RESEARCH AND BI DATA Where is it going? Benefits, opportunities Challenges John Parkinson CPRD, UK
Professor John Parkinson Expert Consultant CPRD MRC Bioinformatics Centre, University of Leeds RESEARCH AND BI DATA
BI DATA AND HEALTHCARE EHR data not that big But COMPLEX Incorrect, Missing, Unexpected/out of normal bounds, Time variable Scan data BI Need methods for reading the binary files mhealth data BI PROs, devices with wifi feeds (1000 entries per second) enetic data BI (= Lifetime EHR) Other Food, sport BI?
BUT IS BI IN OTHER WAYS Complexity means Requires experts Big powerful computing
THE UK SCENE Unique patient ID (NHS number) 99+% EHR primary care records for 25 years About 50 NHS datasets National Mortality PROs NHSEngland 54 million Clinical Practice Research Datalink (CPRD) Synergy of CPRD NHSE NIHR HSCIC
USES OF DATA NHS Intelligence Citizen Involvement Data Audit Surveillance Research
NHS Datasets I o Care 8500Practices 8LRNs % 2 o Care 100% HES +Rx NHS Audits 40 ranularity Very detailed Coded record Lab Path Microb. ICD10 OPCS4 Drugs 100% National central Data ONS Death Birth Demography 100% Linkage on NHS# And Or PC, sex, DOB Social Care Data Environ. Air Poll. Food Data? Consented Clinical Trials Biosample PROs PHE Cancer BINOCAR 100% Disease Reg. 5 Local Data enetic data Digital path. data Approved Researchers
OPPORTUNITIES AND BENEFITS 1. Cardiovascular outcomes linked data 2. Drugs in pregnancy need max data 3. Drugs for children CTs? max data 4. New products who Rxs first 5. Vaccines large population data 6. Infection tracking 7. Cancer acute or chronic 8. Clinical Trials faster, more cost effective 9. Drugs in hospital 10.Dynamic Risk scoring
PERMISSIONS AND OVERNANCE
CPRD (MHRA) Summary Data Stewardship Map Key to Symbols Ch L CoE A Charter or statement Privacy Enhancing Technology Physical Security Legal agreement overnance/risk minimisation approvals Contracts of Employment Right of Audit Standard Operating Procedures
CPRD (MHRA) Summary Data Stewardship Map Inbound (Data Acquisition) Patient Communication Materials: Posters, leaflets, website copy Ch Right to opt-out Ch Data Subject Domain End to End I Policy I Tool- Kit & I Management Clinical data Primary care Clinical Data with Personal IDS Separation of Clinical Data and Personal IDs Practice Identity Keyed Names, addresses removed Clinical Data with Pseudo ID1 using Key 1 Data encrypted Data on Secure N3 network Data to High Security CPRD Data Centre Data Warehouse With Pseudo ID1 Data only accessible to Limited Documented /Controlled Users CoE A L Data encrypted Key 2 Data to Research Data Store With Pseudo ID2 In High Security Data Centre 1 IDs only NHS#, DOB, DOD, Sex and postcode Pseudo ID 1 NO Links between Data centre 1 and 2 Ch Ch Data on Secure N3 network Other NHSClinical Data with Personal IDS A L Data handled as with Primary Care Data IDs Data to High Security IC Data Centre Linker file Only keys and ID derivations Ch NHS Domain (Data Source) CoE CPRD Trusted Linkage Domain CPRD Data Domain
CPRD (MHRA) Summary Data Stewardship Map Outbound (Data Provision) CPRD Caldicott uardian Data CPRD Data Domain End to End I Policy MHRA/CPRD Senior Information Risk Officer CPRD Information Asset Owner Data to Research Data Store With Pseudo ID2 In High Security Data Centre 1 Data only accessible to Limited Documented/Controlled Users CoE L I Tool- Kit & I Management A BMA/RCP Joint IT committee for whole process Health Research Authority Confid. Advisory Committee (CA) Ethics (REC) approval MHRA Independent Scientific Advisory Committee (ISAC) approval RSA Key + Certificate only computers Joint work on disclosive potential L A CPRD old Research Organisations Small cell Policy Research Users CPRD Other data Data Cutting Researcher Domain CPRD checks Research Output Research Output report
EU DATA PROTECTION CHANES???? Research as we know it will not be possible BIAS, 25% non responders ive patients option of Dissent rather than Consent In 25 years NO known disclosures Existing DP Acts for Health OK.
THANK YOU! John. parkinson@cprd.com