background Self-reports of psychotic symptoms are considered to be unreliable; never-theless, in longitudinal cohort studies these self-reports turn out to have high prognostic validity. This apparent paradox is subjected to scrutiny in a prospective study.
method 4081 persons without any previous history of psychotic disorders were interviewed by means of the Composite International Diagnostic Interview (CIDI) at baseline and 3 years later (T2). 'An unrecognised symptom' was defined as a psychotic symptom self-reported at baseline but rejected by the clinician in a re-evaluation procedure at baseline.
results An 'unrecognised symptom' turned out to be a relatively powerful predictor for the development of a new psychotic disorder at T2 (odds ratio 27.5; 95% confidence interval 4.5 - 123.4). The positive predictive value of an unrecognised symptom was 0.05; (95% confidence interval 0.040 - 0.053). However, in the group of unrecognised symptoms the positieve predictive value was only higher in respondents who had also had sub-clinical psychotic experiences (0.10; 95% confidence interval 0.01-0.19).
conclusions On its own, clinical evaluation of psychotic symptoms is no guarantee of the predictive value of psychotic disorders. A more accurate prediction can be made by continuing self-reported sub-clinical experiences with clinically assessed psychotic experiences.