressed?: no
Any sub-groups (e.g. ethnicgroups): no
Normal range(s) foroutcomes: not reported
How outcomes assessed:lab., nurse measure and self-report
Validated: yes except activity,food frequency, eatingbehaviour inventory
Timing of outcomes same forboth groups: yes
Length of follow-up: 12 months post-intervention(16 months from inception)
continued
Health Technology Assessment 2003; Vol. 7: No. 22
151
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
ResultsNo physiological measures differed between groups, therefore results were reported for all 3 groupscombined
Methodological commentsAllocation to treatment groups: method of randomisation not reportedBlinding of outcome assessors: BP assessment blinded, others not reportedAllocation concealment: not reportedAnalysis by ITT: noComparability of treatment groups: reported that there were no differences in groups in pretreatmentphysiological measuresMethod of data analysis: hypothesis tests (ANOVA), no confidence intervalsSample size/power calculation: not reportedAttrition/drop-out: 3/53, not reported from within groups
General commentsGeneralisability: participants self-selected to participate on basis of advertisements or suggestion fromphysician, therefore may be more motivated than average patient; however, this would be true acrossconditionsConflict of interests: no mentionOther: none
Quality criteria (CRD Report 4) RCTs
Outcome Behaviour group Nutrition group Standard care Difference between groups
Weight (kg) –1.78 –3.03 –3.43 NS
1. Was the assignment to the treatment groups really random? Unknown2. Was the treatment allocation concealed? Unknown3. Were the groups similar at baseline in terms of prognostic factors? Reported4. Were the eligibility criteria specified? Yes5. Were outcome assessors blinded to the treatment allocation? Unknown6. Were the point estimates and measure of variability presented for the primary outcome measure? Partially7. Did the analyses include an ITT analysis? Unknown8. Were withdrawals and drop-outs completely described? Partial
Appendix 8
152
Reference and design Intervention Participants Outcome measures
Surname and year: Wing et al., 198639
Source: published
Country: USA
Setting: community andhome
Language: English
Trial design: RCT
Common treatmentcomponents: all sessions:individual weigh-in, BGmeasurement, discussion ofbehaviour modification forweight control. Given astandard behavioural weightcontrol programme. A dailycalorie goal set. Caloriebooks and self-monitoringdiaries were distributed.Patients asked to self-monitortheir food intake and to walkto exercise. Behaviourmodification techniques werepresented. All patientsdeposited $85 which couldbe earned back for meetingtreatment contingencies
Treatment intervention =Glucose monitoring group:
Providers:
Topics: focused on therelationship between weightloss and BG control. Taught tomonitor BG and valuesrecorded on a self-monitoringform; both the form and usedstrips were returned to theoffice at each meeting.Patients encouraged to keepBG levels normal by adjustingcaloric intake and expenditure.
Sessions: weekly meeting for12 weeks, monthly meetingsfor the next 6 months andfollow-up sessions at 9 and12 months
Treatment changes:
Training of trainers:
Theory:
Mode:
Control intervention =weight control group.Focused on weightreduction. BG levels checkedat each meeting so adjustscould be made tomedication, but no praise orreinforcement was given forBG control. Sessions asintervention group.
Duration of intervention:12 weeks
Eligibility: Type 2 diabetes,between 35 and 65 years;20% above ideal weight forheight; use of OHAmedication or insulin forcontrol of BG; diagnosis ≥ 30years.
Exclusion criteria: patientshaving prior experience withhome monitoring of BG
How selected: About 2/3were self-referred; 1/3referred by their physicians
Numbers involved: N = 50,(25 weight control group, 25glucose monitoring group)
Nos on insulin: weightcontrol = 48%, glucosemonitoring = 52%
Type of diabetes: all 2
Duration of diabetes: notgiven
Baseline measurements ofoutcome parameter:
FBG: weight control group(N = 22) 207 ± 70.5,glucose monitoring group (N= 22) 209.2 ± 69.7
HbA1 (%): weight controlgroup (N = 21) 10.86 ±2.00, glucose monitoring (N= 22) 10.19 ± 2.51
Weight (kg), mean ± SD:weight control group (N =22) 96.35 ± 23.57
Gender (% male): weightcontrol group = 20%, glucosemonitoring group = 24%Overall 39 women/11 men
Age (years): overall average54 years; weight controlgroup = 54.0, glucosemonitoring group = 53.5
Ethnic groups: not given
Losses to follow-up: 5(10%): 3 from weightcontrol group and 3 fromglucose monitoring group
Primary outcomes used:glycosylated haemoglobin(HbA1)
Secondary outcomes used:self-reported depression,weight (kg), FBG, BP,triglyceride levels, totalcholesterol levels, HDLcholesterol, decreases inmedication (others reportedonly for 12 weeks)
Individual preferred learningstyle addressed?: no
Any sub-groups (e.g. ethnicgroups):
Normal range(s) foroutcomes: FBG levels =60–120 mg/dl, HbA1 = 6.5± 0.5%
How outcomes assessed:Beck depression inventoryscale for depression (self-report), BP nurse, lab.physiological measures, self-report compliance
Validated: yes
Timing of outcomes same forboth groups: yes
Length of follow-up: 12months from inception
continued
Health Technology Assessment 2003; Vol. 7: No. 22
153
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Methodological commentsAllocation to treatment groups: randomisation blocked according to sex and % overweight, no otherdetailsBlinding of outcome assessors: nurse unaware B/P, HbA1 not applicable, others unclearAllocation concealment: not statedAnalysis by intention to treat: noComparability of treatment groups: no significant differences between groups reportedMethod of data analysis: repeated-measures analysis of variance used to compare physiological changes inpatients in the two groups. p values givenSample size/power calculation: noAttrition/drop-out: reports 10%, however, numbers for outcomes also reduced but no details
General commentsGeneralisability: Approximately two-thirds of patients were self-referred (and perhaps more motivated),so may not be generalisable to all patientsOther:
Quality criteria (CRD Report 4) RCTs
Reference and design Intervention Participants Outcome measures
Outcomes Weight control group Glucose monitoring group Difference between groups(n = 22) (n = 23)
HbA1 (%) 10.44 ± 2.16 10.19 ± 2.29
Beck depression inventory No data provided
FBG (n = 22) 210 ± 73.1 216.2 ± 58.7
Decreases in medication (%) Oral agents 64 Oral agents, 73 NSInsulin 64 Insulin 83
Serum lipids did not differ between groups. Analysis for BP, triglyceride levels, total cholesterol levels, HDL cholesterol onlytested before and after.
Compliance: assessed by self-report records and by a‘marked item’ technique.Patients used 89.1% of theassigned strips duringtreatment and 70.2% duringthe follow-up period. Theydetected 86.7% of themarked items duringtreatment and 62.8% duringfollow-up
1. Was the assignment to the treatment groups really random? Unknown2. Was the treatment allocation concealed? Unknown3. Were the groups similar at baseline in terms of prognostic factors? Reported4. Were the eligibility criteria specified? Yes5. Were outcome assessors blinded to the treatment allocation? Adequate6. Were the point estimates and measure of variability presented for the primary outcome measure? Partially7. Did the analyses include an ITT analysis? Unknown8. Were withdrawals and drop-outs completely described? Reported
Appendix 8
154
Reference and design Intervention Participants Outcome measures
Surname and year: Samaraset al., 199740
Source: published
Country: Australia
Setting: community – hospitaloutpatient clinic
Language: English
Trial design: RCT
Treatment intervention:
Topics: initially a needsassessment undertaken usingfocus groups of outpatientswhere contributing factorsfor exercise non-compliancewere identified and classified.Strategies to overcomebarriers, build self-esteemand motivation and provideprofessional and peersupport. Safe exercise,exercise-specific education toimprove confidence, copingwith diabetes and exercise,self-esteem issues, decisionmaking, goal setting andachieving mastery andenjoyment in exercise
Provider: designed andundertaken by nurseeducator, also involvedexercise physiologist,dietitian, group facilitator andphysician
Sessions: monthly sessionsfor 1 h followed by amoderately paced aerobicexercise session.
Delivery: group intervention,in person
Treatment changes: unclear
Training of trainers:
Theory: health promotionmodel ‘proceed-precede’(ref. given)
Control intervention:usual treatment withassessment visits at baseline,6 and 12 months and routineclinic visits
Duration of intervention: 6months (after programmeexercise sessions stillavailable to interventiongroup)
Eligibility/exclusion criteria:Type 2 diabetes, aged 40–70years, performing less than 1 h exercise per week.Excluded if history or signs ofischaemic heart disease,current smoker, poorcomprehension of English
How selected:endocrinologists completedquestionnaires on all theirpatients 40–70 years old atroutine clinic for 2 months
Numbers involved: N = 26(intervention 13, control 13)
Nos on insulin: intervention3, control 4
Sulphonylurea: intervention5, control 5
Metformin or diet alone:intervention 5, control 4
Type of diabetes: 2
Duration of diabetes: notreported
Baseline measurements ofoutcome parameter (mean± SE):
HbA1c: intervention 5.6% ±0.3, control 6.8% ± 0.6 (notsignificant)
BMI: intervention 32.3 ±1.1, control 35.7 ± 1.6
Weight: intervention 83 ±3.6, control 98.2 ± 3.4
Skinfolds: intervention 99.4± 6.0, control 119.4 ± 9.4
% Body fat: intervention 40.3± 1.7, control 40.3 ± 2.4
Waist:hip: intervention 0.94± 0.1, control 0.94 ± 0.08
Activity score: intervention164 ± 28, control 168 ± 16
Total cholesterol:intervention 5.6 ± 0.3,control 5.6 ± 0.2
HDL cholesterol:intervention 1.1 ± 0.1,control 1.1 ± 0.1
Triglycerides: intervention 3.1± 1.1, control 2.3 ± 0.3
Primary outcomes used:HbA1c, QoL (SF-36)
Secondary outcomes used:BMI
Individual preferred learningstyle addressed?: no
Any sub-groups (e.g. ethnicgroups): those managed withmetformin or diet alone andthose taking sulphonylurea orinsulin therapy
Normal range(s) foroutcomes: not reported
How outcomes assessed:physiological measures lab.,QoL self-report, activity =meter
Validated?: HbA1c yes, QoLby SF36: yes.
