Abstract
OBJECTIVE: Early intervention and prevention of psychosis remain a major challenge. Prediction would be greatly advanced with improved ability to identify individuals at true risk, which, at present, is moderate at best. The authors tested a modified strategy to improve prediction by selecting a more homogeneous high-risk sample (attenuated positive symptom criteria only, age range of mid-teens to early 20s) than is currently standard, combined with a systematic selection of neurodevelopmental deficits. METHOD: A sample of 101 treatment-seeking adolescents (mean age, 15.9 years) at clinical high risk for psychosis were followed clinically for up to 5 years (mean follow-up time, 3.0 years, SD=1.6). Adolescents were included only if they exhibited one or more attenuated positive symptoms at moderate to severe, but not psychotic, severity levels. Cox regression was used to derive a risk index. RESULTS: The overall conversion rate to psychosis was 28.3%. The final predictor model, with a positive predictive validity of 81.8%, consisted of four variables: disorganized communication, suspiciousness, verbal memory deficits, and decline in social functioning during follow-up. Significant effects also suggest narrowing the risk age range to 15-22 years. CONCLUSIONS: Clinical high risk criteria that emphasize disorganized communication and suspiciousness while also including compromised verbal memory and declining social functioning have the potential to improve predictive accuracy compared with attenuated positive symptoms used alone. On the resulting risk index (a weighted combination of the predictors), low scores were interpreted as signifying minimal risk, with little treatment necessary, high scores as suggesting aggressive intervention, and intermediate scores, although less informative, as supporting psychosocial treatment.