Transcript The Scottish Cancer Registry and linked diabetes register

The Scottish
electronic diabetes register and
cancer registry and their linkage
Sarah Wild, University of Edinburgh
Thanks to David Brewster, Director of the
Scottish Cancer Registry
June 2010
Background - Scotland
• Population approximately 5.1 million people, 14
Health Boards
• Health data identified using Community Health
Index which is used for routine linkage of
hospital admissions, cancer registrations and
mortality records
• Prescribing data available from primary care
records – encashment data from pharmacies
available soon
Background – diabetes register
• Scottish Care Information – Diabetes
Collaboration (SCI-DC) is an electronic
population-based diabetes register populated by
daily downloads from primary and secondary
care systems, including prescribing
• Data collection started in Tayside and
Lanarkshire with national data collected from
2000 onwards
• Core data set includes demographic and
relevant clinical data – completeness >95%
(except ethnicity)
• Access for research through Scottish Diabetes
Research Network (SDRN) epidemiology group
Number of people included in the
Scottish electronic diabetes register
Source: Scottish Diabetes Survey 2009
2008 data extract
Data available for approx 270,000 people
Based on people with data on age, sex and
SIMD recorded on SCI-DC who were alive
in April 2008 (approx. 204,000 of 209,000)
there were:
• 26,242 people with T1DM (crude
prevalence 0.51%)
• 177,783 people with T2DM (crude
prevalence 3.46%)
GROS 2008 population estimates used for denominator
Validation of recorded type 1
diabetes and prevalence
Ratio of observed prevalence to expected prevalence of type 1 diabetes estimated from Tayside age-sex
prevalence applied to other Health Boards' population data
Type I
1.60
1.40
1.20
1.00
0.80
0.60
0.40
0.20
&
s
fri
e
um
nd
et
la
Sh
G
al
lo
n
te
r
W
es
D
G
re
Health Board of Residence
wa
y
es
Is
l
le
y
Va
l
rth
Fo
Lo
th
i
an
y
ne
O
rk
m
pi
an
G
ra
hi
re
rk
s
na
H
ig
h
la
n
La
G
la
sg
at
er
Ay
d
e
&
C
lyd
Fi
fe
rs
rd
e
ow
&
rs
h
ire
Bo
ys
id
e
Ar
ra
n
0.00
Ta
Ratio of Observed to Expected Prevalence
1.80
Age and sex standardised prevalence
of type 2 diabetes by Scottish Index of
Multiple Deprivation quintile
Age-sex standardised
prevalence %
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
1 (least
deprived)
2
3
SIMD quintile
4
5 (most
deprived)
Background – cancer registry
• Cancer data collected since 1958 with full
dataset including stage, (for breast, colorectal
and cervical cancer) and information on
treatment for all patients collected from 1997
• Approx. 40,000 registrations/ year
• Total of over 1,200,000 records
• Data quality monitored using routine indicators,
computer validation, data exchange with
specialist tumour registries and ad hoc studies of
data accuracy and completeness of
ascertainment.
Collection of Scottish Cancer Registry data
• Electronic capture and reformatting of ‘raw’ data
• Probability matching to related incoming
records and to previous registrations
• Application of multiple primary rules and rules to
deal with metastatic disease
• Creation of ‘best guess’ provisional registrations
• Completion and validation by staff based in
hospitals using medical records at least 6
months after date of diagnosis
Source data
•
•
•
•
•
•
Hospital discharge records
Radiotherapy, oncology, haematology
Pathology records (8 different lab systems)
Mortality records
Cancer audit data (not for all cancers)
Other, eg paper records from private
hospitals
• Deaths from General Register Office for
Scotland
Invasive cancer of the cervix uteri in Scotland
Age-standardised incidence and mortality rates per
100,000 person-years at risk by SIMD 2006 category
16
14
Rate per 100,000
12
10
Incidence
8
Mortality
6
4
2
0
1 (Least deprived)
2
3
Deprivation quintile
4
5 (Most deprived)
Oesophageal adenocarcinoma in UK women:
results of a case-control study
Variable
Adjusted OR (95% CIs) P for trend
BMI at age 20
Q1 <19.49
Q2 19.49-20.95
Q3 20.96-22.66
Q4 >22.67
1
0.86 (0.17-4.32)
4.90 (0.86-28.02)
6.04 (1.28-28.52)
0.002
Total fruit consumption (per week)
Q1 <12.00
Q2 12.01-18.04
Q3 18.05-25.72
Q4 >25.73
1
0.42 (0.09-2.03)
0.37 (0.05-2.59)
0.08 (0.01-0.49)
0.002
Breastfeeding
No children
Had children but never breast fed
Up to 6 months
>6 months
1
0.66 (0.06-6.88)
0.30 (0.04-2.30)
0.13 (0.01-1.40)
0.005
Source: Cheng et al. Br J Cancer 2000; 83: 127-32
Factors influencing population-based survival data
Data quality factors
Tumour-related factors
Completeness of ascertainment
Accuracy of registration
Completeness of follow-up
‘Death certificate only’ registrations
Extent of disease
Site (and subsite) of tumour
Morphology
Tumour biology
Host factors
Health care-related factors
Age
Sex
Race/Ethnicity
Co-morbidity
Socio-economic status
Behaviour (including awareness of
Screening
Diagnostic facilities
Treatment facilities
Quality of treatment
Follow-up care
cancer symptoms and compliance
with treatment)
Linkage pilot (2005 data, 2 Health
Boards): validation of diabetes
recording
• 47% of 82,958 hospital records after a
diagnosis of diabetes include a mention of
diabetes (71% of SMR01 records for MI after
diagnosis of diabetes)
Of the 4,777 death certificates, 9.6% gave
diabetes as the underlying cause of death and
a further 39% mentioned diabetes as a
contributory cause
Background to linkage project
• The Scottish Government Health
Department funded the linkage of 2008
SCI-DC data to cancer (2007) and hospital
admission/mortality (2008) data.
