'Use of linked health care data for research: experiences with the Hampshire Health Record' - Paul Roderick

Use of linked health care data for research: experiences with the Hampshire Health Record (HHR) P RoderickAcademic Unit of Primary Care and Population Sciences, University of Southampton, UK Contact: [email protected]

Thank you

Research group goes back 2010 Range of work grown CLAHRC*

AcknowledgementsHugh Sanderson (founding clinical director) NHS South, Central and West Commissioning Support Unit and the Hampshire Health Record Information Governance Group Elen Hall Matt Johnson (UoS analyst)

Colleagues*

Hampshire Health RecordAnalytical Database *Pseudonymised, individual, linked extracts of primary and secondary care records

Data from about 130 general practices in Hampshire. Total registered population about 13 million people (80%)

Laboratory data from Portsmouth and Southampton hospitals (600k)

Hampshire Health Record

Started 2003-4 as Electronic Hampshire clinical Repository PASComm hlth Southern1 ccg social careGoverance ..opt out, widely advertised, and one register 6k/2.85m 0.2% StricUses Structured Query Language (SQL) - programming language designed for managing data held in a relational database management system t rules of access

Uses Epidemiology of conditions Frequency, determinants, processes of care, outcomes, costs Risk prediction Evaluation of interventionsPrimary or secondary care (natural expts)Benefits of HHR Linkage to existing studies; trial feasibility, recruitment Whole population (Social care, housing)

Understanding Variations in Outcome in COPD: Early Results of an Observational Study using Routine Clinical DataL Josephs, M Johnson, P Roderick, D M Thomas

GP=hosp*

Methods Retrospective observational study using routine individual patient-anonymised dataSelected Read Codes used to define and characterise a prevalent cohort with a practice diagnosis of COPD as at 31/12/102 year follow up of outcomes:Hospital admissions (respiratory)A/E attendances (respiratory)Mortality

*

*

Smoking status recorded in 21,068 patients (99.2%)37.8% current smokers51.0% ex-smokers10.4% never smokers

*Smoking

*

ConclusionsIn a broad unselected UK primary care COPD population, highlights:high proportion of patients still smoking (>1 in 3)poor prognosis of COPD: one in ten patients died during the 2 years, a third of with a hospital admission

*

Antibiotic use in care homes High use in Nursing homes especially for Urinary tract infection (UTI)Why--overtreatment for non specific symptoms; asymptomatic bacteruria common Retrospective cohort 2012. Postcode used to identify nursing home. Sample 8.2% of 1.24 million >75, of whom 7.3% in care homes.

P Sundvall BMC Geriatrics 2015

Antibiotic prescriptions/100Adjusted odds ratio for NH UTI 2.2 if no catheter, 1.4 if catheter

HHR=A with CDSC re MSRA surveillanceLinkage hosp to community

Chronic kidney disease (CKD) Identified people with prevalent (2008) and incident CKD stage 35 between 2008 and 2011 was identified from the UK Hampshire Health Record (HHR) using eGFR values.

Two values of eGFR 60 mL/min per 173 m or no previous eGFR value)

QOF-registered CKD identified by relevant Read codes.

S Fraser et al BMC Family Practice 2015

CKD Results 88 practices with continuous pathology records 2008-2013

Total over 18 population =498,631

Prevalent CKD at end 2008 =24,021 (4.8%)

Incident CKD 2008-2013 =15,736

QoF register and Urinary ACR testing

Incident cohort:46% had a record of ever having an uACR63% QoF registered

Ckd reg assoc low egfr female dmhtcvdTimely acr assoc reg dm higher egfrTIMELY cKD vreg and ACr poor*

The Hampshire AKI StudySimon FraserPaul RoderickMark UniackeMatthew JohnsonBorislav DimitrovDavid CullifordLily Yao

*

Incidence and consequencesof community and hospital AKIIncidence and consequencesof community and hospital AKIPREPOSTValidate community AKI prediction tool

HHRStrengthsPopulation basedUsed for individual care Laboratory dataScope for extension Geography, dataWider edu/env/housing Free text

LimitationsData qualityMissingenter, leave Exclusions from HHR No cause of deathNo microbiology/radiology dataLogistics size, coding

ExclusionsSTD, ToP, Abuse , IVF, marital, prison

Practical issues re size of databases (AKI), complex algorhythms based on biochem, Missing ethnicity, ACR

Ongoing

EPIDEMIOLOGY /VARIATION Dementia, Liver disease pathways Cancer survivor prevalence Multimorbidity, frailty, treatment burdenMaternity-child health Fuel poverty EVALUATIONAtrial fibrillation detection (Watch BP), NHS Vascular ChecksAlcohol detoxification Benefits of HHR per se

STREAMBr Ca*

HHR Users I think it is fantastic and use it all the time. - GPHaving access to results and reports from other hospitals saves time and resources. Hospital consultant.help make an informed decision; provides the patient with confidence. - GPPossibly the Hampshire Health Record saved her life - GP it enables me to work more closely with consultants in developing plans of care - Nurse SpecialistThe patient is prescribedthe correct and appropriate medication as soon as possible. - Community Pharmacist

*

*

Thank you

Research group goes back 2010 Range of work grown CLAHRC*Colleagues*Started 2003-4 as Electronic Hampshire clinical Repository PASComm hlth Southern1 ccg social careGoverance ..opt out, widely advertised, and one register 6k/2.85m 0.2% StricUses Structured Query Language (SQL) - programming language designed for managing data held in a relational database management system t rules of access GP=hosp*

*

*

*

*HHR=A with CDSC re MSRA surveillanceLinkage hosp to community Ckd reg assoc low egfr female dmhtcvdTimely acr assoc reg dm higher egfrTIMELY cKD vreg and ACr poor**

ExclusionsSTD, ToP, Abuse , IVF, marital, prison

Practical issues re size of databases (AKI), complex algorhythms based on biochem, Missing ethnicity, ACR STREAMBr Ca*

*

*