Skip navigation

The PCDQ (Primary Care Data Quality) Programme


The Primary Care Data Quality Programme (PCDQ) is helping primary care groups to make sense of local data and information systems as they look for ways to improve care for their cardiac patients. This paper describes the process developed by the team, progress so far, and some of the lessons from the work. The programme is based at St George's Medical School in South London.

Background to the programme

Ischaemic Heart Disease affects up to 6% of the population [1]. There is hard evidence that such patients benefit from optimal secondary and primary prevention measures [2]. It is impossible to know whether these interventions are being carried out if quality data is not being recorded and fed back. The interventions have been included in the recently published National Service Framework for Coronary Heart Disease [3]. The latter has focused the minds of Primary Care Organisations (PCOs) on to questions about how this agenda might be delivered [4].

The overall aims of the programme are:

Table 1 provides a detailed description of the specificeducational objectives for the programme.

Table 1 - Educational Objectives


1. To value clinical data
2. Address Clinical Governance agenda across PCO
3. Encourage openness and sharing of relevant Clinical Governance and data quality issues.


1. All clinicians should be able to code effectively within their clinical system.
2. Skill to download or have supplied the MIQUEST processor.
3. Skills to run the queries within individual practices, supported by PCDQ team.
4. Develop the skills to give feedback, both within individual practices & within large group meetings.
5. To understand the principles of data validation.


1. To be able to describe the clinical governance cycle.
2. To define the key elements of the core data set.
3. To understand Read clinical terms, and its migration to SNOMED/CT (Systematised and Nomenclature of Medicine - Clinical Terms).
4. Knowledge of key targets within the National Service Framework for coronary heart disease.
5. To understand the concept of numbers needed to treat and confidence intervals as methods for describing absolute risk.
6. Understanding MIQUEST functionality and how data is anonymised.
7. Knowledge of some of the relevant theories about change management.

Five Elements of PCDQ

The programme [5] provides a mechanism for delivering this agenda, by supporting the Clinical Governance Lead and the PCO as a whole with:

1. Anonymised data extraction from practices using MIQUEST (Morbidity Information Query and Export Syntax),

2. Processing that data to make it informative, and presenting it at twice yearly meeting to the whole PCO.

3. Running local MIQUEST searches in the practices. These provide lists of patients that need to be targeted for intervention. These lists can be produced by GP if needed. "Audit on a plate"

4. Supporting education about the evidence, and providing Read Code training as required by the PCO.

5. Participation of the PCO in the future development of the PCDQ programme. This occurs via twice yearly workshops.

The Programme Team is a small research group based at the Department of General Practice & Primary Care at St George's Hospital Medical School. It has four years experience of MIQUEST (formerly as the South Thames Miquest Project [6]). The team consists of two GPs who hold academic posts at St George's and two support staff experienced in General Practice & IT. For the South Thames MIQUEST Project working with one PCG funding came from a consortium in the NHS Executive South East Region. The funding for the other two PCGs was provided by MSD.

1. Using MIQUEST

MIQUEST software was commissioned by the NHS to allow standardised data extraction from a variety of proprietary GP computer systems [7]. The software is unfriendly and not for the faint-hearted. At present EMIS, Meditel, Vamp Vision and Micromedic have working MIQUEST interpreters. Torex Premiere and others are currently developing interpreters. All GP Clinical systems will eventually need to be MIQUEST compliant to achieve official accreditation [8].

2. Presenting analyses to the PCO

The data goes through sophisticated processing so that its presentation has maximum impact. Data is presented to give comparisons between practices and to show individual practice data. The team prefer to present this data at a meeting of the whole PCO. They believe the group dynamic is important in stimulating change. See Figure 1.

3. Local MIQUEST searches

These are run in the individual practices. The output is a list of patients who need an intervention, listed by GP. This patient identifiable data is not removed from the individual practices where it is run.

4. Education

The programme aims to be educationally focused. The style is non-judgemental. The "good-guys" are those who get involved in improving data quality and eventually patient care. There are set objectives to achieve in collaboration with colleagues in the programme.

5. Developing and Evaluating PCDQ

PCDQ is a programme for change. It has achieved change: improved prevalence, more aspirin and statins prescribed, more blood pressure and smoking status recorded in IHD. The team needs to work with colleagues to discover the elements that represent solutions to problems, as well as the lessons learnt. They actively take structured feedback. Evaluation is being carried out using an Action Research methodology [9].

