Peter Schofield

advertisement
The Lambeth DataNet
- how UK primary care data can inform
service development and research
Peter Schofield
MRC Population Health Scientist
Department of Primary Care and Public Health Sciences
Division of Health and Social Care Research
Kings College London
peter.1.schofield@kcl.ac.uk
The unique benefits of UK primary care data
•
Universal health care -> free at the point of use
•
National Institute for Health & Care Excellence (NICE) guidelines -> ensure
standardised treatment
•
Quality and Outcomes Framework (QOF) - national incentive scheme ->
standardised data collected
•
Primary care –> first point of access
•
-> GP holds central care record for each patient
•
National data
– QOF register -> national data at practice level
– CPRD, THIN, QRESEARCH -> national data at patient (sample) level
•
Local data
– Small number of schemes -> comprehensive local data at patient level
Clapham Common
Brixton Market
Socioeconomic
diversity in
Lambeth
Deprivation
scale
Most deprived
Least deprived
Lambeth is the 19th most
deprived (of 354 local
authorities in England)
Ethnic diversity
in Lambeth
Proportion nonwhite British
Lambeth has the 2nd highest
proportion of Black
residents in the UK (25.8%)
The Lambeth DataNet
GP records and patient profile data covering Lambeth GP practices
What socio-demographic information do we have?
•
•
•
•
•
Residential area (detailed): 99.9% of patients
Ethnicity (census - 16 categories): 81%
Main language spoken: 73%
Country of birth: 56%
Religion: 47%
What clinical information do we have?
•
•
•
•
Potentially anything coded by the GP: i.e. diagnoses, test results, prescribing ..
QOF indicators are particularly reliable
Can be linked with secondary care (hospital records, psychiatric records)
Also linked to area level data (e.g. census, area deprivation etc.. )
How is the data collected?
• Previously extracted from practice computer systems using MIQUEST software
– required specialist expertise to run queries
• Currently extracted centrally using EMIS Web
– easier to run queries on the data
– potential to update and interrogate dataset in real time
When is this data collected?
• Typically we have obtained a new extraction annually
• Can now analyse longitudinally patient records going back to 2008
How much data do we have?
• Currently all patient records for 51/52 GP practices in Lambeth
• Approximately 350,000 patients in total
Who owns the data?
• Ultimately the data is owned by Lambeth GPs
• Responsibility for approving research and organising the extractions is
delegated to the DataNet steering group
Socio-demographic profile*
Encrypted ID
“Read” codes
*data here is fictitious but in the same format as original records
Clinical record*
*data here is fictitious but in the same format as original records
Why is this useful for research?
-
Lambeth is well placed for research into ethnic and socio-economic inequalities
-
Typically better socio-demographic information compared to national data
-
Larger sample of non-white ethnic groups than other national datasets
- i.e. national survey data / national health records datasets
-
Local data -> allows analysis in context
-
Whole population data => Includes information about healthy controls
-
Studies so far
Ethnic inequalities in blood pressure control
Marked ethnic inequalities were found, with black patients significantly less likely to
achieve BP targets than white (OR 0.73). Once ethnicity was accounted for social
deprivation made little or no difference.
Ethnic inequalities in the treatment of hypertension
National guidelines recommending specific therapy for hypertension management
often ignored. Applies to both black and white patients equally.
Cardiovascular Risk Scores
We compared the results of four different CVD prediction scores: FRAMINGHAM, QRISK2, ASSIGN and ETHRISK. We found all except Q-RISK2 over-predicted risk for the
black population.
Ethnicity and stroke risk in patients with atrial fibrillation
Stroke risk is higher among South Asians with AF. Under-prescription of anticoagulants
persists in all ethnic groups.
Ethnic differences in the management of patients with psychosis
No ethnic differences in physical health care BUT Black patients > 2 x likely to be on
depot anti-psychotics.
Ethnic isolation and ethnic inequality in risk of psychosis
Controlling for deprivation, Black people in high own ethnic density areas
showed no difference in psychosis rates but in low ethnic density areas were
at significantly greater risk.
