On diagnosing rare disorders rarely: appropriate use of screening instruments

Abstract

The main aim of this study was to determine whether child mental health professionals who regularly use questionnaires to screen for mental disorders know that these questionnaires have a low predictive value when the base rate of a disorder is low. The study was based on a representative sample of professionals who used questionnaires regularly to screen for mental disorders. They were set a problem in which a clinic sample was screened with a questionnaire that, at a certain cut-point, had 80 % sensitivity and 80 % specificity, and in which the true base rate of disorder was 10%. Only 10% (5/48) of respondents answered correctly that just 30 % of individuals who scored above this cut-point would actually have the disorder and more than half of respondents believed that 80 % would have the disorder. Both users and designers of questionnaires need to be more aware of and explicit about their drawbacks as screens for mental disorders.