Timing of outcomes same forboth groups: yes
Length of follow-up: 12months from baseline
continued
Health Technology Assessment 2003; Vol. 7: No. 22
155
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Methodological commentsAllocation to treatment groups: no details of method of randomisationBlinding of outcome assessors?: not reportedAllocation concealment?: not reportedAnalysis by ITT?: no drop-outs reportedComparability of treatment groups: weight significantly higher, BMI and skinfolds marginallysignificantly higher in control group at baselineMethod of data analysis: ANOVA and Mann–Whitney statistics employed. Standard deviation given insome cases. No confidence intervals givenSample size/power calculation: not reportedAttrition/drop-out: not reported
Reference and design Intervention Participants Outcome measures
Outcome Intervention group Control group Difference between groups(values are changes from baseline, mean ± SE)
HbA1c (%) +0.86 (0.29) +0.86 (0.27) NS
QoL No data presented
BMI –0.1 (0.5) +0.29 (0.45) NS
Weight (kg) +0.14 (1.09) +0.79 (1.09) NS
Skinfolds +6.18 (2.2) –3.7 (4.8) NS
% Body fat +1.2 (0.5) +1.1 (0.9) NS
Waist:hip –0.02 (0.02) +0.01 (0.001) NS
Activity score (Mets) +1 (12) –23 (11) NS
Total cholesterol (mmol/l) –0.22 (0.27) –0.33 (0.18) NS
HDL cholesterol (mmol/l) –0.01 (0.04) –0.07 (0.04) NS
Triglycerides (mmol/l) –0.46 (1.02) –0.23 (0.23) NS
FBG (mmol/l) +0.97 (0.64) +1.5 (0.98) NS
Fasting insulin –3.3 (3.5) +1.5 (2.2) NS
Subgroup: metformin or +0.4 ± 0.3 +1.5 ± 0.14 p = 0.02diet alone HbA1c (changes from baseline)
Subgroup: metformin or +1.1 ± 0.3 +3.1 ± 0.4 p = 0.003diet alone FBG (changes from baseline
FBG: intervention 9.3 ± 1.0,control 7.9 ± 0.7
Fasting insulin: intervention22.4 ± 4.1, control 21.4 ±2.2
Gender (M/F): intervention4/9, control 6/7
Age ranges (years):intervention 60.5= (SE 7.8),control 60.5= (SE 2.1)
Ethnic groups: not reported,varied cultural backgrounds
Losses to follow-up: assumenil
Compliance: full
Appendix 8
156
General commentsGeneralisability: small sample size, smokers excludedConflict of interests: funding support not mentionedOther:
Quality criteria (CRD Report 4)
1. Was the assignment to the treatment groups really random? Unknown2. Was the treatment allocation concealed? Unknown3. Were the groups similar at baseline in terms of prognostic factors? Reported4. Were the eligibility criteria specified? Yes5. Were outcome assessors blinded to the treatment allocation? Unknown6. Were the point estimates and measure of variability presented for the primary outcome measure? Partial7. Did the analyses include an ITT analysis? No losses reported8. Were withdrawals and drop-outs completely described? No losses reported
Health Technology Assessment 2003; Vol. 7: No. 22
157
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Reference and design Intervention Participants Outcome measures
Surname and year: Wing et al., 198841
Source: published
Country: USA
Setting: unclear
Language: English
Trial design: RCT
Common procedure toboth groups: weight controlprogramme. Participated in alecture-discussion onbehavioural weight control,given individualised caloriegoals and recorded all intake.Taught about caloric valuesof food groups and trained inportion size estimation.Exercise (walking) wasstressed, and given graduallyincreasing exercise goals.Other lessons focused onbehavioural strategies forcontrolling cues for eating,dealing with social situationsinvolving food, changingcognitions about food,motivation and self-reinforcement and problemsolving. Deposited money atstart, and refunded for everypound of weight lost and forattending
Both groups given freeglucometers and asked tomonitor BG 12 times/week.Trained in its use
Intervention 1: self-regulation education:
Topics: extensive training inhow to use SMBGinformation; this info. wasgiven gradually over thecourse of the programme.Meetings 1–5 givenhomework tasks todemonstrate the effect ofdiet and exercise on BGcontrol, and given examples,these were then discussed atlater group meetings.Meetings 6–9 given goals forBG which were good andfair. Monitored how manywithin each range. Thentaught to use the readings toself-regulate their behavioursusing reinforcement.Meetings 10–13 refundeddeposit money for behaviourchanges and other criteriaused in previous phases. Notasked to adjust treatments inresponse to SMBG
Provider:
Eligibility criteria: >20%overweight, 30–65 years,met NDDG (1979) criteriafor Type 2
How selected: newspaperadvertisements used torecruit
Numbers involved: total N = 20, intervention 1 =10,intervention 2 = 10
Nos on insulin: none
Tablets: 16
Diet alone: 4
Type of diabetes: 2
Duration of diabetes: notreported
Baseline measurements ofoutcome parameter (mean± SE):
HbA1: intervention 1 10.57± 0.44%, intervention 210.54 ± 0.55%
BMI: 35.4 ± 1.05
Gender (M/F): 7, 13
Age ranges (years): average53.3 (range 38–60).
Ethnic groups: not reported
Losses to follow-up: 3 intotal, 1 in intervention 1, 2 inintervention 2 (1 death, 2refusals)
Compliance: all attended all16 weeks
Primary outcomes used:HbA1
Secondary outcomes used:BMI
Individual preferred learningstyle addressed?: no
Any sub-groups (e.g. ethnicgroups): no
Normal range(s) foroutcomes: HbA1 6.1 ±0.5%
How outcomes assessed?:lab.
Validated?: yes
Timing of outcomes same forboth groups: yes
Length of follow-up: 68weeks from inception
continued
Appendix 8
158
Methodological commentsAllocation to treatment groups: not describedBlinding of outcome assessors?: not described – not relevant for HbA1
Allocation concealment?: not describedAnalysis by ITT?: noComparability of treatment groups: no report of any differences in baseline, many characteristicsreported per total N only.Method of data analysis: ANOVA for repeated measures of the two treatment groups pretreatment and 1year. Standard error of mean reportedSample size/power calculation: not reportedAttrition/drop-out: percentages reported
General commentsGeneralisability: self-selected sampleConflict of interests: Biodynamics supplied glucometers and strips for SMBGOther:
Reference and design Intervention Participants Outcome measures
Outcome (mean ± SE) Intervention group 1 Intervention group 2 Difference between groups(n = 9) (n = 8)
HbA1 (%) 10.8 ± 0.8 9.71 ± 0.78 Time × condition interaction, NS(based analysis on baseline of those attending for follow up)
Weight (kg) (BMI not 86.6 ± 5.6 94.8 ± 5.9 Time × condition interaction, NSreported at follow-up) (based analysis on baseline of
those attending for follow-up)
Sessions: 13 sessions
Delivery: in person
Treatment changes:treatment changes in bothgroups monitored byphysician and followedstandard algorithm
Training of trainers:
Theory:
Intervention 2: self-monitoring education:
No additional training inusing SMBG information (asintervention 1 group had).
Duration of bothinterventions: 13 meetingsover 16 weeks (held weeklyfor 10 weeks and every 2 weeks for the following 6 weeks). Follow-upmeetings held every 2 weeksfor the next 3 months and atmonthly intervals for thefollowing 3 months. 10months total.
Were care programmesidentical: unclear
Health Technology Assessment 2003; Vol. 7: No. 22
159
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Quality criteria (CRD Report 4) RCTs
1. Was the assignment to the treatment groups really random? Unknown2. Was the treatment allocation concealed? Unknown3. Were the groups similar at baseline in terms of prognostic factors? Reported4. Were the eligibility criteria specified? Yes5. Were outcome assessors blinded to the treatment allocation? Not applicable6. Were the point estimates and measure of variability presented for the primary outcome measure? Partial7. Did the analyses include an ITT analysis? Inadequate8. Were withdrawals and drop-outs completely described? Partial
Appendix 8
160
Reference and design Intervention Participants Outcome measures
Surname and year: Gillilandet al., 200242
Source: published
Country: USA
Setting: community
Language: English
Trial design: CCT (3 groups)
Intervention 1: family andfriends (FF):
Topics: culturally appropriatediabetes education materials,skill building, social support.Three core areas: exercise,diet and support. Sessionsnamed: get more exercise;eat less fat; eat less sugar;together we can (how toget/receive support); stayingon the path (maintenance oflifestyle changes)
Intervention used NativeAmerican values, NativeAmerican foods, informationon diet and exercise andvideos featuring NativeAmericans. Consistent withNative American learning,stories and prayers wereused. There were writtenmaterials, as well as food andphysical activitydemonstrations. Activities toencourage discussion andsharing of stories about livingwith diabetes. Group physicalactivities and shared healthymeal
Provider: mentor led
Sessions: 5 sessions, ~6weeks apart for ~2 h
Delivery: in person in groupswith family and friends
Treatment changes:
Training of trainers: bilingualcommunity mentors trainedon each session
Theory: social learning theory
Intervention 2: one-on-one (OO):
Same written materials asgiven to FF but in individualsessions for approximately45 minutes
Control: usual care (UC):usual schedule of clinic visitsand activities. All participantsreceived comprehensivediabetes care includingprofessional and patienteducation. This group did notreceive culturally specificintervention materials
Eligibility criteria: all NativeAmerican women and men inlocal diabetes registries ≥18years old, mentally andphysically able and resided inone of 8 communities
How selected: placed intogroups by community ofresidence
Numbers involved: 104evaluable patients providedboth baseline and follow-updata (see below). 32 in FF;39 in OO; 33 in UC
Nos on insulin: total = 19: 2FF, 10 OO, 7 UC
Tablets: total = 63: 25 FF, 23OO, 15 UC
Diet alone: total = 22: 5 FF,6 OO, 11 UC
Type of diabetes: 2
Duration of diabetes (mean± SD): FF 8.1 (5.3), OO 8.3(6.4), UC 10.0 (6.6)
Baseline measurements ofoutcome parameter (mean± SD):
HbA1: FF 8.3 (1.9), OO 9.2(2.3), UC 7.9 (2.0)
BMI: FF 31.0 (5.6), OO 31.2(6.8), UC 32.0 (6.1)
Weight (lb): FF 174.6 (35.4),OO 172.2 (37.2), UC 168.9(33.8)
Diastolic BP (mmHg): FF 80(9), OO 81 (12), UC 78 (10)
Cholesterol (mg/dl): FF 199(51), OO 218 (50), UC 193(43)
Triglycerides (mg/dl): FF 224(147), OO 290 (214), UC214 (154)
Sex (M/F); FF 9/23, OO10/29, UC 3/30
Age (years) (mean ± SD): FF60.2 (12.1), OO 59.9 (13.4),UC 60.2 (11.8)
Ethnic groups: all participantsNative American
Primary outcomes used:HbA1c, weight
Secondary outcomes used:diastolic BP, cholesterol,triglycerides
Individual preferred learningstyle addressed?: no
Any sub-groups (e.g. ethnicgroups): no
Normal range(s) foroutcomes: HbA1 notreported
How outcomes assessed:laboratory
Validated: yes
Timing of outcomes same forboth groups: yes
Length of follow-up: ~1 yearfrom inception
continued
Health Technology Assessment 2003; Vol. 7: No. 22
161
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Methodological commentsAllocation to treatment groups: by communityBlinding of outcome assessors: not reported, not of concern for laboratory measuresAllocation concealment: N/AAnalysis by ITT?: noComparability of treatment groups: at baseline groups differed in HbA1c, in number of patients receivingOHAs, in hypertension. These differences were incorporated into statistical analysesMethod of data analysis: ANOVA for continuous variables, chi-squared or Fisher’s exact tests for discretevariables. ANCOVA for intervention differences in HbA1c and weight. Covariates were sex, age, durationof diabetes, medication use, two preintervention determinations of annual change in HbA1c and factorssignificantly different at baselineSample size/power calculation: none reported. Study size likely underpowered to detect differences in twointerventionsAttrition/drop-out: More women than men and more obese than non-obese participants were evaluableParticipants in usual care were more likely to drop-out
General commentsGeneralisability: Compared with the overall population of diabetic patients in the included communitiesthe patients who were evaluable seem generally representative. However, the evaluable patients weremore likely to be women and older. Relatively high drop-out rate is a concern for generalisabilityConflict of interests: none reportedOther:
Reference and design Intervention Participants Outcome measures
Outcome (mean ± SD) FF intervention OO intervention Control – usual care Difference between group group groups (across 3 arms)
HbA1 adjusted mean +0.5 (0.3) +0.2 (0.3) +1.2 (0.4) p < 0.05change Combined interventions vs
control, p < 0.05
Weight (lb) –2.0 1.5) –1.8 (1.5) +1.7 (1.8) NS
Combined interventions vs control, p = 0. 05
Diastolic BP (mmHg) –6.5 (2.0) –0.4 (1.7) –0.3 (2.1) p < 0.05Combined interventions vs control, NS
Cholesterol –22 (11) –20 (11) –10 (16) NSCombined vs control, NS
Triglycerides –178 (78) –48 (48) –69 (63) NSCombined vs control, NS
Duration of bothinterventions: sessionsconducted during 10 monthperiod
Were care programmesidentical: yes
Losses to follow-up: 206volunteered to participate,47 withdrew beforereceiving intervention, 42dropped out duringintervention, 13 did not haveinformation on covariates,104 were evaluable
Compliance: all evaluablepatients received fullintervention
Appendix 8
162
Quality criteria (CRD Report 4) CCTs
Were the groups similar at baseline in terms of prognostic factors? ReportedWere the eligibility criteria specified? YesWere outcome assessors blinded to the treatment allocation? UnknownWere the point estimates and measure of variability presented for the primary outcome measure? AdequateDid the analyses include an ITT analysis? InadequateWere withdrawals and drop-outs completely described? PartiallyWere participants likely to be representative of the intended population? No
Health Technology Assessment 2003; Vol. 7: No. 22
163
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Appendix 9
Data extraction: patients with either Type 1 or Type 2 diabetes
Reference and design Intervention Participants Outcome measures
Surname and year:Bloomgarden et al., 198744
Source: published
Country: USA
Setting: diabetes clinic at ateaching hospital
Language: English
Trial design: RCT
Treatment intervention:
Providers: nurse educatorand nutritionist
Topics: understandingdiabetes, foot skin and dentalhygiene, insulinadministration, emergencies,risk factors for macrovasculardisease.
