Prescription intentions in relation to therapeutic information: a study of 117 French general practitioners

In 1987 we conducted a mailed questionnaire survey involving 250 GPs, randomly drawn from the 3061 GPs in the 'Rhone-Alpes' region in France, in order to study how general practitioners (GPs) react to information about drugs in terms of their prescribing practices. The aim of the questionnaire was to invetigate the GPs reactions (prescription intentions) to 25 statements containing information concerning drugs. These included results from randomized clinical trials with adequate clinical criteria (pertinent information), but there were also some statements containing non-relevant information such as intermediate criteria, physiopathological or pharmacological information, and some containing general information such as advice from colleagues, the established position of the drug etc. The GPs were also asked through which channels they commonly received therapeutic information (i.e. medical journals, conferences). A total of 117 GPs returned completed questionnaires. We found the prescription intentions, for pertinent information to be between 76.9% and 95.7%, whilst the intentions, as a result of personal knowledge and/or success with a drug were around 93%. More theoretical information resulted in prescription intentions which were more widely scattered (between 23.1% and 80.3%), and for external advice the intentions were not as high but they were also widely scattered (between 3.4% and 65%). The search for latent dimensions corresponding to GPs reactions to therapeutic information, with both principal component analysis and Rasch Modelling, showed that two orthogonal latent dimensions, i.e. 'sensitivity to clinical and theoretical information', and 'sensitivity to external standards', best explained the responses to the questionnaire. These two diemsnions appeared to be independent of age, sex, medical school and type of practice (urban, rural). The use of the journal 'Prescrire' by GPs was found to be significantly associated (P < 0.005) with low scores, or good quality of perception of pertinent information in the first dimension. The use of specialists' prescriptions was associated with similar scores for the first dimension, but also with poor quality of perception of pertinent information scores (i.e. high scores) for the second dimension. These results could be used to draw up proposals for the improvement of post-graduate medical education, which should take into consideration these two dimensions of therapeutic information assessment by doctors, in order to obtain better quality of perception profiles for information assessment and prescription by doctors. Key words--prescription intentions, drugs, general practitioners, pertinent information, medical education, randomized controlled trials, questionnaires, Rasch analysis, principal component analysis