Influence of new evidence on prescription patterns.

BACKGROUND It is currently accepted that no drug can enter clinical practice without proved efficacy in clinical trials. Improving patient care requires that the results of clinical evaluation be translated into practice. Results of studies are conflicting, but there is support that well-executed, clinically relevant randomized trials published in highly visible clinical journals can have an effect on patterns of medical practice.

METHODS We evaluated the potential impact of the publication in a leading journal of different drug studies (metformin, alendronate, terazosin, and finasteride) on the prescription behavior of generalists and specialists. Using a health maintenance organization (HMO) prescription drug database, we analyzed the incidence of new prescriptions written by generalists and specialists from a university-affiliated HMO before and after the publication date of the studies.

RESULTS The proportions of new prescriptions changed between a 6-month period before publication and a 6-month period after publication. The rate for alendronate increased from 31.7% to 43.2% of all prescriptions for specialists (P = NS) and from 8.8% to 38.9% for generalists (P < .01). The rate for metformin increased from 26.7% to 46.4% for specialists (P = .04) and from 7.9% to 24.2% for generalists (P < .01). The rate for alpha1-blockers decreased from 48.7% to 38.9% (P = NS) for specialists and increased from 20.7% to 60% for generalists (P < .01). The rate for finasteride decreased from 40.9% to 19.64% for specialists (P < .01) and from 22.11% to 11.3% for generalists (P = .01).

CONCLUSIONS The change in the prescription patterns of all physicians showed a clear temporal association with the publication of new evidence. The greater change observed for generalists could be explained by their lower baseline use of the drugs and a more conservative behavior that might defer the adoption of new treatments until they are supported by strong evidence published in major journals.