Context: It has been suggested that attenuated psychotic symptoms (APSs) reported by people who do not have psychotic disorders signal risk for later severe mental illness.
Objective: To investigate this suggestion using follow-up assessments of hospitalization for clinical diagnoses of nonaffective psychotic and other psychiatric disorders.
Design: Longitudinal cohort study of self-reported APSs with outcome assessment of severe mental illness obtained through linkage with a national hospitalization case registry.
Setting: Israel.
Participants: A stratified full probability sample of 4914 persons aged 25 to 34 years who were screened for psychopathology in the 1980s.
Main outcome measure: Subsequent psychiatric hospitalization was ascertained using the psychiatric hospitalization registry, with a mean follow-up of 24 years.
Results: After removing subjects with diagnosable psychotic disorders at baseline, 57.2% of the remaining sample reported at least 1 weak (infrequent) APS and 14.3% reported at least 1 strong (frequent) APS in the year preceding the assessment. Self-reported APSs predicted risk of later hospitalization for nonaffective psychotic disorders, mostly during the 5 years after baseline (adjusted odds ratio = 4.31; 95% CI, 2.21-8.41; positive predictive value = 1.27%; population attributable risk fraction = 33%). Also, APSs increased the risk of later hospitalization for other psychiatric disorders, albeit to a lesser extent (adjusted odds ratio = 2.21; 95% CI, 1.02-4.82).
Conclusions: Self-reported APSs signal risk for later nonaffective psychotic disorders but are not clinically useful as predictors. The difference between these population-based data and the high-risk literature in terms of the positive predictive value (1% vs 10%, respectively) and the time window of transition (5 years vs 12 months, respectively) can be attributed to the selective enrichment strategies that produce high-risk samples.