A Working Process

The process of working with primary care groups involves four steps: preparation, data extraction, analysis and feedback.

1. Preparation

Contact is made with practices and information gathered concerning GP computer system usage etc. The participating practices are then advised and assisted with the installation of the MIQUEST interpreter - this is the software made by the various GP system suppliers to enable practice computers to run MIQUEST searches. This is followed by final checks on system readiness. PCDQ can start work when either the whole PCO has MIQUEST compliant practices or when a proportion does. It can be done both ways, and in the latter situation added in new practices, as they become MIQUEST capable.

2. Data extraction

Pre-prepared customised discs are brought to the practices and the first data collection is performed, extracting only anonymised data. The only limitation within the primary care organisation is the inability to extract data from non-MIQUEST compliant practices. If the PCO could obtain the data by other means then it could be included in the analysis.

3. Data processing and analysis

The data is analysed using a series of additional calculations from the MIQUEST output files. This output is displayed in a format, which readily illustrates important absolute and comparative values from all IHD patients within each practice. Figure 2 provides an illustration of the raw data.

Figure 2: Example of raw data











































4. Feedback

The programme centres on six monthly meetings with the whole primary care group, with every practice represented at these meetings. The meetings consist of an educational programme and the presentation of pooled, anonymised data. The practices are not identified and the group can choose to share, or not, their identify among the group.

The data is fed back to the practices in a way that will convey maximum meaning. Initially the age/sex profiles of patients in the practices are presented so that the whole PCG is aware of practices that might have an unusual age/sex profile (e.g. a practice that looks after a University would probably have a low prevalence of IHD). The team shows the age/sex profile for coronary heart disease in each practice. Unusual variations often represent miscoded patients. They look at the control of blood pressure and demonstrate the time interval since the last recorded blood pressure.

Next they look at patients with a diagnosis of IHD but with no record of aspirin prescription. The "query-set" is complex. It excludes or includes patients with an Aspirin contraindication, an indication that it is obtained OTC (Over the Counter), or currently receiving Warfarin. Patients cholesterol, smoking habits, and whether smoking advice has been given are examined. Figure 3 provides an example of how data is presented at feedback meetings.

The thresholds presented to the PCO can be varied to meet their wishes - e.g. to present BP control as being below 160/90 or 145/85 or any other combination the PCO chooses to regard to indicate poor control. Not only can the approach be adjusted accordingly, as the data presented is being secondarily processed, but it can come back to the next feedback session and re-present data previously collected around a different threshold. For example a PCO may choose to start with a target cholesterol level of 5.0mmol, but new evidence or change in policy may make next years target 4.8 mmol. Re-processed old data can show progress against the new target.

Local Queries

Between the monthly meetings local queries are run in individual practices. While these queries are running, the data collectors are able to give one to one basic Read code training to any members of the team who wish it. The local query run produces patient identifiable data. These are never removed from practice premises. The local queries produce lists by registered doctor of patients who require intervention.

Read code training

The team also produces guidelines on how and why to code information (customised for each of the major versions of the GP computer suppliers). In addition, ideal data sets are suggested. However queries are wide ranging. They are designed to trawl widely, though all the Read Codes likely to have been used to maximise the identification of CHD patients, even when ideal Read codes have not been used. E.g. A patient coded as having had a coronary artery bypass grafting will be included in the IHD search (even though this in coding terms is an "operation" and not part of the IHD coding hierarchy).

Programme timetable

The programme is designed to run over three years - see Table 2. It seeks a rate of change that busy clinicians cannot merely cope with, but can actively engage in. The work can be slowed or accelerated to fit in with the wishes of the individual primary care organisation. The team wishes to work alongside more PCOs to develop the programme further. Although it has to be self-funding it does deliver the PCO the disease register needed for the CHD NSF as well as provides an opportunity to network with the other PCOs in the programme.

