How reliable are stroke patients' reports of their numbers of general practice consultations over 12 months?

Data on primary health care use are frequently used in economic evaluations. However, it is unclear how patient self-reports of their number of consultations with their general practitioner (GP) relate to actual consultations in the electronic records. These data are crucial if self-reports are used to conduct economic evaluations.

To report the accuracy of stroke patients' self-reports of their number of primary care consultations over a 12-month period by comparison with practice-held electronic records. We also recorded the number of contacts required to collect service use data from the practices.

We contacted 65 practices requesting electronic consultation records over 12 months for 115 stroke patients who took part in a trial of home blood pressure monitoring. Consultation rates from the electronic records were compared with patients' self-reported number of consultations from a questionnaire covering the same period.

Fifty-one practices (78%) responded. Patients' questionnaires (n = 83) reported a mean of 5.7 consultations with their GP per year compared with 7.2 in the electronic records (difference 1.6, 95% confidence interval 0.5-2.7, P < 0.01). The mean time taken to obtain records from practices was 6 weeks.

Patients modestly under-reported the number of consultations they had with a GP. Obtaining patient records from practices required more effort than obtaining information from patient questionnaires at the same time as assessing main trial outcomes. If patient self-reports of health care usage are used in economic evaluations in primary care, researchers should consider validating a sample against electronic records.