Primary Care data signposting Outline - SPCR · Primary Care data signposting CPRD, THIN and other databases Evangelos (Evan) Kontopantelis1 1Institute of Population Health, University

Primary Care data signpostingCPRD, THIN and other databases

Evangelos (Evan) Kontopantelis1

1Institute of Population Health, University of Manchester

Oxford, 21 Sep 2015

Kontopantelis (IPH) Data signposting 21 Sep 2015 1 / 33

Outline

1 Primary Care DatabasesCoverage and numbersStructureour approachtoolsresults

2 General Practice datasets

3 OtherPopulation datasetsHospital episode statisticsLinking and mapping


The Clinical Practice Research DatalinkCPRD

Established in 1987, with only a handful of practicesSince 1994 owned by the Secretary of State for HealthIn July 2012:

644 practices (Vision system only: in Eng mainly London, SE, SC,NW, WM; see /pubmed/23913774)13,772,992 patients (≈5m active)covering ≈7.1% of the UK population

Access to the whole database is offered and costs ≈£130,000 paOffers the ability to extract anything adequately recorded inprimary care and construct a usable dataset


The Health Improvement Network databaseTHIN

Established in 2003 as a collaboration between In PracticeSystems Ltd and CSD Medical Research UK (EPIC)Now part and parcel of UCLIn May 2014:

562 practices (Vision system only, 50-60% overlap with GPRD)11.1m patients (3.7m active)covering ≈6.2% of the UK population

Usually offered under a 4-year license which costs £119,000Similar structure to CPRD and possibly more efficient patientmatching for socio-demographic characteristics


http://www.ncbi.nlm.nih.gov/pubmed/23913774

QResearch

Collaboration with the University ofNottinghamIn May 2014 reports:

754 practices (EMIS systems: biggestUK provider)over 13m patients (??m active)covering ≈7% of the UK population?

Datasets limited to 100k patients forexternalsPublication list, 90-95%: Vinogradova,Coupland and/or Hippisley-Cox


ResearchOne

Collaboration between TPP and the University of LeedsIn May 2014 reports:

??? practices (SystmOne: Yorkshire&H, East Mid, East Eng, NE)GP, Community Care, Hospital Care.30m research recordscovering ≈?% of the UK populationcosts?

New potentially important playerUniformity of SystmOne and central databases for TPP systemslikely to provide better quality data at lower cost


GP clinical systemsQuality of care and choice of clinical computing system, BMJ Open 2013

North East

North West

London

West Midlands

Yorkshire & the Humber

South West

East Midlands

East of England

South Central

South East Coast

(90.2,90.5](89.9,90.2](89.6,89.9](89.3,89.6](89,89.3](88.7,89][88.4,88.7] EMISIn Practice SystemsTPPMicrotestISoft

NOTE: Chart size proportional to number of practices in area

Average practice scores by Strategic Health Authority, 2010−11

Overall reported achievement (62 indicators)and GP systems suppliers

North East

North West

London

West Midlands

Yorkshire & the Humber

South West

East Midlands

East of England

South Central

South East Coast

(5.5,5.7](5.3,5.5](5.1,5.3](4.9,5.1][4.7,4.9] LVVision 3ProdSysOneXPCSSynergyPractice ManagerPremiere

NOTE: Chart size proportional to number of practices in area

Average practice scores by Strategic Health Authority, 2010−11

Overall exception reporting (62 indicators)and GP systems products


Export formatfrom SQL

Broken down to numerous tables, due to volume of the dataText files need to be imported into powerful analysis/databasemanagement softwareSome of the information available:

Patient birthyear, sex, marital status, smoking/drinking status,height, weight and BMIClinical, referral, therapy, test and immunisation events

All events are entered in codes (lookup tables available)Everything (likely to be recorded by a GP) can be identified,provided one knows which codes to look for and in which tablesBUT a manual search on all the codes is not possible andautomated processes are required


Primary Care Databases structurebased on CPRD

Event filesClinical: all medical history data (symptoms, signs and diagnoses)Referral: information on patient referrals to external care centresImmunisation: data on immunisation recordsTherapy: data relating to all prescriptions issued by a GPTest: data on test records

Look-up filesMedical codes: READ codes, ≈100k availableProduct codes: ≈80k availableTest codes: ≈300 available


