Twenty studies were included (outcome data for 5,313 participants, n=1,634 placebo, n=553 risperidone, n=642 olanzapine, n=605 quetiapine, n=1,028 aripiprazole and n=851 paliperidone extended-release). Sample sizes in the individual studies ranged between 69 and 630. Six studies examined risperidone, six examined olanzapine, four examined quetiapine, six examined aripiprazole and three examined paliperidone.
All of the atypical antipsychotic drugs significantly improved total PANSS scores (overall effect size -11.6, 95% CI -13.3 to -10.0). Effect sizes for the individual agents ranged from -14.9 (95% CI -17.6 to -12.3) for olanzapine to -9.5 (95% CI -11.7 to -7.2) for aripiprazole. PANSS positive score was significantly improved (overall ES -3.7, 95% CI -4.2 to -3.1). Effect sizes for individual agents ranged from -4.3 for risperidone and olanzapine (risperidone 95% CI -5.7 to -2.8 and olanzapine 95% CI -5.3 to -3.4) to -2.6 (95% CI: -3.4, -1.7) for aripiprazole. There was significant improvement in PANSS negative score (overall effect size -2.4, 95% CI -2.9 to -2.0). Effect sizes for individual agents ranged from -3.4 (95% CI -4.2 to -2.7) for olanzapine to -1.3 (95% CI -2.6 to -0.07) for quetiapine. One study on olanzapine included patients with a lower mean PANSS total score than in other studies and its exclusion reduced the effect size for olanzapine.
Improvement on CGI-S score with atypical antipsychotic agents was -0.5 overall (95% CI -0.6 to -0.4). Effect sizes for individual agents ranged from -0.8 (95% CI -1.1 to -0.5) for risperidone to -0.3 (95% CI -0.4 to -0.2) for aripiprazole.
Paliperidone extended-release, olanzapine and risperidone tended to have lower rates of withdrawals due to lack of efficacy than all atypical antipsychotic agents together (compared to placebo, OR overall 0.39, 95% CI 0.34 to 0.45). Rates tended to be higher than the mean for aripiprazole and quetiapine.
Overall withdrawal for any reason had an odds ratio of 0.52 (95% CI 0.46 to 0.58) compared to placebo. Rates were clearly lower than the mean for paliperidone extended-release and risperidone and higher than the mean for quetiapine and aripiprazole.
There was no significant difference in rates of withdrawals due to adverse events for all the atypical antipsychotic agents taken together compared to placebo. Results were similar for the individual agents except olanzapine, which had higher withdrawal rates due to adverse effects; the extent to which this result was significant was unclear.
Atypical antipsychotic drugs led to significant weight gain compared to placebo (OR 2.84, 95% CI 2.3 to 3.5). Odds of weight gain were lowest with paliperidone extended-release (OR 1.75, 95% CI 1.29 to 2.37) and highest with olanzapine (OR 4.56, 95% CI 3.46 to 6.01). There was no obvious influence on this outcome of any of the assessed confounders (age, sex and duration of therapy).
Atypical antipsychotic drugs were associated with increased odds of somnolence compared to placebo (OR 1.7, 95% CI 1.39 to 2.09). Odds of somnolence were lower than the mean with paliperidone extended-release and aripiprazole and higher than the mean with risperidone and olanzapine.
Overall, there was no significant difference in agitation between atypical antipsychotic drugs and placebo. Agitation was claimed to be lower than placebo for paliperidone extended-release and for quetiapine, but the significance of the result was uncertain.
Results for all safety outcomes were consistent when lower effective dose ranges were included.