Nutrition sessions covered:individual diet instruction,basic nutrition, weight lossand the diabetic diet, foodpurchasing and meal planning
Sessions:
Treatment changes:
Training of trainers:
Theory:
Mode: usual care plus 9group education sessionsoffered to each patient.Separate sessions in Spanish.
Used card games, films andslides
Control intervention:
Usual care – available to allpatients in both groups.Patients had contact at eachvisit with their physician anda nurse who reviewedmedications and specificproblems
Duration of intervention:
Programme lasted 1.6 ± 0.3years in education group and1.5 ± 0.3 years in controlgroup
Eligibility: all insulin-treatedpatients. None were excludedby design of the study
How selected: all insulin-treated diabetics on the clinicregistry as of September 1979
Numbers involved: 345agreed to participate
302 returned for examinationby physician: 145 in educationgroup, 157 in control group
Nos on insulin: all
Type of diabetes: 2,intervention 76%, control =65%
Duration of diabetes (years)(± SD): intervention 13 ± 8,control: 14 ± 9
Baseline measurements ofoutcome parameter (mean± SD):
HbA1c: intervention 6.8 ±2.1, control 6.6 ± 2.0
Knowledge: intervention 5.3± 1.6, control 5.3 ± 1.7
Gender (M/F): intervention50/77, control 72/67
Age ranges (years) (± SD):intervention 56 ± 12,control 59 ± 13
Ethnic groups: interventionwhite = 6%; black = 41%,Hispanic = 31%; controlwhite = 6%; black = 29%,Hispanic = 35%
Losses to follow-up: 79 (38in intervention and 41 incontrol). 345 agreed to
Primary outcomes used:HbA1c
Secondary outcomes used:development of foot lesions,diastolic and systolic BP inhypertensive subgroup, useof medical care, BMI, footlesions scores, FBG,behaviour score,triglycerides, HDL and LDLcholesterol, insulin dosage
Individual preferred learningstyle addressed: not stated
Any sub-groups: graduatesand non-graduates.Hypertensives
Normal range(s) foroutcomes: 1.8–4.8% totalHb.? Knowledge andbehaviour score
How outcomes assessed:knowledge and behaviourscores derived from previousliterature
Validated: knowledge score –possible?
Timing of outcomes same forboth groups: longer ineducation group by 1 month
Length of follow-up same asduration of intervention: 1.6± 0.3 years in educationgroup and 1.5 ± 0.3 years incontrol group
continued
Appendix 9
164
Methodological commentsAllocation to treatment groups: method of randomisation not statedBlinding of outcome assessors: not statedAllocation concealment: not statedAnalysis by ITT: noComparability of treatment groups: control group had more frequent foot lesions; education group hadhigher FBG and number of hospitalisations in previous yearMethod of data analysis: hypothesis tests (t-test, ANOVA). Standard deviations and p values givenSample size/power calculation: yes. Large enough to detect a difference in means between the groups inHbA1c of >1% with α = 0.5 and a power of 0.95Attrition/drop-out: reported
General commentsGeneralisability: no participants tended to be older (>70), required assistance to travel to the clinic, morelikely to be maleConflict of interests: none mentionedOther:
Quality criteria (CRD Report 4)
Reference and design Intervention Participants Outcome measures
Outcome (mean ± SD) Intervention group Control group Difference between groups(n = 127) (n = 139)
HbA1c 6.1 ± 2.0 6.3 ± 2.0.
Knowledge score 5.8 ± 1.6 5.3 ± 1.7 p < 0.007
BMI Men: 29.1 ± 4.6 Men: 27.7 ± 4.3Women: 32.1 ± 6.9 Women: 32.9 ± 7.0
Glucose (mg/dl) 179 ± 73 185 ± 76
Foot lesions (none/minor/severe) 61/56/10 48/75/16
Behaviour score 4.3 ± 1.6 4.1 ± 1.6
No differences between groups for sick days, hospitalisations, emergency room visits, outpatient visits, triglycerides, HDLand LDL cholesterol, insulin dosage (data not shown). No differences in HbA1c in those attending ≥ 7 sessions and those <7 sessions. Among hypertensive patients, no differences between groups (no data shown).
participate and 266 completedfinal assessment n = 127intervention, 139 control
Compliance: of the 145patients in the interventiongroup, 82 attended at least 7classes and regarded asgraduates of the programme;20 attended 3–6 classes, 30attended 1–2 classes and 17failed to attended any
1. Was the assignment to the treatment groups really random? Unknown2. Was the treatment allocation concealed? Unknown3. Were the groups similar at baseline in terms of prognostic factors? Reported4. Were the eligibility criteria specified? Yes5. Were outcome assessors blinded to the treatment allocation? Unknown6. Were the point estimates and measure of variability presented for the primary outcome measure? Adequate7. Did the analyses include an ITT analysis? Unknown8. Were withdrawals and drop-outs completely described? Partially
Health Technology Assessment 2003; Vol. 7: No. 22
165
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Reference and design Intervention Participants Outcome measures
Surname and year: Glasgowet al., 199745
Source: published
Country: USA
Setting: in the office of 2internists who are primarycare providers and part of alarge medical group
Language: English
Trial design: RCT
Treatment intervention:
Providers: researcher seenafter physician visit
Topics: patient-centred goalsetting and problem solving,and dietary self-helpmaterials. Producedindividualised goal-settingplan to lower fat intakebased on patients’ eatinghabits and barriers to dietaryself-management. Patientswith higher self-efficacyscore received a take-homevideo. Patients with lowerefficacy levels returned for a30-minute interactive video,more personalised
Sessions: 20 minute initially,then telephone follow-up at1 and 3 weeks, and 3 and 6months to review progress,adjust strategies and mailmaintenance information. At9 months received a copy ofa book ‘On the human sideof diabetes’. Interventiondelivered by research staff
Treatment changes:
Training of trainers:
Theory:
Mode: an additional 5–10minute touchscreen dietarybarriers assessment whichgenerated immediatefeedback forms
Control intervention:Usual care = quarterlymedical care (regularassessment and follow-up,plus the initial touchscreencomputer assessment),telephone contact; 3 weeks,6 months, given book at 9months
Duration of intervention:9 months
Eligibility criteria: having Type1 or 2 diabetes, at least 40years old, being primarilyresponsible for one’s owndiabetes dietary self-management
How selected: thosescheduled for visit received aletter encouragingparticipation. Randomisedfrom the physician practice
Numbers involved: 206 totalintervention N = 108,control N = 98
Nos on insulin: intervention68%, control 66%
Type of diabetes: 2,intervention 76%, control81%
Duration of diabetes (years):intervention 13.0 (9.9),control 13.7 (12.2)
Baseline measurements ofoutcome parameter:
HbA1c: intervention 7.9,control 7.9
Food questionnaire:intervention 2.26, control2.20
BMI: intervention 30.4,control 30.5
Cholesterol: intervention217, control 223
Gender (M/F) (%):intervention 37/63, control40/60
Age ranges (years):intervention 61.7 (SD 12.1),control 63.1 (SD 10.5)
Ethnic groups: not given
Losses to follow-up: 16%(16.7 vs 15.3)
Compliance: assume 100%
Primary outcomes used:HbA1c
Secondary outcomes used:patient satisfaction, BMI,dietary self-managementquestionnaire, serumcholesterol
Individual preferred learningstyle addressed: no
Any sub-groups (e.g. ethnicgroups): no
Normal range(s) foroutcomes: not given
How outcomes assessed:Patient Satisfactioninstrument contained 7 itemsassessing the office visit.Food Habits Questionnaire(FHQ) measuring fourdimensions of fat-relateddietary habits
Validated: the Kristal FHQ isvalidated. Patient SatisfactionMethods was developed forthis study (not reported)
Timing of outcomes same forboth groups: yes
Length of follow-up: 12months from inception
continued
Appendix 9
166
Methodological commentsAllocation to treatment groups: randomised using a table of random numbersBlinding of outcome assessors: not statedAllocation concealment: not statedAnalysis by ITT: noComparability of treatment groups: well matched – no significant differences on any variablesMethod of data analysis: a series of MANCOVA and ANCOVAs to identify specific measures on whichthere were treatment effects. p values given. No measure varianceSample size/power calculation: noAttrition/drop-out: 16% – no differences between groups
General commentsGeneralisability: 61% of eligible patients (those that had scheduled an outpatient visit) agreed to participateConflict of interests: (funding support mentioned?)Other: Costs for the Brief Intervention were $137 per participant. As there were no significant effects onHbA1c an economic analysis not conducted. ? normal range for diet questionnaire. Different Ns fordifferent outcomes
Quality criteria (CRD Report 4)
Outcomes Intervention group Control group Difference between groups
HbA1c levels (N = 161) 7.8 7.8
BMI (N = 164) 30.5 30.4
Serum cholesterol (N = 167) 208 226 p < 0.01
Food Habits Questionnaire (FHQ) (n = 170) 2.06 2.26 p < 0.01
1. Was the assignment to the treatment groups really random? Adequate2. Was the treatment allocation concealed? Unknown3. Were the groups similar at baseline in terms of prognostic factors? Reported4. Were the eligibility criteria specified? Yes5. Were outcome assessors blinded to the treatment allocation? Unknown6. Were the point estimates and measure of variability presented for the primary outcome measure? Partially7. Did the analyses include an ITT analysis? Unknown8. Were withdrawals and drop-outs completely described? Unknown
Health Technology Assessment 2003; Vol. 7: No. 22
167
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Reference and design Intervention Participants Outcome measures
Surname and year: Raji et al.,200246
Source: published
Country: USA
Setting: unclear
Language: English
Trial design: RCT
Intervention 1 (intensivegroup education):
Topics: core elementsrecommended by ADA (seeref.)