Diabetes example

Size of the tables prohibits looking at codes one by oneInstead we use search terms to identify potentially relevant codesin the look-up tables and create draft lists

Example (Search terms for diabetes)String search in Medical codes: ’diab’ ’mell’ ’iddm’ ’niddm’READ code search in Medical codes file: ’C10’ ’XaFsp’String search in Product codes file: ’insulin’ ’sulphonylurea’’chlorpropamide’ ’glibenclamide’


Diabetes example...continued

Clinicians go through the draft lists and select the relevant codesThree sets of codes are created, corresponding to:

QOF criteriaConservative criteriaSpeculative criteria

Using the finalised code lists we search for events in the Clinical,Referral, Immunisation, Therapy and Test filesProcess involves much work in code writing, hence use of anappropriate statistical package like Stata or R is essential


Primary Care Databases toolsCPRD/THIN based but applicable to all

Search commandspcdsearch in Stata and Rpcdsearch in Rcode list extraction algorithmModelling conditions and health care processes in Electronic Health Records: an application to Severe Mental

Illness with the Clinical Practice Research Datalink, under review

Code listsclinicalcodes.orgWebsite with freely available developed code listsClinicalCodes: An Online Clinical Codes Repository to Improve the Validity and Reproducibility of Research

Using Electronic Medical Records, PLOS ONE 2014

Data extractionrEHR (github.com)R package for manipulating and analysing EHR datarEHR: An R package for manipulating and analysing Electronic Health Record data, under review


Primary Care Databases toolsCPRD/THIN based but applicable to all

Power calculationsipdpower in Statamixed-effects power calculation through simulationsSimulation-Based Power Calculations for Mixed Effects Modelling: ipdpower in Stata, JSS in print

Cleaning BMImibmi in StataCleaning and multiple imputation for missing BMI dataLongitudinal multiple imputation approaches for Body Mass Index: the mibmi command, Stata Journal under

review

General Multiple imputationtwofold in StataMultiple imputation for longitudinal datasetsApplication of multiple imputation using the two-fold fully conditional specification algorithm in longitudinal clinical

data, Stata Journal 2014


Non-incentivised aspects of careSample of 148 representative practices from the CPRD

Achievement rates improvedfor most indicators in thepre-incentive periodSignificant initial gains inincentivised indicators but nogains in later yearsNo overall effect onimprovement rate for nonincentivised aspects in 2004-5But by 2006-7 achievementrates significantly below thosepredicted by pre- trends

Figure

Mean achievement rate of 148 general practices for quality of care indicators from 2000-1 to 2006-7. Performance indicatorsgrouped by activity and whether they were incentivised under the QOF scheme, which came into force from 2004-5. (Themean rate is the mean of the adjusted means for the individual indicators within each group)

Reprints: http://journals.bmj.com/cgi/reprintform Subscribe: http://resources.bmj.com/bmj/subscribers/how-to-subscribe

BMJ 2011;342:d3590 doi: 10.1136/bmj.d3590 Page 12 of 12

RESEARCH

Effect of financial incentives on incentivised andnon-incentivised clinical activities: longitudinalanalysis of data from the UK Quality and OutcomesFrameworkTim Doran clinical research fellow1, Evangelos Kontopantelis research associate1, Jose M Valderasclinical lecturer 2, Stephen Campbell senior research fellow 1, Martin Roland professor of healthservices research3, Chris Salisbury professor of primary healthcare4, David Reeves senior researchfellow 1

1National Primary Care Research and Development Centre, University of Manchester, Manchester M13 9PL, UK; 2NIHR School for Primary Care

Research, Department of Primary Health Care, University of Oxford, Oxford OX3 7LF; 3General Practice and Primary Care Research Unit, University

of Cambridge, Cambridge CB2 0SR; 4Academic Unit of Primary Health Care, University of Bristol, Bristol BS8 2AA

AbstractObjective To investigate whether the incentive scheme for UK generalpractitioners led them to neglect activities not included in the scheme.

Design Longitudinal analysis of achievement rates for 42 activities (23included in incentive scheme, 19 not included) selected from 428identified indicators of quality of care.

Setting 148 general practices in England (653 500 patients).

Main outcome measures Achievement rates projected from trends inthe pre-incentive period (2000-1 to 2002-3) and actual rates in the firstthree years of the scheme (2004-5 to 2006-7).