Predictors of first episode psychosis for black population in Lambeth
Ethnic Density
1
1st quintile (most dense)
2nd quintile
3rd quintile
4th quintile
5th quintile (least dense)
IRR
1.00
2.50**
3.59***
5.39***
5.24**
95% CI
(1.37-4.58)
(1.87-7.00)
(2.48-11.69)
(1.95-14.07)
Analysis at lower super output area (LSOA) level –approx. 1,500 people
Much smaller effect size shown at (typically used) middle SOA level – approx.
7,500 people
Ethnic density and rates of first episode psychosis for black
population in Lambeth
6
5
4
3
2
1
0
1st quintile
2nd
- most
quintile
dense
3rd
quintile
4th
quintile
5th
quintile least dense
Current follow up study
Revisit ethnic density analysis using:
1) Lambeth DataNet linked with psychiatric data – to validate and enhance
primary care mental health data
2) Data pooled with similar primary care data collected for East London practices
Other current projects using similar linked data
• Location of care for people with severe mental illness: implications for
service use and costs
– Investigating the shift from secondary to primary care for severe
mental illness using Lambeth DataNet linked to psychiatric records
• Mental & physical health in Lambeth
– Investigating physical health care of people with severe mental illness
using linked data
Other projects
• Effect of multi-morbidity on blood pressure control – also using data
pooled with health records from East London
• Ethnic inequalities in smoking and COPD rates
Data Governance
•
Ethical approval
NHS ethics committees now treat secondary data analysis as audit => ethical approval usually not
required
•
Local governance approval
BUT obtaining local approval for data sharing can be a challenge:
•
Care.data
Recent attempt to introduce a comprehensive national patient records research dataset
-> public outcry over potential for data sharing abuses
 now much more difficult to obtain local data sharing approval
•
NHS re-organisation
Local Primary Care Trusts (PCTs) recently disbanded
-> approval process now being re-organised
•
Data Linkage approval
Currently difficult to obtain for sensitive data
Relies on “accredited safe haven”
Lambeth DataNet and Audit
 Identify inequalities
• e.g. under-diagnosed diabetes in black community
 Identify unmet needs
• e.g. under-use of anti-coagulants in AF
 Apply targeted interventions
 Inform re-design of services
Public health reports feeding back individual practice performance in key
clinical areas:
• Hypertension
• Atrial fibrillation
• Diabetes
• Dementia
• Severe mental illness
• Depression
• Use of Psychological Therapies
Summary
• The UK is uniquely well placed to provide electronic health records
for audit & research
• Local data enables whole population studies in context
• Lambeth DataNet is a particularly valuable resource for research
into ethnic inequalities
• Data linkage has enabled a range of studies examining the interface
between primary and secondary care
• Obtaining local governance approval can be a challenge
• Local data facilitates local audit
Lambeth DataNet Publications
Mathur, R., E. Pollara, S. Hull, P. Schofield, M. Ashworth and J. Robson (2013). Ethnicity
and stroke risk in patients with atrial fibrillation. Heart.
Schofield P, Crichton N, Chen R (2012) Methods for assessing cardiovascular disease
risk in a UK black population. Heart.
Schofield P, Baawuah F, Seed P, Ashworth M. (2012) Managing hypertension in general
practice: a cross sectional study of treatment and ethnicity. British Journal of General
Practice
Schofield P, Ashworth M, Jones R (2011) Ethnic isolation and psychosis: re-examining
the ethnic density effect. Psychological Medicine
Schofield P, Saka O, Ashworth M., (2011) Ethnic differences in blood pressure
monitoring and control in Lambeth British Journal of General Practice.
Pinto, R., M. Ashworth, P. Seed, G. Rowlands, P. Schofield, R. Jones (2010) Differences
in the primary care management of patients with psychosis from two ethnic groups: a
population based cross-sectional study. Family Practice.
Kumarapeli, P., R. Stepaniuk, S. de Lusignan, R. Williams and G. Rowlands (2006).
Ethnicity recording in general practice computer systems. Journal of Public Health
Download