Provider: physician, nurse,nutritionist, pharmacist,exercise physiologist, socialworker, diabetes educator
Sessions: 3.5 days
Treatment changes:
Training of trainers:
Theory:
Mode: structured curriculumthrough lectures, groupdiscussions and supervisedexercise.
Two meals and snacksprovided to reinforce thenutritional instruction.
Patients then returned tousual care.
Groups of 4–6 participants
Intervention 1 (passiveeducation):
Topics: general diabetesmanagement, nutrition,coronary artery disease, footcare
Provider:
Sessions:
Treatment changes:
Training of trainers:
Theory:
Mode: educational materialsmailed to participants, homesevery 3 months, for 12months including booklets(15–45 pages)
Control intervention: alsoused data from another 56matched patients from thosewho declined randomisation,but not randomised.Matched on age, sex andbaseline HbA1c to passivegroup. Hb measured at 12 ±3 months from screening
Duration of intervention:intervention 1 3.5 days,intervention 2 once every 3months for 12 months
Eligibility criteria: elevatedHbA1c (>8.5%) within 30days randomisation, ≥ 18years, able to exercise,available to participate, ableto understand written andspoken English.
Excluded if: significant eyedisease limiting visual acuity,urine protein >2g/dl,coronary artery diseasesymptoms and/or lowerextremity amputation thatlimited exercise capacity
How selected: hospital lab.data screened for patientswith HbA1c >8.5%
Numbers involved: 106(intervention 1: 50,intervention 2: 56)
Nos on insulin: 39%
Tablets: 46%
Combination: 15%
Type of diabetes: notreported, assume mixed
Duration diabetes: notreported
Baseline measurements ofoutcome parameter: HbA1cintervention 1: 10%,intervention 2: 9.9%
Gender: 99% male
Mean age (years) : 60 ± 3
Ethnic groups: not reported
Losses to follow-up:intervention 1, 1 (no reasongiven); no report forintervention 2
Compliance: not reported
Primary outcomes used:HbA1c
Secondary outcomes used:numbers on oral medication,insulin, combination
Individual preferred learningstyle addressed? no
Any sub-groups: no
Normal range(s) foroutcomes: not reported
How outcomes assessed?:HbA1c, lab., assumemedication from patientrecords
Validated?: yes
Timing of outcomes same forboth groups: yes
Length of follow-up frominception: 12 months
continued
Appendix 9
168
Methodological commentsAllocation to treatment groups: not reportedBlinding of outcome assessors?: not applicableAllocation concealment?: not reportedAnalysis by ITT?: yesComparability of treatment groups: baseline characteristics only given for total group, except HbA1c
which was not differentMethod of data analysis: point estimates only, used hypotheses testsSample size/power calculation: not reportedAttrition/drop-out: reported for intervention 1 only, assume nil for intervention 2
General commentsGeneralisability: majority maleConflict of interests: from research bodies, not commercialOther: method of recruitment means high motivated patients. Possible that medical therapy intensifiedduring trial
Quality criteria (CRD Report 4) RCTs
Outcome Intervention 1 group Intervention 2 group Non-randomised Difference control between groups
HbA1c 8.0 8.0 NS
HbA1c joint intervention 8.0 (SD 1.4) 8.6 (SD 1.8) p < 0.03versus control
Numbers on medication: Pie chart only with no Oral monotherapy indication of gaugeOral combinationInsulin/oral combinationinsulin
1. Was the assignment to the treatment groups really random? Unknown2. Was the treatment allocation concealed? Unknown3. Were the groups similar at baseline in terms of prognostic factors? Unknown4. Were the eligibility criteria specified? Yes5. Were outcome assessors blinded to the treatment allocation? Unknown6. Were the point estimates and measure of variability presented for the primary outcome measure? Inadequate7. Did the analyses include an ITT analysis? Adequate8. Were withdrawals and drop-outs completely described? Reported
Health Technology Assessment 2003; Vol. 7: No. 22
169
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Reference and design Intervention Participants Outcome measures
Outcome (mean ± Group A (education Group B (education) Group C (control) Difference between SEM) 2 years and support) groups
HbA1c 6.6 ± 0.3 6.5 ± 0.2 8.4 ± 0.7*a *from group A, p < 0.05afrom group B, p < 0.05
Knowledge 38 ± 1 36 ± 1* 34 ± 1* *from group A, p < 0.05
QLa 26 ± 1 25 ± 1 23 ± 1** **from Group A, p < 0.01
Surname and year: Gilden et al., 199247
Source: published
Country: USA
Setting: diabetes clinic,Veterans Affairs MedicalCenter
Language: English
Trial design: CCT, threegroups matched for age andduration of diabetes
Group A: education andsupport groupeducation:
Providers: diabetologist,nurse educator, dietitian,social worker, psychologist,podiatrist, pharmacist
Topics: general knowledge ofdiabetes, nutritional and drugmanagement, social worksupport services, stressmanagement, self-caretechniques (SMBG, generalhealth habits, foot care)
Sessions: six, one per week
Support group:
Provider: self-directed bypatients but additionaleducation by providers listedabove
Topics: continuing education,coping skills, groupdiscussion, structured socialactivities
Sessions: monthly for 18months (ref. 2 describeseducation)
Treatment changes:
Training of trainers:
Theory:
Mode:
Group B: education onlyas described above
Group C: no intervention
Eligibility/exclusion criteria:no eligibility criteria reported
How selected: attendedsame clinic
Numbers involved: total: 32
Group A (treatment): 11
Group B (treatment): 13
Group C (control): 8
Nos on insulin: not reported
Tablets: not reported
Diet alone: not reported
Type of diabetes: notreported
Duration of diabetes (years):10 ± 2, range 3–6
Baseline measurements ofoutcome parameter (GroupA only; mean ± SEM):
Knowledge: 36 ± 4 QoL:QLa 22 ± 2, QLb 38 ± 10,QLt 62 ± 13
Stress: 12 ± 3
Family involved: 28 ± 5
Social activity: 9 ± 4
Gender: all male
Age ranges (years): mean 68±1.3 (SEM), range 57–82
Ethnic groups: not reported
Losses to follow-up: notreported
States patients not includedin analysis if participants inother education programmesduring 2 years, suggestsnumbers may not be correct
Primary outcomes used:HbA1c, QoL (two subscales,QLa, QLb)
Secondary outcomes used:knowledge, stress, family involvement, socialactivities, depression (Zung’sMood Scale)
Individual preferred learningstyle addressed?: No
Any sub-groups (e.g. ethnicgroups): no
Normal range(s) foroutcomes: HbA1c 3.0–6.1%,depression 25–50 = normal,others not reported
How outcomes assessed: allmeasures except HbA1cwere self-report
Validated: questionnairesreported as validated(reference given)
Timing of outcomes same forboth groups: yes
Length of follow-up: 2 yearsfrom programme inception
continued
Appendix 9
170
Methodological commentsAllocation to treatment groups: matched for age and diabetes duration. No additional informationBlinding of outcome assessors: not reportedAllocation concealment: not reportedAnalysis by ITT: not reportedComparability of treatment groups: “no significant differences on questionnaire variables between GroupsA and B prior to support group intervention”. No other informationMethod of data analysis: t-tests and ANOVA, SEM, no confidence intervalsSample size/power calculation: no power calculation reportedAttrition/drop-out: no information on attrition
General commentsGeneralisability: no information about inclusion criteria. Owing to small groups, setting and all maleparticipants, generalisability may be limitedConflict of interests: no mention
Quality criteria (CRD Report 4) CCTs
Outcome (mean ± Group A (education Group B (education) Group C (control) Difference between SEM) 2 years and support) groups
QLb 53 ± 5 45 ± 5 41 ± 2** **from Group A, p < 0.01
QL total 78 ± 5 71 ± 6* 64 ± 3** *from group A, p < 0.05**from Group A, p < 0.01
Stress 14 ± 1 14 ± 1 11 ± 1* *from group A, p < 0.05
Family involvement 26 ± 1 28 ± 3** 24 ± 2* *from group A, p < 0.05**from Group A, p < 0.01
Social activities 8 ± 1 10 ± 1 12 ± 1** **from Group A, p < 0.01
Depression 43 ± 6 51 ± 3 56 ± 2(higher = more depression)
Pervasive affective 2.3 ± 0.2 2.7 ± 0.2* 3.4 ± 1** *from group A, p < 0.05disturbance **from Group A, (higher = more p < 0.01depression)
QLa = more demanding and intensive life-style changes due to diet, exercise and other general factors. QLb = lessdemanding behaviours including medication compliance and self-testing. Higher scores indicate better knowledge and betterperception of QoL.
Were the groups similar at baseline in terms of prognostic factors? UnknownWere the eligibility criteria specified? NoWere outcome assessors blinded to the treatment allocation? UnknownWere the point estimates and measure of variability presented for the primary outcome measure? AdequateDid the analyses include an ITT analysis? UnknownWere withdrawals and drop-outs completely described? UnknownWere participants likely to be representative of the intended population? No
Health Technology Assessment 2003; Vol. 7: No. 22
171
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Albano MG, Jacquemet SP. Patient education anddiabetes research: a failure. Going beyond theempirical approaches. Acta Diabetol 1998;35:207–14.
Beebe C, O’Donnell M. Educating patients with type 2diabetes. Nurs Clin North Am 2001;36:375–86.
Brown SA. Effects of educational interventions indiabetes care: a meta-analysis of findings. Nurs Res1988;37:223–30.
Brown SA. Studies of educational interventions andoutcomes in diabetic adults: a meta-analysis revisited.Patient Educ Couns 1990;16:189–215.
Brown SA. Meta-analysis of diabetes patient educationresearch: variations in intervention effects acrossstudies. Res Nurs Health 1992;15:409–19.
Brown SA, Hedges LV. Predicting metabolic control indiabetes: a pilot study using meta-analysis to estimatea linear model. Nurs Res 1994;43:362–8.
Clement S. Diabetes self-management education.Diabetes Care 1995;18:1204–14.
Corabian P, Harstall C. Patient diabetes education in themanagement of adult type 2 diabetes. AlbertaHeritage Foundation for Medical Research; 2001. No.HTA 23: Series A.
Diabetes UK. Patient education for effective diabetesself-management. London: Diabetes UK; 2002.
Eakin EG, Bull SS, Glasgow RE, Mason M. Reachingthose most in need: a review of diabetes self-management interventions in disadvantagedpopulations. Diabetes Metab Res Rev 2002;18:26–35.
