Early prediction of antipsychotic nonresponse among patients with schizophrenia.

OBJECTIVE: Schizophrenia guidelines recommend waiting several weeks before major changes in antipsychotic treatment are implemented. If, however, nonresponders could be identified shortly after treatment initiation, considerable time could be saved by rapidly switching such patients to potentially more effective alternatives. We therefore attempted to identify what degree of nonresponse shortly after initiation of antipsychotic drug treatment predicts nonresponse at 4 weeks.
METHOD: Individual patient data from 7 randomized, controlled antipsychotic drug trials including 1708 patients with schizophrenia or schizophreniform disorder according to DSM-III-R or DSM-IV criteria and positive symptoms (mean +/-SD age of 36.0 +/-10.9 years; 1054 men and 654 women) were pooled. Receiver-operator curves and logistic regression analyses were used to predict nonresponse at week 4 from the percentage Brief Psychiatric Rating Scale (BPRS) score change at weeks 1 and 2. Three criteria for nonresponse at week 4 were examined: less than 25% BPRS score reduction, less than 50% BPRS score reduction, and "no remission."
RESULTS: Cutoffs predicting nonresponse at 4 weeks with 90% specificity were virtually no response at week 1 (less than 3%-7% BPRS score reduction) and less than 15%, 25%, and 20% at week 2 for the 3 nonresponse criteria described above, respectively. However, to predict less than 25% BPRS score reduction with a positive predictive value of 80%, the cutoff needed was 0% BPRS score reduction at week 2. When the cutoffs identified were entered in logistic regression analyses together with other parameters, they remained the strongest predictors of nonresponse.
CONCLUSIONS: Patients with no improvement of symptoms during the first 2 weeks of treatment are unlikely to respond at week 4 and may benefit from a change of treatment.