• Ethical and Caldicott guardian approval
obtained
• Early work based on improving quality of
data on date of diagnosis, type of
diabetes, prescribing data, glargine &
cancer
• Further linkages to renal register,
maternity records, hepatitis C register
Characteristics by treatment group
Non-glargine
insulin
Non-glargine plus
Insulin
glargine insulin glargine only p valuea
Characteristic
Data presented as
Subjects
Sex, % women
Age, years
% (n)
% (n)
median (LQ, UQ)
92.7 (18,455)
48.8 (8,996)
65 (56,72)
5.2 (1,033)
48.5 (501)
57 (49, 66)
2.1 (411)
52.6 (216)
69 (59,77)
0.13
<0.0001
BMI, kg / m2
Systolic BP, mmHg
Diastolic BP, mmHg
HbA1c, %
Duration of diabetes
5 years
Prior Insulin
5years on insulin
Age at diagnosis, years
Prior cancer
Ever
5 years ago
Any CVD
Ever smoked
Use of oral glucose-lowering
drugs at baselineb
On three or more oral glucoselowering drugs at baseline
In the two most deprived SIMD
quintiles
mean (SD)
mean (SD)
mean (SD)
mean (SD)
30.4 (6.1)
139.3 .2(22)
76 (12)
8.5 (1.7)
27.9 (5.7)
136.1 (19.5)
77 (12)
9.0 (1.7)
30.1 (6.2)
140.5(20.5)
77(13)
9.3 (1.8)
<0.0001
0.19
0.006
<0.0001
% (n)
83.2 (15,360)
76.8 (793)
75.2 (309)
<0.0001
% (n)
median
37.2 (6,867)
53
47.1 (487)
46
4.6 (19)
59
<0.0001
<0.0001
% (n)
% (n)
% (n)
% (n)
7.4 (1,357)
3.6 (664)
18.6 (3,441)
29.9 (4,335)
5.0 (52)
2.7 (28)
12.1 (125)
28.4 (247)
9.3 (38)
5.6 (23)
22.1 (91)
24.8 (77)
0.59
0.35
0.016
0.001
% (n)
28.9(4559)
23.4 (169)
80.0 (295)
<0.0001
% (n)
4.0 (745)
3.1 (32)
29.4 (121)
<0.0001
% (n)
47.1 (8,686)
35.4 (366)
54.0 (222)
<0.0001
Risk of cancer by treatment group
aModel
1 adjusts for prior cancer, type of diabetes and calendar year, and is stratified by sex ; timescale is age
2 further adjusts for baseline metformin, sulfonylurea and other oral hypoglycaemic drugs at baseline
cModel 3 further adjusts for diabetes duration, HbA , diastolic BP, systolic BP and deprivation quintile
1c
dModel 4 further adjusts for smoking ever and BMI, but note the reduction in available sample size
bModel
Cancer registrations and deaths in
2008 Scottish diabetes linked data
Year
Registrations
Deaths
2001
2535
845
2002
2693
1033
2003
2818
1240
2004
3041
1313
2005
3029
1465
2006
3155
1585
2007
3151
1703
Only includes first registration of each cancer in an individual person. Includes
ICD-9 codes 140-239 EXCEPT 173; ICD-10 codes C00-C97 EXCEPT C44
Distribution of cancer registrations
by type/date of diagnosis of diabetes
Type1
Type 2
Cancer prior to
diagnosis of DM
469
12065
Cancer after
diagnosis of DM
1516
19113
Date diagnosis
of DM unknown
50
730
Number of site specific cancers
following diagnosis of DM
Colo-rectal
Lung
Breast
Prostate
Pancreas
Endometrial
Liver
Type 1
204
192
221
92
86
38
48
Type 2
3360
3078
2292
2092
812
689
565
Acknowledgements
• SCI-DC data are available for analysis by members of
the Scottish Diabetes Research Network thanks to the
hard work of numerous NHS staff who enter the data
and people and organisations (the Scottish Care
Information –Diabetes Collaboration [SCI-DC] Steering
Group, the Scottish Diabetes Group, the Scottish
Diabetes Survey Group, the managed clinical networks
managers and staff in each Health Board) involved in
setting up, maintaining and overseeing SCI-DC .
• Financial support for the work was provided by the
Scottish Government and the Wellcome Trust through
the Scottish Health Informatics Programme