Table 2: PCDQ Programme

Year 1

0 months Introductory Meeting

1. Defining the Population

2. Getting the big four right (Cholesterol, BP, Aspirin, Smoking)

2 weeks Building PCO Profile
3 weeks System preparation
1 month First data collection
6-8 weeks First presentation - PCDQ I
10 weeks Local queries
6 months Second data collection
6-7 months Second presentation - PCDQ II
7-8 months Local queries

Year 2

1 year Third data collection
Tackling the co-morbidities (B-blockers post MI, anticoagulants in AF, ACE in CHF)   Third presentation - PCDQ III
  Local query run
18 months Fourth data collection
  Fourth presentation - PCDQ IV
  Local query run
2 years Fifth data collection

Year 3

Primary prevention (high risk groups) 3rd year Risk calculation for diabetics, hyperlipidaemia, hypertensives
  Vth and VIth presentations


Results so far

First results are derived from three PCGs. The team has looked at the mean change per practice and the range of change at the practice level across all the PCGs. The data is collected from 25 practices with a combined population of 216 000 patients. The team has completed a first analysis at the individual practice level to see whether there are different characteristics among the practices that were able to change.

Across the first PCGs there is a measurable improvement in recording of the diagnosis, measurement and control of blood pressure, prescribing of aspirin and absolute increase in the number of statin prescriptions. So far the team has worked for over 6 months with 3 PCGs and has started work with five others. See Figure 4 - Change in IHD recording and management.

The percentage of patients with a computer record of IHD rose in most practices, the practices in which it fell either had trawled adult notes to look for confirmation or found that erroneous diagnoses had been made in the under 40s age group. Blood pressure recording rose. It rose least in those practices that had the biggest rise in IHD prevalence. The increase in percentage of IHD patients on statins represented over 120 new patients started on statins in each of the PCGs. A smaller number of patients were newly started on Aspirin. The fall in percentage of aspiring prescribing to IHD patients occurred like the BP recording in those practices with the biggest increase in IHD prevalence.

Smoking data has been collected but the recording habits are so inconsistent that the results are not presented here. This is an area for further development or to seek consensus of what should be recorded, and what feedback mechanism is of most use to PCO clinicians and managers.


It can be argued that the increase measured is only a change in data-quality and that care already given is being recorded electronically for the first time. This accounts for some of the change, but as repeat prescribing was largely electronic in all the practices - how can the new statin and aspirin prescriptions be explained if not representing an improvement in care?

Primary care is currently exposed to multiple exhortations to improve secondary and eventually primary prevention in coronary artery disease. It is possible that the change here is due to these external stimuli and not related to the PCDQ programme. Only a controlled trial would tell. However, as the evidence about interventions and what practitioners and patients want is changing it was decided that it would be better to allow changes to evolve. Flexibility is built into the programme, and will, for example, allow the team to modify the presentation of smoking data to see if it will affect change. The team believes the narrow focused area of work, as well as the commitment to a sensible pace of change, are key to the success of the programme.

Involvement with PCDQ cannot be a complete solution. Moreover, there is no preclusion to member practices also being part of other data collection or data quality schemes. The team hopes that the learning environment they are creating within and between PCOs will enable the programme to develop.


PCDQ has achieved change in the first three PCOs that they have worked with. More patients are recorded as having IHD, more of them have electronic recordings of their blood pressure, and there are absolute increases in the number of patients on aspirin and prescribed statins. They hope that this improvement will be sustained. The team has an in-built rolling, formative evaluation methodology to ensure that they identify the solutions that the project provides, and learn the lessons where it needs to be improved or changed.

There are now 23 PCOs using the system (April 2002)

For more information contact

Claire Yates

Department of General Practice

St George's Hospital Medical School

London SW17 ORE

Telephone: 020 8725 5661

Fax: 020 8767 7697

Mobile: 07971 989237


PCDQ Website:


1 Foot A. Taking Cardiac Care - being part of SUCEED. Impact Sept 2000.

2 NNTs for Cardiac Interventions. Bandolier.

3 Department of Health. The National Service Framework for Coronary Artery Disease. 2000.

4 McColl A, Roland M. Knowledge and information for clinical governance. BMJ 2000;321:871-874

5 PCDQ (Primary Care Data Quality Project) - General Practice, St George's Hospital Medical School.

6 STMP (South Thames MIQUEST Project)

7 Official MIQUEST download site.

8 NHS Information Authority, Requirements for Accreditation of GP computer systems.

9 Lau F. A review of the Use of Action Research in Information System Studies. In Lee AS, Lienbenau JL (eds) Information Systems Research: Information systems and Qualitative Research. Cahpan & Hill London 1997:31-68. URL: This is being done using an Action Research methodology.