Historical Overview

HISTORICAL OVERVIEW

The full history of the GPRD, since its creation in 1987, has been well documented elsewhere (Lawson, Sherman and Hollowell, 1998; Wood and Coulson, 2001). Since 2000, the database has been managed by the Medicines and Healthcare Products Regula-tory Agency (MHRA) under its remit to safeguard public health. It is used within the Agency to provide evidence for the evaluation of risks and benefits of marketed medicines. As such, the GPRD forms a critical part of the UK MHRA drug safety system.

Signals obtained through the spontaneous reporting scheme for suspected adverse drug reactions (known as the Yellow Card Scheme in the United Kingdom) may be tested in the GPRD. Such hypothesis testing is in GPRD conducted not only by the MHRA but by pharmaceutical companies, academics and other regulators.

Given the importance of the GPRD to public health, the MHRA has made an extensive investment in staff and information technology required to store and obtain access to the data. The MHRA developed the Full Feature GPRD that has now been available for over 5 years. It provides world-wide users with online access to a data warehouse. It was developed as not every researcher has access to the large data storage capabilities required to house the full data set or the experience of using powerful data manip-ulation and analysis tools as available in SAS or STATA. Researchers can access, through the Full Feature GPRD data warehouse, anonymised patient records alongside markers relating to the quality of the data, which are set during the data loading process. These markers include

•   the acceptable patient flag, which relates to the internal consistency of key patient data including age, gender and registration status, and

•   the practice up to standard date, which defines the first date at which the practice to which the patient is registered met the GPRD-derived mini-mum standards for data recording quality (Wood and Martinez, 2004).

In addition to the quality markers, a variety of other parameters, which increase the research utility of the data, are calculated during the data-loading process. Owing to the rapid increase in size of the GPRD data set and the volume of queries being run across the Web system, the GPRD has recently become available in a variety of other formats. This includes flat files for large subsets of the GPRD that can be loaded into statistical software packages.