The number needed to treat for all-cause medication discontinuation in the treatment of schizophrenia: consistency across world geographies and study designs.

The number needed to treat (NNT) for all-cause medication discontinuation in large, industry-sponsored, non-randomized, observational studies conducted across world geographies was compared with NNTs from CATIE, an 18-month, NIMH-sponsored, randomized study. NNTs (with 95% confidence intervals) were calculated using data from 3 large Lilly-sponsored, non-randomized, observational studies (EU-SOHO, IC-SOHO, and US-SCAP, n=20 957). Group differences at medication initiation were adjusted by Cox regression modeling. These NNTs were compared with published NNTs for CATIE (phase 1). NNTs for olanzapine vs. risperidone and for olanzapine vs. quetiapine were similar across the observational studies and similar to those of CATIE. The NNTs for olanzapine vs. oral typical antipsychotics were similar across the observational studies but demonstrated a somewhat stronger effect size than the NNT reported for olanzapine vs. perphenazine in CATIE. NNTs for all-cause treatment discontinuation (a proxy measure of a medication's effectiveness from patients' and clinicians' perspectives) appear to be consistent across study designs (non-interventional, observational vs. RCT), study sponsorship (industry vs. independent), and across world geographies, suggesting that antipsychotics differ in this measure.