Thirty-one trials with 2,320 patients were included in the review.
Fifteen trials compared clozapine with chlorpromazine and 11 provided data on the number of patients with at least one EPS. The pooled estimate showed that fewer patients in the clozapine group experienced EPS (NNT 7, 95% CI: 4, 25, P=0.008). The pooled estimate also showed that more patients in the clozapine group achieved a clinically significant improvement (NNT 7, 95% CI: 4, 33, P=0.02).
Four trials compared olanzapine with chlorpromazine. Fewer patients in the olanzapine group experienced at least one EPS, but this was not statistically significant (P=0.07). More patients in the olanzipine group showed a clinically significant improvement (NNT 5, 95% CI: 2, 50, P=0.03).
One trial compared quetiapine with chlorpromazine. There was no significant difference in the number of patients experiencing EPS, but more patients in the quetiapine group experienced a clinically significant improvement (P=0.05).
Remoxipride was compared with chlorpromazine and thioridazine in 4 trials. There were no significant differences between the groups.
One trial compared risperidone with methotrimeprazine. Patients in the risperidone group showed a greater clinical improvement (NNT 3, 95% CI: 2, 100, P=0.04).
Zotepine was compared with chlorpromazine and perazine in 5 trials. There were no significant differences between the groups.
Dose of low-potency antipsychotic had a significant effect: the pooled risk difference (RD) for studies using doses lower than 600 mg/day chlorpromazine was not statistically significant (RD=0.01, 95% CI: -0.03, 0.04, P=0.7), while that for studies with higher doses was statistically significant (RD -0.26, 95% CI: -0.37, -0.16; NNT 4, 95% CI: 3, 6, P<0.0001).
The meta-regression analysis showed no effect of study quality, washout phase, or year of publication on the outcomes.
The funnel plot analysis suggested that relevant unpublished studies that showed no benefit of new-generation drugs might not have been included in the review.