Researchers have provided critical input to help the MCA develop the Full Feature GPRD. Some indicated that, in the past, they had been precluded from using the GPRD or using it as extensively as they might. The cost of access, requirement for extensive software and hardware to store the data, and the need for sophisticated programs to extract and analyse the data were cited as rate limiting factors in its use, and this has informed the development approach adopted for the new database and services offered from it. The approach has focused on:
• Development of a robust data model. GP computer systems will continue to evolve to support GPs in managing their practices more efficiently, while the nature of the diagnostic and treatment data will remain relatively static. Hence, the data model that has been developed is based on the elements of a patient record rather than on the data structures required by the GP computer software. This will facilitate future collection and storage of additional data from practices using both InPS software and from other potential data sources.
• Availability to researchers of the raw data collected from GPs. Researchers will continue to have access to the anonymised data as they were recorded in the GPs' practices.
• Provision of ''value added'' information to researchers, e.g. on data quality, which is clearly distinguished from the raw data.
• Access to data subsets (data marts), a frozen dataset and to a continuously updated database.
• Provision of hierarchical dictionaries to support data retrieval. Historically, InPS practices used OXMIS medical terms. Currently they use Read Clinical Terms as mandated by the UK Department of Health. It is likely that SNOMED Clinical Terms will be adopted as the UK National Health Service standard in the future. Similarly, various drugs dictionaries are used by GPs in the United Kingdom. Researchers using GPRD will be able to analyse the data using Read Clinical Terms or the Medical Dictionary for Regulatory Activities (MedDRA ) terminology and the ATC or GPRD Drugs/ Product dictionary to support comprehensive data retrieval. For example, a search on fluoxetine will retrieve all information on the drug substance, drug substance variants, formulations and licensed products via one query.
• Reducing the opportunity costs of using GPRD. The total size of the new database is in the range of one terabyte. To date, GPRD data have been distributed as flat files to users. The MCA is offering on-line access to the Full Feature GPRD which:
— reduces the cost of accessing the database since researchers will no longer require extensive hardware and software on site;
— enables a wider range of services to be provided by the MCA in keeping with the MCA's vision for increasing GPRD use; and
— ensures that all users have access to the same base data.
The technical architecture for the Full Feature GPRD is summarised in Figure 29.1. All GPRD data are stored in an Informix database called the Operational Data Store, an area which also includes medical and drugs terminology information and operational data to support the GPRD Data Collection and Processing Management
Was this article helpful?