Carroll et al have produced an immensely valuable piece of work in their recent study of the DST (Archives 1981;38:15-22). However, their recommendation that it be used widely by internists and family practitioners deserves a word of warning.
Galen and Gambino have emphasized that the probability of a positive laboratory test indicating the presence of a disease is heavily dependent on the prevalence of the disease in the population tested.1 The statement by Carroll et al that a positive DST at the 5-μg level indicates 94% probability of the presence of melancholia is true only if the population tested contains 58% melancholiacs, as theirs did. If a generalist began to give the test to everyone with depressive symptoms, it would not seem unreasonable if there were only 10% true melancholiacs in the population he was testing. At this prevalence, the predictive value of a positive test would be