Fain JA, Nettles A, Funnell MM, Charron D. Diabetespatient education research: an integrative literaturereview. Diabetes Educ 1999;25:7–15.
Goodall TA, Halford WK. Self-management of diabetesmellitus – a critical review. Health Psychol 1991;10:1–8.
Griffin S, Kinmouth AL, Skinner C, Kelly J. Educationaland psychological interventions for adults withdiabetes. London: British Diabetic Association; 1998.
Huang ES, Meigs JB, Singer DE. The effect ofinterventions to prevent cardiovascular disease inpatients with type 2 diabetes mellitus. Am J Med2001;111:633–42.
Krishna S, Balas EA, Spencer DC, Griffin JZ, Boren SA.Clinical trials of interactive computerized patienteducation: implications for family practice. J FamPract 1997;45:25–33.
Montani S, Bellazzi R, Quaglini S, D’Annunzio G. Meta-analysis of the effect of the use of computer-basedsystems on the metabolic control of patients withdiabetes mellitus. Diabet Technol Ther 3:347–56.
Norris SL, Engelgau MM, Narayan KM. Effectiveness ofself-management training in type 2 diabetes: asystematic review of randomized controlled trials.Diabetes Care 2001;24:561–87.
Padgett D, Mumford E, Hynes M, Carter R. Meta-analysis of the effects of educational and psychosocialinterventions on management of diabetes mellitus. JClin Epidemiol 1988;41:1007–30.
Thomas N. Assessment: the crucial stage in diabeteseducation. Adv Clin Nurs 1999;3:67–74.
Whittemore R. Strategies to facilitate lifestyle changeassociated with diabetes mellitus. J Nurs Scholarsh2000;32:225–32.
Appendix 10
List of reviews and systematic reviews retrieved
Health Technology Assessment 2003; Vol. 7: No. 22
173
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Internal validity of economic evaluationsa
Appendix 11
Internal/external validity of economic evaluations
Item Kaplan et al., 198765 Glasgow et al., 199745
1. Well-defined question � �
2. Clear description of alternatives � �
3. Reasonable study type � ? Cost-effectiveness results not related to health outcomes
4. Effectiveness established Effectiveness data used from the Effectiveness data used from the study undertaken, with statistically study undertaken, with statistically significant differences in the significant differences in the outcomes measures used outcomes measures used
5. Estimates related to population risks ? ?
6. Relevant costs and consequences � Intervention costs � Intervention costsidentified
7. Costs and consequences � Costs for intervention based � Costs for intervention based measured accurately on resource use documented on resource use documented
in the trial reported in the trial reported.
� Consequences from trial data � Consequences from trial data
8. Costs and consequences valued � × Costs valued credibly, credibly consequences not
9. Differential timing considered Analysis within trial period only 1-Year analysis(11–18 months)
10. Incremental analysis performed � �
11. Sensitivity analysis performed � ×
12. Modelling conducted reasonably Modelling of health benefits only, ?from previous study
a? means unclear or unknown; � means item included or judged as acceptable to be internally valid; × means factor notincluded or judged unacceptable to be internally valid.
Appendix 11
174
External validity of economic evaluationsa
Item Kaplan et al., 198765 Glasgow et al., 199745
1. Patient group – are the patients in the ? Efficacy obtained from patients ? Efficacy obtained from patients study similar to those of interest in with NIDDM, in USA. Self-selecting having Type 1 or 2 diabetes, at England and Wales? patient group least 40 years old, being primarily
responsible for one’s own diabetes dietary self-management
2. Health care system/setting – × US health care provider setting × US primary care providers comparability of available alternatives?; forming part of a large medical similar levels of resources?; no group. Analysis from the untoward supply constraints?; perspective of the health care institutional arrangements comparable? organisation
3. Treatment – comparability with clinical ? Treatment in US hospital setting ? Treatment US centres.management?
4. Resource costs – comparability between × US cost data ? US cost datastudy and setting/population of interest?
5. Marginal versus average costs – what difference does this make? × ?
a? means unclear or unknown; � means judged item suitable to generalise to England and Wales with or without some re-adjustment; × means factor judged not suitable as either not possible to see how an adjustment could be made easily inshort/medium term or relevant data unavailable.
Health Technology Assessment 2003; Vol. 7: No. 22
175
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Methods – economic evaluation (see Appendix 9 for clinical data extraction)Base year prices: not statedPerspective: healthcare organisationCosts: included costs for intervention package (computer hardware, software), materials includinghandouts, pamphlets, supplies, labour costs for health educators, nurses, physicians and support staff,postage and telephone charges. Capital costs depreciated over year one. Did not include facility spaceand labour costs for training (of educators) (these were considered in sensitivity analyses)Outcomes: from the study undertaken and reported (see above and Appendix 9 for clinical detail)Discounting: no discounting undertaken (costs occurred within 1 year)
Results – economic evaluation (see Appendix 9 for clinical data extraction)Base case: costs for the delivery of the Brief Intervention were reported at US$137 per participant. Costswere combined with outcomes data on fat consumption, saturated fat consumption and serum cholesterol(there were no significant effects on HbA1c). The marginal cost per unit improvement in these outcomeswere:
$62 per reduction of each per cent in dietary fat
Appendix 12
Data extraction of economic evaluations
Reference and design Intervention Subjects Outcome measures
Surname and year: Glasgowet al., 199745
Source: published – patient educationand counseling
Country: USA
Setting: in the office of 2internists who are primarycare providers and part of alarge medical group
Language: English
Trial design: RCT
Economic evaluation/type:cost-effectiveness analysis
Treatment intervention:
Providers: researcher seenafter physician visit
Topics: patient-centred goalsetting and problem solvingand dietary self-helpmaterials
Sessions: 20 minutes initially,then telephone follow-up at1 and 3 weeks and 3 and 6months to review progress,adjust strategies and mailmaintenance information. At9 months received a copy ofa book ‘On the human sideof diabetes’. Interventiondelivered by research staff
Control intervention:Usual care = quarterlymedical care (regularassessment and follow-up,plus the initial touch-screencomputer assessment)telephone contact; 3 weeks,6 months, given book at 9months.
Duration of intervention:9 months
See data extraction form inAppendix 9 for further detail
Eligibility criteria: having Type1 or 2 diabetes, at least 40years old, being primarilyresponsible for one’s owndiabetes dietary self-management
How selected: thosescheduled for visit received aletter encouragingparticipation. Randomisedfrom the physician practice
Numbers involved: total N =206, intervention N = 108,control N = 98
Nos on insulin: intervention68%, control 66%
Type of diabetes: 2,intervention 76%, control81%
Duration of diabetes (years):intervention 13.0 (9.9),control 13.7 (12.2)
Compliance: assume 100%
See data extraction form inAppendix 9 for further detail.
Primary outcomes used:HbA1c
Secondary outcomes used:patient satisfaction, BMI,dietary self-managementquestionnaire, serumcholesterol
Individual preferred learningstyle addressed: no
How outcomes assessed:Patient Satisfactioninstrument contained 7 itemsassessing the office visit.FHQ measuring fourdimensions of fat-relateddietary habits.
Length of follow-up: 12months from inception
See data extraction form inAppendix 9 for further detail
Appendix 12
176
$105 per percentage reduction in saturated fat$8 per mg/dl reduction in serum cholesterol
Cost-effectiveness estimates were also presented for three different-sized potential patient groups, toreflect economies of scale (these were similar to the study estimates above)
Sensitivity analysis: not formally presented for the economic evaluation; however, authors do state thatwhere costs were set to include cost of facility space and training, whilst reducing equipment costs bydepreciating equipment costs over 3 years, increases the research model costs by 11% (and adissemination model cost by 1%)
Methodological comments – economic evaluation (see Appendix 9 for clinical data extraction)Outcomes used in cost-effectiveness analysis are intermediate outcomes and are not related by theauthors to health outcomes (e.g. events or complications of disease)
Sensitivity analysis is not reported formally for the economic evaluation.
Caveat: research staff delivered the intervention package
General commentsConflict of interests: project supported by the National Institute of Diabetes, Digestive and KidneyDiseases (USA)
Health Technology Assessment 2003; Vol. 7: No. 22
177
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Methods – economic evaluation (see Appendix 8 for clinical data extraction)Base year prices: 1986 clinical chargesPerspective: not stated (assume that of the healthcare provider)Costs: costs estimated using 1986 clinical charges in San Diego County, USA. Costs comprised: historyand physical examination, laboratory charges, charges for behaviour modification sessions and chargesfor medical consultations. No side-effects noted, so costing of these not undertakenOutcomes: QWB scores at initial interview and 3, 6, 12 and 18 months (QWB score reflects a mean valueover 4 days prior to assessment). QWB scores used to reflect outcomes in terms of well-years. Study usesQWB weights derived from community surveys to reflect social preference or utility (0 = dead to 1 =optimum function). Analysis does not cover any issues related to long-term complicationsDiscounting: no discounting reported
Results – economic evaluation (see Appendix 8, for clinical data extraction)
Base case: total costs for the programme are estimated at ~US$1000. Benefit stated as 0.092 well-years(see note below). Cost–utility estimate presented as $10,870 per well-year.
Sensitivity analysis: sensitivity analysis undertaken on effectiveness parameter, assuming 50% of benefitobserved, providing an estimate of $21,740 per well-year; sensitivity analysis undertaken on effectiveness,assuming benefits last for an additional year, providing an estimate of $5435 per well-year.
Methodological comments – economic evaluation (see Appendix 8 for clinical data extraction)
Based on observations from the reported experimental study.Calculation of benefits using the QWB is an indirect derivation of benefit, based on data reported andmodelled in a previous study. An appendix is presented at the end of the paper to detail the genericmethodological approach taken to derive ‘well-life’ scores via the QWB – no study-specific scores arediscussed in the appendixAnalysis does not cover any issues related to long-term complicationsPatient numbers across groups will be small, and patients in the trial were self-selectingUnsure of the numbers in each of the intervention and control groups
General commentsConflict of interests: funding support for one of the authors from National Institutes of Health, USA
Reference and design Intervention Subjects Outcome measures
Surname and year: Kaplan et al., 198735,65
Source: published – health promotion
Country: USA
Setting: unclear
Language: English
Trial design: RCT
Economic evaluation/type:cost utility analysis, alongsidean RCT
Patients randomly assigned toone of four experimentalconditions: diet, exercise,diet plus exercise oreducation control
See data extraction form inAppendix 8 for detail
Eligibility/exclusion criteria:confirmed diagnosis, fastingplasma glucose >3.62mmol/litre.
