General Practice Research Database provides detailed anonymised dataBMJ 1996; 313 doi: https://doi.org/10.1136/bmj.313.7059.757 (Published 21 September 1996) Cite this as: BMJ 1996;313:757
EDITOR,—Congratulations are due to Nicky Pearson and colleagues on their creation of a database of local morbidity data by aggregating the computerised medical records of general practitioners in Somerset.1 I wish to point out, however, that a large database already exists—namely, the General Practice Research Database, which is owned by the Department of Health; the Office of National Statistics (formerly the Office of Population Censuses and Surveys) is its custodian and operator.
The research database comprises anonymised medical records from general practice on over 3.5 million patients; the records cover about 6.5% of the population of England and Wales and come from around 550 practices.2 The earliest continuous records date from 1987 and most extend back to 1991, giving over 15 million patient years of observation, with considerable longitudinal value. The data are patient based and include detailed information on prescribing and medical history, including information received by the practices from hospitals. As with the Somerset scheme, the data have been recorded as part of general practice rather than as a special exercise and are subject to a range of quality checks before being loaded into the research database. Several studies have shown the data to be of good quality.3 4 5 The database is updated annually.
Data can be provided down to the smallest area that does not permit identification of the individual practice. Thus many health authorities are able to obtain data for commissioning purposes or for research. Because the database holds both therapeutic and diagnostic data at the level of individual patients, more detailed analysis is possible than with the Somerset scheme. Additionally, the General Practice Research Database can provide regional and national baselines for comparative purposes. Use of the database avoids the delay, trouble, and expense incurred in imitating the Somerset morbidity project. The difficult task of finding general practitioners willing to undertake additional work is rendered unnecessary.