Hyperglycaemia and diabetes risk among 100,000 patients
Opportunities and challenges in using routine healthcare data
Dr David McAllisterClinical Lecturer in Epidemiology, University of Edinburgh
and Specialty Registrar in Public Health, NHS Fife
Acknowledgements
• Alistair McLeod and Paul Ewing at Lothian and Fife SCI-stores• Claire MacDonald, Marion Flood and Roma Armstrong at NHS
Greater Glasgow and Clyde Safe Haven• Christopher Hall and Duncan Heather of Health Informatics
Centre• David Bailley, Philip Johnston, Janet Murray at NHS ISD• Sharon Tuck and Cat Graham at Wellcome Trust CRF Statistics
and Epidemiology• Collaborators – Katherine Hughes, Nazir Lone, Naveed Sattar,
Johnny McKnight and Sarah Wild• Chief Scientist Office, Scotland
4
6
8
10
12
0 10 20Time (days)
Glu
cose
(m
mo
l/L)
Normal
Stress hyperglycaemia
Type 2 DM
Illness
Longitudinal perspective
4
6
8
10
12
0 10 20Time (days)
Glu
cose
(m
mo
l/L)
Normal
Stress hyperglycaemia
At risk
Type 2 DM
Illness
Longitudinal perspective
Clinician perspective
4
6
8
10
12
0 10 20Time (days)
Glu
cose
(m
mo
l/L)
Normal
Stress hyperglycaemia
At risk
Type 2 DM
Illness
Previous literature
• specific diseases– coronary disease– stroke– pneumonia
• small (largest n = 2,000)• short duration• considerable loss to follow-up
Routinely collected data
• Uncommon outcome• Representative sample (population)• Range of conditions• Answer 2 questions
- Distribution of glucose- Incidence of type 2 diabetes
Emergency admissionsEm
erge
ncy
adm
issi
ons
- SM
R01
• Scottish Morbidity Record• Codes for :-
– emergency– specialty– diagnosis
• Age, sex, deprivation quintile• 2004-2010
Glucose results – SCI-storeFife
Lothian
Glasgow …
Ayrshire ..
Grampian
Borders
Dumfries …
Forth Valley
Highland
Orkney
Shetland
Tayside
• Disseminating clinical results
• Lab and radiology• Independent• Commercially supported
Emer
genc
y ad
mis
sion
s - S
MR0
1
Diabetes - outcomeEm
erge
ncy
adm
issi
ons
- SM
R01
SCI-
diab
etes
– 3
yea
r in
cide
nce
• SCI-diabetes• Prevalence• Incidence to Nov 2011
Fife
Lothian
Glasgow …
Ayrshire ..
Grampian
Borders
Dumfries …
Forth Valley
Highland
Orkney
Shetland
Tayside
Approvals
• Ethics Committee• NHS Research and Development• Scottish Diabetes Research Network• Privacy Advisory Committee at Information
Services Division (ISD)• Scotland-wide Caldicott Officers (pilot)• Glasgow Safe Haven• Health Informatics Centre
Mar-12 Apr-12 May-12 Jun-12 Jul-12 Aug-12
dw SCI storeGrant
submittedEthics waiver
Caldicott approval
Sep-12 Oct-12 Nov-12 Dec-12 Jan-13 Feb-13PAC
approvalSMR01 extract
Lothian extract
Requested start early
SCIDC
Mar-13 Apr-13 May-13 Jun-13 Jul-13 Aug-13
SQL shared Fife extractGlasgow extract
Extention granted
Sep-13 Oct-13 Nov-13 Dec-13 Jan-14 Feb-14First
submittedTayside extract
Second submission
Mar-14 Apr-14 May-14 Jun-14 Jul-14 Aug-14
Third submission
Accepted PublishedRevisions
Challenge SCI store data• SQL written in Lothian• Resources vs costs• Maintain clinical service• Data governance drift
• Data not research-ready
Emer
genc
y ad
mis
sion
s - S
MR0
1
SCI-
diab
etes
– 3
yea
r in
cide
nce
Fife
Lothian
Glasgow
Lanarkshire
Tayside
PLOS journals require authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception.
PLOS defines the “minimal dataset” to consist of the dataset used to reach the conclusions drawn in the manuscript … to replicate the reported study findings in their entirety
Minimal dataset
• Age (5-year bands)• Sex• SIMD quintile• Glucose (nearest 0.1 mmol/l)• Medical/ Surgical• COPD , MI , Fracture , Stroke (0 or 1)• Diagnosed with type 2 diabetes (0 or 1)• Days from discharge to diabetes• Year of discharge from hospital