Survey method for post-marketing drug surveillance: a demonstration.

Distinguishing characteristics of a community survey method for post-marketing drug monitoring are described, as are results of a pilot study of the method. Representative samples of outpatients treated with tricyclic antidepressants or benzodiazepine anxiolytics were identified in a clinic or pharmacy and interviewed at baseline and specified intervals thereafter in the home or by telephone. Data were collected by lay interviewers trained to use a detailed interview schedule designed with input from experts in pharmacology, relevant clinical disciplines, and survey research. The critical validity question was the degree to which predicted outcomes for the two well-studied drug classes matched observed outcomes. The analytic design involved two contrast groups, three measurement periods, and six key symptom measures. The latter were specific to the disorder (anxiety and depression), to the medications, or to neither the disorders nor the medication (neutral symptoms). Predictions took account of degree and direction of change, as well as differences in profile levels over time. Results conformed precisely to expectations. As one component of a comprehensive system, the method provides an opportunity to examine efficacy as well as safety under conditions that are typically absent or excluded in clinical trials. Data on patterns of prescribing by physicians and use by patients are valuable by-products that are immediately relevant for professional education and product liability. Advantages and limitations of the method are discussed.