Numbers involved: Total N =87, unsure of group numbers
Nos on insulin: 19Tablets: 29Diet alone: 28
Type of diabetes: 2
Duration of diabetes: notrecorded
Gender (M/F): 32/44
Compliance: averageattendance >80% for allgroups
See data extraction form inAppendix 8 for further detail
Primary outcomes used:HbA1c, QoL
Secondary outcomes used:weight in kg
Individual preferred learningstyle addressed?:
How outcomes assessed?:HbA1c, laboratory; QoL, self-report questionnaire, theQWB
Length of follow-up: 18 months from inception
Health Technology Assessment 2003; Vol. 7: No. 22
179
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Appendix 13
Critical appraisal of health economic modelling studies in area of diabetes
Appendix 13
180
Title Model of complications Estimated benefits of Lifetime benefits and costs of The cost-effectiveness of of NIDDM. I Model glycaemic control in intensive therapy as practised different management construction and microvascular complications in the diabetes control and strategies for type I diabetes: assumptions in Type 2 diabetes complications trial a swiss perspective
Authors Eastman et al.75 Vijan et al.114 DCCT72 Palmer et al.73
Year 1997 1997 1996 2000
Modelling assessments should include:1 A statement of the problem Analysis of prevention strategies
for Type 2 diabetes usingmodelling
To evaluate the efficacy ofglycaemic control in Type 2diabetes patients
To examine cost-effectiveness ofalternative approaches to themanagement of Type 1 diabetes
The overall objective of this studywas to determine the healthoutcomes and economicconsequences of differentcombinations of diabetesinterventions in newly diagnosedpatients with Type 1 diabetes inSwitzerland
2 A discussion of the need formodelling vs alternativemethodologies
Implied by the lack of empiricaleconomic evidence though notstated directly
Implied by the lack of empiricaleconomic evidence though notstated directly
Implied by the lack of empiricaleconomic evidence, although notstated directly
Implied by the lack of empiricaleconomic evidence, although notstated directly
3 A description of the relevantfactors and outcomes
Factors included: disease incidenceand progression, hazard rates(dependent on age and clinicalfactors), ethnicity adjustments,mortality sub-model, CVD sub-model. Costs of screening,treatment and disability alsoincluded. This model covers end-stage disease progression. QALYssuggested
Factors included: model coversearly-stage complication only.Lifetime risk, absolute reduction inrisk for blindness covered, nocosts included
Factors included: mortalityincorporated within disease states.Costs of therapy (all direct medicalincluded) stated but not included.Also includes average years freefrom complications, cumulativeincidence, QALYs suggested.Model covers end-stage diseaseprogression
Factors included: cumulativeincidence, mortality incorporatedinto complication sub-models,end-stage disease progression(dependent on demographic andclinical factors). Costs of event +12-month follow-up. Lifeexpectancy and cost per life-yeargained also included as outcome.
continued
Health Technology Assessm
ent2003; Vol. 7: No. 22
181
© Q
ueen’s Printer and Controller of H
MSO
2003. All rights reserved.
Title Model of complications Estimated benefits of Lifetime benefits and costs of The cost-effectiveness of of NIDDM. I Model glycaemic control in intensive therapy as practised different management construction and microvascular complications in the diabetes control and strategies for type I diabetes: assumptions in Type 2 diabetes complications trial a swiss perspective
Authors Eastman et al.75 Vijan et al.114 DCCT72 Palmer et al.73
Year 1997 1997 1996 2000
Modelling assessments should include:4 A description of the model
including reasons for this typeof model and a specification ofthe scope including; timeframe, perspective,comparators and setting. Note:n = number of health stateswithin sub-model
3 complications + CVD:retinopathy (n = 5), neuropathy (n = 3), nephropathy (n = 4),CVD (n = 2). State transitionmodel used to simulate theprogression of Type 2 diabetespatients aged 25–74. Comparatorsused: conventional vs intensiveglycaemic control. Perspective:based on published data andMedicare reimbursement rates(1994 US$). Costed fromviewpoint of single payerresponsible for all direct medicalcosts. Costs and QALYsdiscounted at 5 and 7% per year
2 complications showing early-stage disease only: nephropathy (n = 5), retinopathy (n = 5). Statetransition model used to simulatethe progression of Type 2 diabetespatients aged 45–75 (assumed).Hypothetical drug used. No costs
3 complications modelled.Retinopathy(n = 5), Neuropathy(n = 3), Nephropathy (n = 4).State transition model used tosimulate the progression of type Idiabetes patients aged 13–39.Perspective: Healthcareperspective used for cost-effectiveness (all direct medicalcosts). 1994 US$. Both costs andeffects discounted at 3% per year.
7 complications modelled:neuropathy (n = 5), nephropathy(n = 10), retinopathy(n = 5),AMI(n = 8), stroke (n = 5),hypoglycaemia (n = 3),Ketoacidosis (n=3). Statetransition model used to simulatethe progression of male Type 1diabetes patients aged 19 years(Swiss median age at onset).Comparators used: conventionalinsulin therapy, screening, intensiveinsulin therapy and ACE inhibitorsused in combination. Perspective:Swiss health insurance payer. 1996Swiss CHF. Costs discounted at 3,5 and 6% per year
5 A description of data sources(including subjective estimates),with a description of thestrengths and weaknesses ofeach source, with reference toa specific classification orhierarchy of evidence
Progression rates and cohort:DCCT, WESDR, REP. All hazardrates are provided
Progression and cohort: DCCT,WESDR, REP
Progression rates and cohort:DCCT, WESDR
Progression rates and cohort:DCCT, published sources
Costs: published data and/orprevailing Medicarereimbursement rates
Costs: N/A Costs: resources based on DCCTtrial, Medicare reimbursement
Other: VA cooperative study,Metformin Cooperative Trial
Other: mortality retrieved from US Department of Vital Statistics
continued
Appendix 13
182
Title Model of complications Estimated benefits of Lifetime benefits and costs of The cost-effectiveness of of NIDDM. I Model glycaemic control in intensive therapy as practised different management construction and microvascular complications in the diabetes control and strategies for type I diabetes: assumptions in Type 2 diabetes complications trial a swiss perspective
Authors Eastman et al.75 Vijan et al.114 DCCT72 Palmer et al.73
Year 1997 1997 1996 2000
Modelling assessments should include:
continued
6 A list of assumptions pertainingto the structure of the model(e.g. factors included,relationships, and distributions)and the data
All major assumptionssystematically reviewed
All major assumptions addressedbut not in a systematic manner
All major assumptions addressedbut not in a systematic manner
All major assumptions addressedbut not in a systematic manner
7 A list of parameter values thatwill be used for a base caseanalysis, and a list of the rangesin those values that representappropriate confidence limitsand that will be used in asensitivity analysis
Disease progression rates derivedfrom DCCT and publishedsources. Certain prevalence ratesconsistent with WESDR
Rates of early disease based onDCCT findings. Cohort data usedfor rates of subsequentprogression to later disease.Incidence – DCCT,Microalbuminuria CollaborativeStudy and REP
Base-case rates of progressionretrieved from DCCT andpublished sources. Formulaeshown within literature
Base-case rates of progressionretrieved from DCCT andpublished sources. Non-exhaustivelist provided within the text
8 The results derived fromapplying the model for thebase case
Results derived from applying themodel to the base case aresystematically reported
Results derived from applying themodel to the base case aresystematically reported
Results derived from applying themodel to the base case aresystematically reported
Results derived from applying themodel to the base case aresystematically reported
9 The results of the sensitivityanalyses.Unidimensional; best/worstcase; multidimensional (MonteCarlo/parametric); threshold
Not described within the literature 3-way sensitivity analysisconsidering the impact ofimproved glycaemic control onlifetime risk for blindness. Mainconclusions hold true
Sensitivity analysis conducted toexamine the sensitivity of resultsto changes in incidence andprogression of complications.Decreasing the incidence ofmicroalbuminuria by 50% in theconventional group increased theincremental cost per life-yeargained to $79,883
1-way sensitivity analysis on allcost and probability parameterswas performed, varying oneparameter at a time by ±10%. 1-way sensitivity analysis showedthe annual cost of intensivetherapy had the greatest impacton the total lifetime costs.Reduced risk of AMI and incidenceand progression of MAU withintensive therapy had the greatestimpact on life expectancy
Health Technology Assessm
ent2003; Vol. 7: No. 22
183
© Q
ueen’s Printer and Controller of H
MSO
2003. All rights reserved.
Title Model of complications Estimated benefits of Lifetime benefits and costs of The cost-effectiveness of of NIDDM. I Model glycaemic control in intensive therapy as practised different management construction and microvascular complications in the diabetes control and strategies for type I diabetes: assumptions in Type 2 diabetes complications trial a swiss perspective
Authors Eastman et al.75 Vijan et al.114 DCCT72 Palmer et al.73
Year 1997 1997 1996 2000
Modelling assessments should include:10 A discussion of how the
modelling assumptions mightaffect the results, indicatingboth the direction of the biasand the approximatemagnitude of the effect
Where applicable, all assumptionsare systematically reported andanalysed
Where applicable, all assumptionsare systematically reported andanalysed
Where applicable, all assumptionsare systematically reported andanalysed
Where applicable, all assumptionsare systematically reported andanalysed
11 A description of the validationundertaken, includingconcurrence of experts,internal consistency, externalconsistency, predictive validity
Validity could be strengthened bydata on progression rates andcosts from clinical trials but thesewere not available – results are anapproximation only. Thereforereported results are conservative
Sensitivity analysis resulted in arange of outcomes that do notsubstantially affect the mainconclusions
Results of the analysis extend thefindings of the DCCT trial
Not described within theliterature.
12 A description of the settings towhich the results of theanalysis can be applied and alist of factors that could limitthe applicability of the results
Settings described within thesystematic review
Settings described within thesystematic review
Settings described within thesystematic review
Settings described within thesystematic review
13 A description of research inprogress that could yield newdata that could alter the resultsof the analysis
Data on progression rates andcosts and resource usage fromactual clinical trials couldstrengthen any study
Data on progression rates andcosts and resource usage fromactual clinical trials couldstrengthen any study
Data on progression rates andcosts and resource usage fromactual clinical trials couldstrengthen any study
Data on progression rates andcosts and resource usage fromactual clinical trials couldstrengthen any study
Reproduced from Chilcott J, Wight J, Lloyd Jones M, Tappenden P. The clinical effectiveness and cost-effectiveness of pioglitazone for Type 2 diabetes mellitus: a rapid and systematicreview. Health Technol Assess 2002;5(19).
Health Technology Assessment 2003; Vol. 7: No. 22
185
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Appendix 14
Summary of educational interventions and comparators, with outline estimates on
UK staffing costs
Appendix 14
186
Study Educational intervention/ Resource implications Estimated additional resource input for education and control group outline cost estimate covering UK staffing costs (2001–2)a
Type 1 studies included in the review of clinical effectiveness
continued
Stockholm DiabetesIntervention Study (SDIS):Reichard et al., (multiplepublications) 1988–9621
(SDIS)RCT
Patients: adult – Type 1
Two groups:1. Usual care: instructed to use
SMBG and visited clinic every4th month, many hadfrequent contact over studyperiod
2. Self-management ed. withintensified treatment.Physician provided 2 sessionsof education to individuals orpairs of 2–3 h. Regularcontact over study period viatelephone
Usual care: patients continued with routinediabetes care
Intervention: physician tutoring was through2 initial education sessions and frequentface-to-face telephone contact, initiallyevery 2 weeks then at greater intervals.Physician available to patients ‘on demand’using a pager system
No cost estimates reported by authors.
Southampton estimateStaffing Resource Inputs:
Minimum 5 h physician time per patient (assuming individualeducation).
