Influence of selection of patients on results of clinical trials

The hypothesis is advanced that any population of patients can be divided into those with a good prognosis, a poor prognosis, and an intermediate prognosis in respect to a particular treatment. The outcome of a clinical trial of treatment will be partly determined by the proportion of patients selected for the trial from each of these prognostic groups. The implications of the hypothesis are illustrated by consideration of three recent trials of treatment for psychiatric patients.