Early prediction of clinical and functional outcome in schizophrenia

The objective of this paper was to investigate the prognostic and predictive value of a small panel of independent and clinically important factors based on symptom improvement, baseline cognitive impairment, and weight change during the early treatment phase.

The study sample was based on a double-blind, 6-month continuation study of ziprasidone and olanzapine (N=94). We developed a parsimonious 6-month GAF prediction function using a logistic regression model, and evaluated its predictive accuracy and performance using bootstrap estimates of c-statistics and error in predicted probability.

At up to 6 months of follow-up, 52 (55%) of all subjects treated with ziprasidone or olanzapine met the responder criterion of ≥50% improvement in GAF. At Week 2 (acute phase), the majority of ziprasidone (75%) and olanzapine (70%) patients showed greater than 25% improvement in the BPRS psychotic symptom subscale score. These early psychotic symptom responders (Week 2) showed significantly greater improvement in global functioning than early non-responders at all time points (Week 6 and Month 6) (all p's<0.05), confirming early response as an indicator of continued responsiveness to treatment over at least 6 months. A multivariate prediction function based on baseline neurocognitive scores and GAF, early reduction of psychotic symptoms at 2 weeks, and percentage of weight change observed at 6 weeks (All p's <0.05), showed statistically acceptable predictive performance (boostrap c-statistics=0.8598).

Our findings suggest that a parsimonious model incorporating a psychotic symptom assessment score, baseline neurocognitive performance, and risk of weight gain can be developed for predicting patients' likelihood of achieving favorable, long-term treatment outcomes.