Estimate an additional 2 h physician time per patient per year
Estimated minimum costs per patient:
Minimum staff costs (year 1) £506
Minimum staff costs (year 2 onwards) £145
Education materials per patient Not known
Additional costs for training of educators Not known
Additional capital/set-up costs and on-going quality assurance costs Not known
Terent et al., 198522
RCT
Patients: adult – Type 1
Four groups:
1. Usual care
2. Self management ed. +SMBG
3. Self management ed.
4. SMBG
Groups 2–3 provided byphysician and dietitian for sixhourly lessons during 1 month.SMBG groups had additionalsession. Then seen every 3rdmonth.
Group 4 seen in clinic every 3rdmonth
Usual care: pre-trial checking habits
Interventions (groups 2–4): in addition tostandard therapy, education was deliveredby a physician and dietitian in six 1-hindividual sessions. Patients in SMBGgroups (2 and 4) attended an additionalsession by physician for training in SMBG.Patients received photocopies of materialsused
No cost estimates reported by authors
Southampton estimateStaffing resource inputs:Minimum 6 h physician time and 6 h of dietitian time, per patient
Estimated minimum costs per patient:
Minimum staff costs (year 1) £567
Education materials per patient Not known
Additional costs for training of educators Not known
Additional capital/set-up costs and on-going quality assurance costs Not known
Health Technology Assessm
ent2003; Vol. 7: No. 22
187
© Q
ueen’s Printer and Controller of H
MSO
2003. All rights reserved.
Study Educational intervention/ Resource implications Estimated additional resource input for education and control group outline cost estimate covering UK staffing costs (2001/02)a
continued
Mühlhauser et al., 198723
(Geneva–Düsseldorf model)CCT
Patients: adult – Type 1
Three groups:3. Usual care. Under care of
physician
1. IDTTP: self-managementwith intensified treatment.Group education over 5 days,run by diabetes nurses
2. BDTTP: self-managementwith simple rules for insulinadjustment but ‘conventionaltreatment’. Group educationover 4 days, run by diabetesnurses
Usual care: comprised that of theBucharest Hospital. Individual instructionby physician regarding management ofdisease. Insulin prescribed by theoutpatient unit
Interventions: IDTTP – delivered by 2teaching nurses in a structured 5-dayinpatient education course. Groupsconsisted of ~10 patients. IDTTP patientswere followed up exclusively by thetraining team of 2 physicians and nurses.IDTTP – may also result in theintensification of insulin therapy
BDTTP – delivered by two teaching nursesover 4 days. Follow-up in general diabeticoutpatient unit. Patients could contact thetwo physician and two nurse treatmentand teaching team
No cost estimates reported by authors
Note: methods a little outdated given today’s standard methodsfor self-management in diabetes
Southampton estimateStaffing resource inputs:IDDTP and BDTTP required 2 teaching nurses for minimum of 5days, covering ~10 patients
Estimated minimum costs per patient:
IDTTP – Minimum staff costs (year 1) £163
BDTTP – Minimum staff costs (year 1) £130
Education materials per patient (estimate, based on DAFNE data) £ 94
Additional costs for training of educators Not known
Additional capital/set-up costs and on-going quality assurance costs Not known
Note: the standard treatment in the UK would not include a 4-or 5-day inpatient stay to initiate insulin therapy, therefore thatcost would be incurred where education (IDDTP or BDDTP)was delivered on an inpatient basis
Appendix 14
188
Study Educational intervention/ Resource implications Estimated additional resource input for education and control group outline cost estimate covering UK staffing costs (2001/02)a
continued
Starostina et al., 199424
(Geneva–Düsseldorf model)CCT
Patients: adult – Type 1
Three groups:1. Usual care
2. Self-management withintensified treatment +SMBG
3. Self-management withintensified treatment + urinetesting
Groups 2 and 3, 5-day groupeducation provided by 2physicians
Usual care: the study reports that diabeticpatients in Russia and other former USSRcountries are treated by endocrinologistsin district policlinics and as inpatients inspecial endocrinology departments. Thestructural differences between the UKshould be noted
Interventions (groups 2–3): consisted of a 5-day inpatient based educationprogramme. Intervention groups wereinpatient admissions, admitted fortreatment of diabetes. The DTTP methodswere identical with those described inMühlhauser et al., 1987 (as above), exceptthat teaching was delivered by 2 physicians
Cost data reported by authors in Russian roublesb
Southampton estimateStaffing resource inputs:
The DTTP required 2 physicians for a minimum of 5 days,covering ~10 patients
Estimated minimum costs per patient:
Minimum staff costs (year 1) £578
Education materials per patient (estimate based on DAFNE data) £94
Additional costs for training of educators Not known
Additional capital/set-up costs and on-going quality assurance costs Not known
Note: the standard treatment in the UK would not include a 5-day inpatient stay to initiate insulin therapy, therefore that costwould be incurred where education (IDDTP or BDDTP) wasdelivered on an inpatient basis
Health Technology Assessm
ent2003; Vol. 7: No. 22
189
© Q
ueen’s Printer and Controller of H
MSO
2003. All rights reserved.
Study Educational intervention/ Resource implications Estimated additional resource input for education and control group outline cost estimate covering UK staffing costs (2001/02)a
Other selected studies discussed in the review
a Assuming physician time at £72.29 per hour, nurse time at £21.75 per hour and dietitian time at £22.23 per hour (see the table below for detailed assumptions).b See the section ’Results of the systematic search economic evaluations of patient education models for diabetes’ (p. 53).
DAFNE Study Group, 2002Submission to NICE115
Patients: adult – Type 1
Two groups:1. Usual care.
2. DAFNE Education Group
Usual care: the cost-effectiveness analysispresented by the DAFNE Study Groupassumes that standard care reflects thecurrent standard practice of two or threepre-specified insulin dose injections perday
Intervention: 5-day DTTP delivered by aDSN and dietitian, on an outpatient basis.Patients are also greeted by a physician anda physician participates in some of thegroup sessions. In practice, the DTTPinvolves all patients being on a regime ofmultiple daily injections (i.e. those ontwice-daily insulin injections are switchedto multiple injections)
Cost data reported by DAFNE Study Group. Fully inclusive cost estimate provided, based on all costs averaged over 120 patients attending education programmes per centre per year: estimated cost per patient £545
Southampton estimateStaffing resource inputs:
The DTTP requires 1 DSN and 1 dietitian for a minimum of 5days, covering ~8 patients. We also assume 2 h of physician timeper programme.
Estimated minimum costs per patient:
Minimum staff costs (year 1) £223
Education materials per patient (estimate based on DAFNE data) £ 94
Additional costs for training of educators Not known
Additional capital/set-up costs and on going quality assurance costs Not known
Appendix 14
190
Calculations of NHS staff costs
Costs/staff Consultant physician DNS Dietitian(assume Discretionary (assume G Grade (assume Senior Dietitian, Point 3 on salary scale) nurse; top of salary Grade 1; top of salary
scale, point 5) scale, point 6)
Annual salary (£) 76700 26056 25145
Employer’s National 8119 1994 1910Insurance Contribution (£)
Employer’s pension 5262 1730 1675contribution (£)
Overheadsa (£) 24320 2216 2216
Capital overheadsa (£) 4161 2263 3606
Total annual costs (£) 118562 34259 34552
Working time 41 weeks × 40 h 42 weeks × 37.5 h 42 weeks × 37 h
Cost per hour (£) 72.29 21.75 22.23
Cost per day (£) 578.35 163.14 164.53
Source: salary scales from Southampton General Hospital Trust (2001–2).a Overhead estimates based on data from PSSRU.116
Health Technology Assessment 2003; Vol. 7: No. 22
199
Health Technology AssessmentProgramme
Prioritisation Strategy GroupMembers
Chair,Professor Kent Woods,Director, NHS HTA Programme& Professor of Therapeutics,University of Leicester
Professor Bruce Campbell,Consultant Vascular & GeneralSurgeon, Royal Devon & ExeterHospital
Professor Shah Ebrahim,Professor in Epidemiology of Ageing, University of Bristol
Dr John Reynolds, ClinicalDirector, Acute GeneralMedicine SDU, RadcliffeHospital, Oxford
Dr Ron Zimmern, Director,Public Health Genetics Unit,Strangeways ResearchLaboratories, Cambridge
HTA Commissioning BoardMembers
Programme Director, Professor Kent Woods, Director,NHS HTA Programme,Department of Medicine andTherapeutics, Leicester RoyalInfirmary, Robert KilpatrickClinical Sciences Building,Leicester
Chair,Professor Shah Ebrahim,Professor in Epidemiology ofAgeing, Department of SocialMedicine, University of Bristol,Canynge Hall, WhiteladiesRoad, Bristol
Deputy Chair, Professor Jenny Hewison,Professor of Health CarePsychology, Academic Unit ofPsychiatry and BehaviouralSciences, University of LeedsSchool of Medicine, Leeds
Professor Douglas Altman,Professor of Statistics inMedicine, Centre for Statisticsin Medicine, Oxford University,Institute of Health Sciences,Cancer Research UK MedicalStatistics Group, Headington,Oxford
Professor John Bond, Professorof Health Services Research,Centre for Health ServicesResearch, University ofNewcastle, School of HealthSciences, Newcastle upon Tyne
Professor John Brazier, Directorof Health Economics, SheffieldHealth Economics Group,School of Health & RelatedResearch, University ofSheffield, ScHARR RegentCourt, Sheffield
Dr Andrew Briggs, PublicHealth Career Scientist, HealthEconomics Research Centre,University of Oxford, Instituteof Health Sciences, Oxford
Dr Christine Clark, MedicalWriter & Consultant Pharmacist,Cloudside, Rossendale, LancsandPrincipal Research Fellow,Clinical Therapeutics in theSchool of Pharmacy, BradfordUniversity, Bradford
Professor Nicky Cullum,Director of Centre for EvidenceBased Nursing, Department ofHealth Sciences, University ofYork, Research Section,Seebohm Rowntree Building,Heslington, York
Dr Andrew Farmer, SeniorLecturer in General Practice,Department of Primary HealthCare, University of Oxford,Institute of Health Sciences,Headington, Oxford
Professor Fiona J Gilbert,Professor of Radiology,Department of Radiology,University of Aberdeen, LilianSutton Building, Foresterhill,Aberdeen
Professor Adrian Grant,Director, Health ServicesResearch Unit, University ofAberdeen, Drew Kay Wing,Polwarth Building, Foresterhill,Aberdeen
Professor Alastair Gray, Director,Health Economics ResearchCentre, University of Oxford,Institute of Health Sciences,Headington, Oxford
Professor Mark Haggard,Director, MRC ESS Team, CBUElsworth House, Addenbrooke’sHospital, Cambridge
Professor F D Richard Hobbs,Professor of Primary Care &General Practice, Department ofPrimary Care & GeneralPractice, University ofBirmingham, Primary Care andClinical Sciences Building,Edgbaston, Birmingham
Professor Peter Jones, Head ofDepartment, UniversityDepartment of Psychiatry,University of Cambridge,Addenbrooke's Hospital,Cambridge
Professor Sallie Lamb, ResearchProfessor in Physiotherapy/Co-Director, InterdisciplinaryResearch Centre in Health,Coventry University, Coventry
Dr Donna Lamping, SeniorLecturer, Health ServicesResearch Unit, Public Healthand Policy, London School ofHygiene and Tropical Medicine,London
Professor David Neal, Professorof Surgical Oncology, OncologyCentre, Addenbrooke's Hospital,Cambridge
Professor Tim Peters, Professorof Primary Care Health ServicesResearch, Division of PrimaryHealth Care, University ofBristol, Cotham House, CothamHill, Bristol
Professor Ian Roberts, Professorof Epidemiology & PublicHealth, Intervention ResearchUnit, London School ofHygiene and Tropical Medicine,London
Professor Peter Sandercock,Professor of Medical Neurology,Department of ClinicalNeurosciences, University ofEdinburgh, Western GeneralHospital NHS Trust, BramwellDott Building, Edinburgh
Professor Martin Severs,Professor in Elderly HealthCare, Portsmouth Institute ofMedicine, Health & Social Care,St George’s Building,Portsmouth
Dr Jonathan Shapiro, SeniorFellow, Health ServicesManagement Centre, ParkHouse, Birmingham
Current and past membership details of all HTA ‘committees’ are available from the HTA website (www.ncchta.org)
© Queen’s Printer and Controller of HMSO 2003. All rights reserved.