Results Achievement rates improved for most indicators in thepre-incentive period. There were significant increases in the rate ofimprovement in the first year of the incentive scheme (2004-5) for 22 ofthe 23 incentivised indicators. Achievement for these indicators reacheda plateau after 2004-5, but quality of care in 2006-7 remained higherthan that predicted by pre-incentive trends for 14 incentivised indicators.There was no overall effect on the rate of improvement fornon-incentivised indicators in the first year of the scheme, but by 2006-7achievement rates were significantly below those predicted bypre-incentive trends.

Conclusions There were substantial improvements in quality for allindicators between 2001 and 2007. Improvements associated withfinancial incentives seem to have been achieved at the expense of smalldetrimental effects on aspects of care that were not incentivised.

IntroductionOver the past two decades funders and policymakers worldwidehave experimented with initiatives to change physicians’behaviour and improve the quality and efficiency of medicalcare.1 Success has been mixed, and attention has recently turnedto payment mechanism reform, in particular offering directfinancial incentives to providers for delivering high qualitycare.2 In 2004 in the UK the Quality and Outcomes Framework(QOF) was introduced—a mechanism intended to improvequality by linking up to 25% of general practitioners’ incometo achievement of publicly reported quality targets for severalchronic conditions.3

Should these incentives succeed, the potential benefits forpatients with the relevant conditions are considerable.4 Incentivesmight also improve general organisation of care, benefitingprocesses and conditions beyond those covered by theincentives.5 Financial incentives have several potentialunintended consequences, however. For example, they mightresult in diminished provider professionalism, neglect of patientsfor whom quality targets are perceived to be more difficult toachieve, and widening of health inequalities.6 7 Doctors mightalso focus on the conditions linked to incentives and neglectother conditions8 or, where certain activities are incentivisedwithin the management of a particular condition, might neglectother activities for patients with that condition.Practices in England generally performed well on incentivisedactivities in the first year of the UK incentive scheme, andoverall performance improved over the next two years.9-11 It is

Corresponding to: Tim Doran [email protected]

Extra material supplied by the author: Methodological appendix (see http://www.bmj.com/content/342/bmj.d3590/suppl/DC1)

Reprints: http://journals.bmj.com/cgi/reprintform Subscribe: http://resources.bmj.com/bmj/subscribers/how-to-subscribe

BMJ 2011;342:d3590 doi: 10.1136/bmj.d3590 Page 1 of 12

Research

RESEARCH


Patient level diabetes careSample of 148 representative practices from the CPRD

In 2004-5 quality improvedover-and-above thispre-incentive trend by 14.2%By 2006-7 improvement abovetrend smaller at 7.3%Levels of care variedsignificantly for sex, age, yearsof previous care, number ofco-morbid conditions

Recorded quality of primary care forpatients with diabetes in Englandbefore and after the introductionof a financial incentive scheme:a longitudinal observational study

Evangelos Kontopantelis,1 David Reeves,1 Jose M Valderas,2,3

Stephen Campbell,1 Tim Doran1

▸ An additional data ispublished online only. To viewthis file please visit the journalonline (http://bmjqs.bmj.com)

1Health Sciences Primary CareResearch Group, University ofManchester, Manchester, UK2Health Services and PolicyResearch Group, NIHR School forPrimary Care Research,Department of Primary HealthCare, University of Oxford,Oxford, UK3European Observatory of HealthSystems and Policies, LondonSchool of Economics, London,UK

Correspondence toDr Evangelos Kontopantelis,Health Sciences Primary CareResearch Group, 5th floorWilliamson Building,Manchester M13 9PL, UK;[email protected]

Received 29 March 2012Revised 26 June 2012Accepted 12 July 2012Published Online First22 August 2012

To cite: Kontopantelis E,Reeves D, Valderas JM, et al.BMJ Qual Saf 2013;22:53–64.

