Eighteen RCTs with 1,837 patients (based on table data) were included. Three trials scored four on the Jadad scale, nine scored three, and six scored two.
Global cognitive index (18 RCTs): Statistically significant improvement in global cognitive performance was reported for atypical antipsychotics compared with typical antipsychotics (SMD 0.17, 95% CI 0.04 to 0.29). Trial quality, dose of typical antipsychotic (haloperidol dose equivalents), duration of treatment, pharmaceutical industry funding, and type of atypical antipsychotic did not seem to influence the global cognitive index outcome. There was no evidence of statistical heterogeneity between trials (QB(17) 24.11, p=0.12).
Cognitive domains: The atypical antipsychotics were found to be better than typical antipsychotics in the following aspects of cognition: language and verbal comprehension (SMD 0.38, 95% CI 0.15 to 0.62); psychomotricity (SMD 0.29, 95% CI 0.11 to 0.47), and speed of processing (SMD 0.26, 95% CI 0.13 to 0.39). There was no statistically significant difference for each of the other cognitive domains.
Rosenthal's file drawer test indicated that 38 trials were required to change the significant result achieved in the global cognitive index outcome. The correlation test showed that smaller studies were associated with larger effect sizes.