Health Technology Assessment Programme
200
Diagnostic Technologies & Screening PanelMembers
Chair,Dr Ron Zimmern, Director ofthe Public Health Genetics Unit,Strangeways ResearchLaboratories, Cambridge
Dr Paul Cockcroft, ConsultantMedical Microbiologist/Laboratory Director, PublicHealth Laboratory, St Mary’s Hospital, Portsmouth
Professor Adrian K Dixon,Professor of Radiology,Addenbrooke’s Hospital,Cambridge
Dr David Elliman, Consultant inCommunity Child Health,London
Dr Andrew Farmer, SeniorLecturer in General Practice,Institute of Health Sciences,University of Oxford
Dr Karen N Foster, ClinicalLecturer, Dept of GeneralPractice & Primary Care,University of Aberdeen
Professor Jane Franklyn,Professor of Medicine,University of Birmingham
Professor Antony J Franks,Deputy Medical Director, TheLeeds Teaching Hospitals NHSTrust
Mr Tam Fry, HonoraryChairman, Child GrowthFoundation, London
Dr Susanne M Ludgate, MedicalDirector, Medical DevicesAgency, London
Dr William Rosenberg, SeniorLecturer and Consultant inMedicine, University ofSouthampton
Dr Susan Schonfield, CPHMSpecialised ServicesCommissioning, CroydonPrimary Care Trust
Dr Margaret Somerville,Director of Public Health,Teignbridge Primary Care Trust,Devon
Mr Tony Tester, Chief Officer,South Bedfordshire CommunityHealth Council, Luton
Dr Andrew Walker, SeniorLecturer in Health Economics,University of Glasgow
Professor Martin J Whittle,Head of Division ofReproductive & Child Health,University of Birmingham
Dr Dennis Wright, ConsultantBiochemist & Clinical Director,Pathology & The KennedyGalton Centre, Northwick Park& St Mark’s Hospitals, Harrow
Pharmaceuticals PanelMembers
Chair,Dr John Reynolds, ClinicalDirector, Acute GeneralMedicine SDU, OxfordRadcliffe Hospital
Professor Tony Avery, Professorof Primary Health Care,University of Nottingham
Professor Iain T Cameron,Professor of Obstetrics &Gynaecology, University ofSouthampton
Mr Peter Cardy, ChiefExecutive, Macmillan CancerRelief, London
Dr Christopher Cates, GP andCochrane Editor, Bushey HealthCentre, Bushey, Herts.
Mr Charles Dobson, SpecialProjects Adviser, Department ofHealth
Dr Robin Ferner, ConsultantPhysician and Director, WestMidlands Centre for AdverseDrug Reactions, City HospitalNHS Trust, Birmingham
Dr Karen A Fitzgerald,Pharmaceutical Adviser, Bro TafHealth Authority, Cardiff
Professor Alastair Gray,Professor of Health Economics,Institute of Health Sciences,University of Oxford
Mrs Sharon Hart, ManagingEditor, Drug & TherapeuticsBulletin, London
Dr Christine Hine, Consultant inPublic Health Medicine, Bristol South & West PrimaryCare Trust
Professor Robert Peveler,Professor of Liaison Psychiatry,Royal South Hants Hospital,Southampton
Dr Frances Rotblat, CPMPDelegate, Medicines ControlAgency, London
Mrs Katrina Simister, NewProducts Manager, NationalPrescribing Centre, Liverpool
Dr Ken Stein, Senior Lecturer inPublic Health, University ofExeter
Professor Terence Stephenson,Professor of Child Health,University of Nottingham
Dr Richard Tiner, MedicalDirector, Association of theBritish Pharmaceutical Industry,London
Professor Dame Jenifer Wilson-Barnett, Head of FlorenceNightingale School of Nursing& Midwifery, King’s College,London
Current and past membership details of all HTA ‘committees’ are available from the HTA website (www.ncchta.org)
Health Technology Assessment 2003; Vol. 7: No. 22
201
Therapeutic Procedures PanelMembers
Chair, Professor Bruce Campbell,Consultant Vascular andGeneral Surgeon, Royal Devon& Exeter Hospital
Dr Mahmood Adil, Head ofClinical Support & HealthProtection, Directorate ofHealth and Social Care (North),Department of Health,Manchester
Professor John Bond, Head ofCentre for Health ServicesResearch, University ofNewcastle upon Tyne
Mr Michael Clancy, Consultantin A & E Medicine,Southampton General Hospital
Dr Carl E Counsell, SeniorLecturer in Neurology,University of Aberdeen
Dr Keith Dodd, ConsultantPaediatrician, DerbyshireChildren’s Hospital, Derby
Professor Gene Feder, Professorof Primary Care R&D, Barts &the London, Queen Mary’sSchool of Medicine andDentistry, University of London
Ms Bec Hanley, FreelanceConsumer Advocate,Hurstpierpoint, West Sussex
Professor Alan Horwich,Director of Clinical R&D, TheInstitute of Cancer Research,London
Dr Phillip Leech, PrincipalMedical Officer for PrimaryCare, Department of Health,London
Mr George Levvy, ChiefExecutive, Motor NeuroneDisease Association,Northampton
Professor James Lindesay,Professor of Psychiatry for theElderly, University of Leicester
Dr Mike McGovern, SeniorMedical Officer, Heart Team,Department of Health, London
Dr John C Pounsford,Consultant Physician, NorthBristol NHS Trust
Professor Mark Sculpher,Professor of Health Economics,Institute for Research in theSocial Services, University ofYork
Dr L David Smith, ConsultantCardiologist, Royal Devon &Exeter Hospital
Professor Norman Waugh,Professor of Public Health,University of Aberdeen
Current and past membership details of all HTA ‘committees’ are available from the HTA website (www.ncchta.org)
Health Technology Assessment Programme
202Current and past membership details of all HTA ‘committees’ are available from the HTA website (www.ncchta.org)
Expert Advisory NetworkMembers
Mr Gordon Aylward, Chief Executive, Association of British Health-Care Industries, London
Ms Judith Brodie, Head of Cancer SupportService, Cancer BACUP, London
Mr Shaun Brogan, Chief Executive, RidgewayPrimary Care Group, Aylesbury,Bucks
Ms Tracy Bury, Project Manager, WorldConfederation for PhysicalTherapy, London
Mr John A Cairns, Professor of Health Economics,Health Economics ResearchUnit, University of Aberdeen
Professor Howard Stephen Cuckle, Professor of ReproductiveEpidemiology, Department ofPaediatrics, Obstetrics &Gynaecology, University ofLeeds
Professor Nicky Cullum, Director of Centre for EvidenceBased Nursing, University of York
Dr Katherine Darton, Information Unit, MIND – TheMental Health Charity, London
Professor Carol Dezateux, Professor of PaediatricEpidemiology, London
Professor Martin Eccles, Professor of ClinicalEffectiveness, Centre for HealthServices Research, University ofNewcastle upon Tyne
Professor Pam Enderby,Professor of CommunityRehabilitation, Institute ofGeneral Practice and PrimaryCare, University of Sheffield
Mr Leonard R Fenwick, Chief Executive, Newcastleupon Tyne Hospitals NHS Trust
Professor David Field, Professor of Neonatal Medicine,Child Health, The LeicesterRoyal Infirmary NHS Trust
Mrs Gillian Fletcher, Antenatal Teacher & Tutor andPresident, National ChildbirthTrust, Henfield, West Sussex
Ms Grace Gibbs, Deputy Chief Executive,Director for Nursing, Midwifery& Clinical Support Servs., WestMiddlesex University Hospital,Isleworth, Middlesex
Dr Neville Goodman, Consultant Anaesthetist,Southmead Hospital, Bristol
Professor Robert E Hawkins, CRC Professor and Director ofMedical Oncology, Christie CRCResearch Centre, ChristieHospital NHS Trust, Manchester
Professor F D Richard Hobbs, Professor of Primary Care &General Practice, Department ofPrimary Care & GeneralPractice, University ofBirmingham
Professor Allen Hutchinson, Director of Public Health &Deputy Dean of ScHARR,Department of Public Health,University of Sheffield
Professor Rajan Madhok, Medical Director & Director ofPublic Health, Directorate ofClinical Strategy & PublicHealth, North & East Yorkshire& Northern Lincolnshire HealthAuthority, York
Professor David Mant, Professor of General Practice,Department of Primary Care,University of Oxford
Professor Alexander Markham, Director, Molecular MedicineUnit, St James’s UniversityHospital, Leeds
Dr Chris McCall, General Practitioner, TheHadleigh Practice, CastleMullen, Dorset
Professor Alistair McGuire, Professor of Health Economics,London School of Economics
Dr Peter Moore, Freelance Science Writer,Ashtead, Surrey
Dr Andrew Mortimore, Consultant in Public HealthMedicine, Southampton CityPrimary Care Trust
Dr Sue Moss, Associate Director, CancerScreening Evaluation Unit,Institute of Cancer Research,Sutton, Surrey
Professor Jon Nicholl, Director of Medical CareResearch Unit, School of Healthand Related Research,University of Sheffield
Mrs Julietta Patnick, National Co-ordinator, NHSCancer Screening Programmes,Sheffield
Professor Chris Price, Visiting Chair – Oxford, ClinicalResearch, Bayer DiagnosticsEurope, Cirencester
Ms Marianne Rigge, Director, College of Health,London
Professor Sarah Stewart-Brown, Director HSRU/HonoraryConsultant in PH Medicine,Department of Public Health,University of Oxford
Professor Ala Szczepura, Professor of Health ServiceResearch, Centre for HealthServices Studies, University ofWarwick
Dr Ross Taylor, Senior Lecturer, Department ofGeneral Practice and PrimaryCare, University of Aberdeen
Mrs Joan Webster, Consumer member, HTA –Expert Advisory Network
The National Coordinating Centre for Health Technology Assessment,Mailpoint 728, Boldrewood,University of Southampton,Southampton, SO16 7PX, UK.Fax: +44 (0) 23 8059 5639 Email: [email protected]://www.ncchta.org ISSN 1366-5278
FeedbackThe HTA Programme and the authors would like to know
your views about this report.
The Correspondence Page on the HTA website(http://www.ncchta.org) is a convenient way to publish
your comments. If you prefer, you can send your comments to the address below, telling us whether you would like
us to transfer them to the website.
We look forward to hearing from you.