ABSTRACTBackground The UK’s Quality and OutcomesFramework (QOF) was introduced in 2004/5,linking remuneration for general practices torecorded quality of care for chronic conditions,including diabetes mellitus. We assessed theeffect of the incentives on recorded quality ofcare for diabetes patients and its variation bypatient and practice characteristics.Methods Using the General Practice ResearchDatabase we selected a stratified sample of 148English general practices in England, contributingdata from 2000/1 to 2006/7, and obtained arandom sample of 653 500 patients in which23 920 diabetes patients identified. Wequantified annually recorded quality of care atthe patient-level, as measured by the 17 QOFdiabetes indicators, in a composite score andanalysed it longitudinally using an InterruptedTime Series design.Results Recorded quality of care improved for allsubgroups in the pre-incentive period. In the firstyear of the incentives, composite qualityimproved over-and-above this pre-incentive trendby 14.2% (13.7–14.6%). By the third year theimprovement above trend was smaller, but stillstatistically significant, at 7.3% (6.7–8.0%). After3 years of the incentives, recorded levels of carevaried significantly for patient gender, age, yearsof previous care, number of co-morbidconditions and practice diabetes prevalence.Conclusions The introduction of financialincentives was associated with improvements inthe recorded quality of diabetes care in the firstyear. These improvements included somemeasures of disease control, but most capturedonly documentation of recommended aspects ofclinical assessment, not patient management oroutcomes of care. Improvements in subsequent

years were more modest. Variation in carebetween population groups diminished underthe incentives, but remained substantial in somecases.

INTRODUCTIONIn the last 15 years the National HealthService in the UK has undergone a series ofreforms aimed at improving the quality ofcare for people with chronic conditions.These include the creation of the NationalInstitute for Health and ClinicalExcellence, and the introduction ofNational Service Frameworks which setminimum standards for the delivery ofhealth services in specified clinical areas,including diabetes mellitus.1 The quality ofprimary care generally, and of diabetes carein particular, improved in the early 2000s,2

partly in response to these quality improve-ment initiatives.3 In 2004 new contractualarrangements for family doctors allowedthem to opt out of out-of-hours care andlinked financial incentives to quality ofcare under the Quality and OutcomesFramework (QOF),4 the largest and mostambitious pay-for-performance schemeever attempted in health care.5 6

The QOF initially included 76 clinicalindicators, covering a range of processes ofcare (eg, measurement of blood pressure)and intermediate outcomes (eg, glycaemiccontrol). A further 70 indicators coveredaspects of practice organisation and patientexperience of care. Eighteen of the clinicalindicators related to care for patients withdiabetes, reflecting in part the nationalimportance of the disease, the recordedprevalence of which had increased by 75%

ORIGINAL RESEARCH

Kontopantelis E, et al. BMJ Qual Saf 2013;22:53–64. doi:10.1136/bmjqs-2012-001033 53

group.bmj.com on September 13, 2013 - Published by qualitysafety.bmj.comDownloaded from

Recorded QOF care did not vary significantly by areadeprivation before or after the introduction of theincentivisation scheme. However, the effect of the inter-vention did vary with area deprivation: patients attend-ing practices from the most deprived quartile appear tohave gained less from the intervention compared withpatients in the most affluent quartile of practices, by4.9% in 2004/5 and 3.8% in 2006/7.There was significant variation in recorded QOF care

by practice diabetes prevalence rates, but the differenceswere small and diminished over time. The interventioneffect also varied with practice diabetes prevalence.Compared with practices in the first quartile (lowest dia-betes prevalence), the QOF effect was larger for practicesin the second and third quartiles—by 1.4% and 2.1%respectively in the short term (2004/5) and by 3.2% and4.8% respectively in the long term (2006/7).

Sensitivity analysesThe sensitivity analysis (based on untransformed data)agreed with the main analysis in all respects except forthe relationship between the intervention and patientgender. In the sensitivity analysis both the short- andlong-term impact of the pay-for-performance schemewas significantly larger, though small in scale, for femalepatients (1.2%, p=0.048 and 2.5%, p=0.01 respectively).

DISCUSSION

The main aim of pay-for-performance schemes is toincentivise physicians to provide high quality care andthereby improve patient outcomes. Research to datesuggests that pay-for-performance schemes have limitedimpact when implemented in isolation, but when sup-ported by other quality improvement initiatives canhave a positive effect on quality of care.3 We foundthat recorded quality of primary care in the UK, as

measured by the QOF diabetes indicators, was alreadyimproving prior to the introduction of the scheme in2004, and that it improved at an accelerated rate inthe first years of its implementation, supporting thefindings of previous studies.11 23 However, this acceler-ated improvement did not seem to benefit all popula-tion groups equally.

