Lost subjects. Source of bias in clinical research?

This report describes a study of subject refusal as a source of bias limiting the generality of psychiatric research results. Fifty psychiatric patients who refused to participate in a battery of research interviews and psychological tests were compared with 50 participant patients. Between-group comparisons were made on an extensive and detailed set of clinical, treatment, and demographic data. Results revealed a remarkable lack of difference between the participant and refusal groups in demographic features, type or duration of hospitalization, and type or degree of pathology as defined by symptoms, prognostic measures, and diagnoses. These results and a review of related studies suggest that sample bias in psychiatric research resulting from loss of eligible subjects by refusal depends on the amount of institutional contact and the severity of disorder in the sample studied.