Determining patient characteristics for decision analysis support systems using anonymized electronic patient records.

While decision analysis has the potential to improve treatment decisions, the quality of patient specific data has hindered its development. Experience with the General Practice Research Database (GPRD) indicated the possibility of generating such data. The example developed concerns the probability of recurrent stroke and the risks and benefits of treatment with aspirin. Eight criteria were systematically applied to identify suitable patients from the 664,241 cohort of patients in the West Midlands GPRD. The risk of recurrent stroke was estimated using survival analysis, yielding multivariate hazard ratios for 960 patient profiles. These were used to populate a decision analysis model. The study highlights how decision analysis can introduce evidence-based information into the consultation. Data can be transformed into information for the increasingly complex world of general practice. This can clarify the risks and benefits of prescribing, assist GP--patient dialogue and promote shared decision-making. [ABSTRACT FROM AUTHOR]