Strengths and limitations of the studyThe main strength of our study was in the use of datafor individual patients drawn from a nationally represen-tative sample of practices. However, the study is subjectto certain limitations. First, the QOF was introduced uni-versally and was not implemented as part of a rando-mised experiment. The lack of a practice control groupentails that analyses of its effect in quality of care areonly possible using quasi-experimental methods. Resultsobtained with these methods can be method- andassumptions-sensitive; however, the interrupted time-series design is one of the most effective and powerfulof all quasi-experimental designs and is routinely beingused as the best possible approach when such researchscenarios arise.24 Second, we are reliant on the accurateand consistent recording of data by practices; however,usage of clinical computing systems has changed overtime and practices may have exaggerated their perform-ance in response to the financial incentives. This studyhas investigated the quality of recorded diabetes care andthere may be differences with care actually delivered.On the other hand, improved measurement is a neces-sary prerequisite for improved quality of care and onecould argue that it is improved quality of care. Third,most of the measures refer to documentation levels anddo not necessarily lead to the intended improvements inoutcomes if not properly followed-through or the inter-ventions are not offered in an appropriate manner (eg,advising a patient briefly ‘perhaps you should considerquitting smoking’ in order to ‘tick’ the relevant QOFbox is not really a smoking cessation intervention).Fourth, some quality indicators are dependent onothers, for example, indicator DM12 (blood pressurecontrolled) cannot be met unless indicator DM11(blood pressure measured) has also been met. However,we aimed to quantify and assess overall quality of care asmeasured by the whole diabetes domain in the QOFand to be as inclusive as possible. Fifth, the conditionswe modelled to investigate the effect of co-morbiditywere not an exhaustive list and only the presence orabsence was modelled and not the severity of each con-dition. However, the number of chronic co-morbidities isa well-established marker of the overall clinical complex-ity of a patient.25 Sixth, our findings assume that theobserved trends in indicator achievement prior to QOFwould have continued unchanged had the incentive

Figure 2 Aggregate patient level Quality and Outcomes

Framework care and predictions based on the

pre-incentivisation trend.

BMJ Qual Saf 2012;0:1–12. doi:10.1136/bmjqs-2012-001033 9

Original research

group.bmj.com on September 14, 2012 - Published by qualitysafety.bmj.comDownloaded from

2000/1 2001/2 2002/3 2003/4 2004/5 2005/6 2006/7new diagno 44.7 50.4 56.5 65.3 73.4 74.2 74.31-4 years 48.4 53.9 59.4 71.1 80.9 83 83.25-9 years 46.4 51.9 56.8 69.1 78.7 81.4 81.810+ years 45.4 50 55.1 66.7 77.6 79.3 80.4

2000/1 2001/2 2002/3 2003/4 2004/5 2005/6 2006/7new diagnoses 44.7 50.4 56.5 65.3 73.4 74.2 74.31-4 years 48.4 53.9 59.4 71.1 80.9 83 83.25-9 years 46.4 51.9 56.8 69.1 78.7 81.4 81.810+ years 45.4 50 55.1 66.7 77.6 79.3 80.4

40

45

50

55

60

65

70

75

80

85

90

aggr

egat

e re

cord

ed Q

OF

care

scor

e


Withdrawing incentives644 CPRD practices, 2004-5 to 2011-12

Financial incentives partiallyremoved for aspects of carefor patients with asthma, CHD,diabetes, stroke and psychosisMean levels of performancegenerally stable after theremoval of incentives, mainlyin the short termHealth benefits from incentiveschemes may be increased byperiodically replacing existingindicators with new ones

Fig 2 Trends and predictions for removed and related unremoved indicators. For indicators removed in April 2011, predictedscores were compared with back transformed observed scores (from logit). Although back transformed observed scoresagree with raw scores fully in most cases, that might not be true for indicators for which denominators are small and 100%scores are prevalent. This can lead to discrepancies due to empirical logit (that is, score at 100% is back transformed tolower score) and an “unfair” comparison between observed and predicted. Unremoved process related control indicatorswere also plotted (using raw scores as no comparison with predictions exists). Condition related control indicators werenot plotted; vertical lines indicate timing of indicator removal

No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe

BMJ 2014;348:g330 doi: 10.1136/bmj.g330 (Published 27 January 2014) Page 16 of 17

RESEARCH

Withdrawing performance indicators: retrospectiveanalysis of general practice performance under UKQuality and Outcomes Framework

