A total of 60 RCTs (33,360 patients) and 27 observational studies (202,591 patients) were included. The mean duration of follow-up ranged from 12 weeks to two years. The quality of the RCTs was variable; 70% did not report an adequate generation of allocation sequence, 77% did not report adequate allocation concealment, and only 20% were double-blind. The quality of the observational studies ranged from 4 to 23/33 (median 13). The study sample size ranged from 20 to 18,154 patients in the RCTs and 50 to 63,214 patients in the observational studies.
Time to all-cause medication discontinuation (results taken from text)
In RCTs, olanzapine was significantly better than aripiprazole (HR 0.81, 95% CI 0.71 to 0.93; two RCTs), quetiapine (HR 0.68, 95% CI 0.56 to 0.83; six RCTs), risperidone (HR 0.77, 95% CI 0.70 to 0.86; 12 RCTs), ziprasidone (HR 0.73, 95% CI 0.59 to 0.90; six RCTs) and perphenazine (HR 0.68, 95% CI 0.48 to 0.97; three RCTs). There were no significant differences between olanzapine and clozapine (four RCTs), amisulpride (three RCTs) or haloperidol (two studies) in RCTs. Observational studies showed some similar results, but there were some differences; full results presented in the review.
All-cause medication discontinuation rates (results taken from text)
In RCTs, olanzapine was associated with significantly less discontinuation rates than aripiprazole (RR 0.87, 95% CI 0.80 to 0.93; six RCTs), quetiapine (RR 0.69, 95% CI 0.58 to 0.82; nine RCTs), risperidone (RR 0.86, 95% CI 0.81 to 0.92; 23 RCTs), ziprasidone (RR 0.81, 95% CI 0.78 to 0.83; six RCTs), haloperidol (RR 0.75, 95% CI 0.66 to 0.85; 16 RCTs), perphenazine (RR 0.78, 95% CI 0.64 to 0.95; three RCTs) and sulpiride (RR 0.56, 95% CI 0.32 to 0.96; one RCT). There were no significant differences between olanzapine and clozapine (eight RCTs) or amisulpride (four RCTs). Observational studies showed some similar results; full results were presented in the review.
There was evidence of significant statistical heterogeneity for several of the analyses. There was no evidence of publication bias apart from one analysis with observational studies. The dose of olanzapine did not alter results. Other analyses were reported.