• F Clinard, C Sgro, M Bardou, M Dumas, P Hillon, and C Bonithon-Kopp.
• Centre d'Epidémiologie de Population, Faculté de Médecine, 7, Boulevard Jeanne d'Arc, BP 87900, 21079 Dijon Cedex, France. fclinard@u-bourgogne.fr
• Pharmacoepidemiol Drug Saf. 2001 Jun 1;10(4):329-38.

PurposeIn order to assess biases occurring in primary care prescription studies, we compared non-steroidal anti-inflammatory drug (NSAID) prescribing patterns reported by general practitioners directly (GP-based survey) and from a pharmacy-based survey of general practitioner prescribing (pharmacy-based survey).MethodsVolunteer GPs of the administrative area of Côte d'Or (France) returned a mailed questionnaire on NSAID prescribing patterns for consecutive patients seen during a 2-month period. In order to obtain a reference basis, pharmacies of the same administrative area provided all prescriptions that included NSAIDs during a 1-week period originating in general practice.ResultsThe rate of participation was 25% for the GPs and 40% for the pharmacies. Participant GPs were representative of GPs of the area with regard to sex, year of graduation and practice area but pharmacies from rural areas were over-represented. The GP-based survey and the pharmacy-based survey provided respectively 770 and 1050 prescriptions. There were no differences between either survey in the type of NSAIDs prescribed and in the most frequently associated drugs. GPs who volunteered in the GP survey prescribed NSAIDs more frequently orally and at higher doses than GPs involved in the pharmacy-based survey. They also prescribed more gastroprotective drugs, especially in the elderly. None of these results could be explained by differences in patient characteristics and GP practice areas.ConclusionGPs who actively participate in prescription surveys exhibit prescribing patterns that fit better with official recommendations than the average. Although selection biases cannot be ruled out, it is suggested that some changes in GP prescription habits may have been induced by the survey itself.

29 keywords...

hide…