OPEN ACCESS

Evangelos Kontopantelis senior research fellow 1 2, David Springate research associate 1 3, DavidReeves reader 1 3, Darren M Ashcroft professor 4, Jose M Valderas professor 5, Tim Doran professor 6

1NIHR School for Primary Care Research, Centre for Primary Care, Institute of Population Health, University of Manchester, Manchester M13 9PL,UK; 2Centre for Health Informatics, Institute of Population Health, University of Manchester; 3Centre for Biostatistics, Institute of Population Health,University of Manchester; 4Centre for Pharmacoepidemiology and Drug Safety Research, Manchester Pharmacy School, University of Manchester;5Institute for Health Services Research, UE Medical School, University of Exeter, Exeter, UK; 6Department of Health Sciences, University of York,York, UK

AbstractObjectives To investigate the effect of withdrawing incentives onrecorded quality of care, in the context of the UK Quality and OutcomesFramework pay for performance scheme.

Design Retrospective longitudinal study.

Setting Data for 644 general practices, from 2004/05 to 2011/12,extracted from the Clinical Practice Research Datalink.

Participants All patients registered with any of the practices over thestudy period—13 772 992 in total.

Intervention Removal of financial incentives for aspects of care forpatients with asthma, coronary heart disease, diabetes, stroke, andpsychosis.

Main outcomemeasuresPerformance on eight clinical quality indicatorswithdrawn from a national incentive scheme: influenza immunisation(asthma) and lithium treatment monitoring (psychosis), removed in April2006; blood pressure monitoring (coronary heart disease, diabetes,stroke), cholesterol concentration monitoring (coronary heart disease,diabetes), and blood glucose monitoring (diabetes), removed in April2011. Multilevel mixed effects multiple linear regression models wereused to quantify the effect of incentive withdrawal.

Results Mean levels of performance were generally stable after theremoval of the incentives, in both the short and long term. For the twoindicators removed in April 2006, levels in 2011/12 were very close to2005/06 levels, although a small but statistically significant drop wasestimated for influenza immunisation. For five of the six indicatorswithdrawn from April 2011, no significant effect on performance wasseen following removal and differences between predicted and observedscores were small. Performance on related outcome indicators retainedin the scheme (such as blood pressure control) was generally unaffected.

Conclusions Following the removal of incentives, levels of performanceacross a range of clinical activities generally remained stable. Thisindicates that health benefits from incentive schemes can potentially beincreased by periodically replacing existing indicators with new indicatorsrelating to alternative aspects of care. However, all aspects of careinvestigated remained indirectly or partly incentivised in other indicators,and further work is needed to assess the generalisability of the findingswhen incentives are fully withdrawn.

IntroductionAs part of wider efforts to improve the quality and efficiencyof healthcare, purchasers worldwide have experimented withlinking performance indicators to financial incentives,reputational incentives, or both, within pay for performance andpublic reporting schemes. As the clinical evidence base andpolicy priorities change over time, indicator sets must beperiodically reviewed and individual indicators modified,removed, or replaced. Within financial incentive schemes,indicators may also be removed because achievement rates havereached a ceiling, thereby allowing new indicators, for whichimprovement is possible, to be introduced.1

Incentives are intended to improve performance by changingphysicians’ behaviour, but evenwhen this approach is successfulthe change may be temporary. If the incentive is necessary tomaintain high performance levels, its withdrawal will result inlower achievement rates and a loss of performance gains. Thismay occur, for example, because better performance requiresadditional staffing resource that depends on the incentivepayments or because physicians’ expectations of reward arealtered. Depending on the nature of the incentives and the extent

Correspondence to: E Kontopantelis [email protected]

Extra material supplied by the author (see http://www.bmj.com/content/348/bmj.g330?tab=related#webextra)

No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribe

BMJ 2014;348:g330 doi: 10.1136/bmj.g330 (Published 27 January 2014) Page 1 of 17

Research

RESEARCH


Quality and Outcomes FrameworkQOF datasets

Pay for performance scheme that started in 1/4/2004Costs over £1bn paVoluntary scheme but participation over 99.9%Freely available on Health & Social Care Information Centre(HSCIC), by financial year:

NHS practice code and list sizePrevalence on 15 key chronic conditions (e.g. diabetes, asthma,CHD, COPD etc)Practice level performance on various clinical indicators for theseconditionsPractice level exception rates for each indicator


General Medical ServicesGMS datasets

Data from around 2000Information on general practicesAvailable on request (not free but cheap) from the HSCIC, bycalendar year:

