Current and Future Developments in General Practice Research Database

CURRENT AND FUTURE DEVELOPMENTS IN GENERAL PRACTICE RESEARCH DATABASE

The data in the GPRD is a person-level data set with the linkage to other information currently undertaken within the primary care system by the GP and his staff. This is, in many ways, the ideal situation as the GP, or other primary care healthcare professionals, do the disease coding at the time of consultation. Increas-ingly, the information within the UK NHS is being communicated electronically using the NHS number unique for each patient. Table 27.2 lists the major changes in GPRD data collection that are happening. Laboratories are now sending biochemistry results electronically to the GP, and this information can be loaded electronically into the GP medical records. Over the years 2002–05, the amount of biochem-istry data in the GPRD has increased three fold due to increases in the number of tests undertaken, the fact that tests are grouped for common requirements even when only one result was requested and auto-matic recording of these data (with better recording of results within normal reference ranges). Figure 27.1 shows the number of laboratory results as recorded in the GPRD over calendar year. It is likely that the introduction of the QOF has also directly impacted on the recording of tests in the GPRD. The frame-work focuses on key disease areas such as diabetes, hypertension and asthma; it is expected that specific tests are conducted (and the results recorded in the electronic record) in patients with diagnoses in these key disease areas. For instance, diabetic patients are expected to have a record of HbA1c or equivalent in the previous 15 months. Recording of HbA1c in 2004 was 13% higher than in 2003; the requirement under QOF to measure and record HbA1c levels for diabetic patients makes it likely that the increase in record-ing is not simply because of the general increase in recording of test results in patient records resulting from the electronic transmission of test results.

The quality of GPRD recording of lifestyle factors such as weight, BMI, smoking and alcohol use is continuously improving and is not as reported by Ilkanoff et al. (2005) a limitation to using GPRD. The reason is that NHS has undertaken initiatives to improve data recording in GP practices and has linked quality standards in care and data recording to practice reimbursement. The GPRD currently only records prescriptions as written by the GP, but due to NHS IT initiatives, data from pharmacies on dispensed drugs may also become available over the coming years.

As of 2007, the GPRD will be using a trusted third party to enable record linkage to other NHS data sets. This linkage is planned for practice-level socio-economic class and complete death certificate information. It will be done at regular intervals and available to all researchers. Other linkages will only be available subsequent to ethics and scientific proto-col approval and only for that specific study as these linkages bring additional requirements for governance and data privacy. Detailed information on hospitalisa-tions (including main procedures and number of bed days) may also become available.

The following programmes have been developed to allow even better use of the GPRD:

·    Risk-management programmes: Pharmaceutical companies are now required to submit Risk-Management Plans to regulatory authorities for newly approved drugs, dose changes and new indications. Systematic data collection on a large cohort of drug users in routine clini-cal practice is an important element of risk management. The GPRD Group has developed the Risk Management Knowledge and Track-ing programme, which allow the monitoring of outcomes in drug users and importantly the key background information required for case assessment.

·    Surveillance programmes: Patients prescribed a drug can be followed for selected outcomes. Further information (including hospitalisation records) may then be requested. This information can then be used to assess the causality of the individual cases and also to estimate overall risks.

·    Randomised simplified trials: Subject to appro-priate approval (including the patient’s informed consent and approval by an ethics committee), it will be possible in selected practices to randomise patients to various treatments. Patients can then be followed using routine data collection to evaluate the outcomes. Confounding by indication is a major concern in pharmacoepidemiological research, and this randomisation can overcome bias due to baseline differences.

·    Prospective data collection: Subject to appropriate approval (including the patient’s informed consent and approval by an ethics committee), additional information can be obtained through the GP. This can include genetic samples. Pharmaco-genetic studies could be conducted to evaluate the effect of genetic polymorphisms on the response of drug treatment.

Current research governance guidance is to separate the scientific and ethical elements of protocol review, and the GPRD Group is currently working to implement new plans that will put its research governance arrangements on a robust footing with regard to current best practice. From March 2006, the Independent Scientific Advisory Committee (ISAC) for MHRA database research will be responsible for the scientific review of protocols for research using GPRD data. Members of this independent committee are appointed following a formal recruitment exer-cise run by the NHS Appointments Commission. The committee membership includes expert epidemi-ologists and statisticians as well as GPs and a lay member. Whilst the committee’s remit with regard to protocols is confined to the scientific aspects of the proposed research, it will have the ability to refer protocols for further ethical review by an NHS Research Ethics Committee (REC) where the proposed research is not covered by the existing ethics approval.