NHS practice code, list size, contract type, full address (includingpostcode, sha, pct, lsoa)Number of GPs, FTE, names, country/area qualified, sex, agePatient counts by age group and sex

Part of the Workforce theme: more info for other healthprofessions


Patient SatisfactionGP Patient Survey

Data from 2007Run by Ipsos MORI, data collected twice a yearStratified random sampling of patients to collect data onsatisfaction with GP servicesData freely at the practice and higher levels, weighted (to matchpatient population) and unweighted satisfaction scores on:

access, making an appointment, waiting times speaking to GP ornurse, ease of accesslast GP and last nurse appointment, opening hours, overallexperienceand many more domains


Primary Care MortalityPCM database

Data from 2006Managed by the HSCIC andaccessible remotelyMonthly and annual extracts ofindividual record level data ondeaths supplied by ONS:

registered GP/practice,patient details e.g. age,causes of death, NHS no

Data for use by LocalAuthorities and NHSorganisations only


Census 2011 datasetsbut also 2001, 1991 etc

Information aggregated at various levels, as low as lower superoutput area (LSOA) levelFreely available from the ONS websites, including:

Counts by age groups and sexHealthEthnicityReligionOccupationQualificationsHousehold-accommodation


Deprivation datasetsIndex of Multiple Deprivation (IMD): 2004, 2007, 2010, 2014

Important covariate, available at the 2001 LSOA levelEngland only (although there is a Welsh IMD as well)Free at the Neighbourhood Statistics ONS websiteAggregate of 7 domains:

IncomeEmploymentHealth deprivationEducation and skillsHousingCrimeEnvironment

2010 range was 0.5-87.8 (9.8 and 30.2 for 25th and 75th centiles)


Mortality datasetsFrom 1998

As counts available at the LSOA level (2001 or 2011) but specialrequest to the ONS mortality teamAs standardised mortality rates freely available but at electoralward level or higher from the main ONS websiteSpecific mortality causes available:

using ICD-10 codes from 2001, ICD-9 beforecounts at the LSOA level can be broken down by sex and age-group


Admitted patient care datasetand outpatient

Data more or less available from 1989Patient-level data, with various organisational markers:

GP, SHA, PCT, site of treatmentAvailable upon request from the HSCIC, including:

patient characteristics (incl IMD), admissions, discharges,episodes, clinical, maternity, psychiatric

Additional sensitive info: dob, NHS number, patient residencepostcode, LSOA etcData for outpatient care available from 2003: similar but lessdetailed


Critical care data

Data available from 2008Add-on dataset which should be matched with inpatient dataset,on request from the HSCICIncludes:

critical care datesadmission typesupport infocritical care levelsdischarge info


Accident and Emergency data

Data available from 2007Similar covariate and organisation info as inpatient-outpatientdatasets Available upon request from the HSCIC, with info on:

attendancesclinical diagnosisclinical investigationclinical treatment

Additional sensitive info: dob, NHS number, patient residencepostcode, LSOA etc


Lookup tables

To combine datasets reported at different levelsUsually the postcode is the best start, if knownThe UK Data Service (previously UK Borders) contains tables tohelp merge data at various levels, at 1991, 2001, 2011 or 2013boundaries:

PCTsWardsLSOAsSHAsClinical Commissioning Groups (CCGs formerly PCTs)NHS Area Teams


Spatial mapping

After merging at a geographical level spatial coordinates areuseful for plotting or accounting for spatial correlations inregression analysesONS Geoportal holds various digital vector boundaries files(shapefiles) for 2001, 2011 and more recent geographies:

LSOAsPCTs-CCGsSHAsRegions


OverviewHealth Sciences related

QOF

PC Mortality Database

GMS

GP patient satisfaction

other LSOA-PCT

LSOA-SHA

Postcode-LSOA

other

Admitted Patient Care

Outpatient

Critical CareA&E

other

Census

Mortality

Deprivation

other

Terminology

Training resources

Classifications

other

SHAs

LSOAs

PCTs (CCGs)

other

CPRD

QResearch

THIN

ResearchOne


Comments and questions: [email protected]


Documents

Primary Care data signposting Outline - SPCR · Primary Care data signposting CPRD, THIN and other databases Evangelos (Evan) Kontopantelis1 1Institute of Population Health, University