Purpose: Atypical neuroleptics, especially clozapine, are clinically more efficacious than the typical neuroleptic haloperidol. We examined the possibility of a further ranking in efficacy of the atypical drugs by using a previously described Bayesian statistical method that included both direct and indirect comparisons simultaneously.
Method: We used the results of 30 double blind, randomized studies including 37 comparisons of four atypical neuroleptics (aripiprazole, clozapine, olanzapine and risperidone) and haloperidol, head-to-head or against placebo. These studies reported the rate of responders, i.e. the number of patients with a certain degree of improvement divided by the total number of patients in the group. We calculated the success ratios for drugs against placebo and thereafter the relative success ratios for one drug against another, altogether 10 comparisons. With uniform priors, we calculated and ranked posterior estimates of success ratios with the use of the BUGS statistical program and 10 000 iterations.
Result: We ranked the medications according to their success ratios. We found that aripiprazole was the least effective among the new antipsychotics with a success ratio of 1.44 (posterior credibility interval (CI) (1.24-1.66)), compared to clozapine with a success ratio of 1.75 (CI (1.51-1.98)), olanzapine with a success ratio of 1.67 (CI (1.48-1.88)) and risperidone with a success ratio of 1.65 (CI (1.47-1.85)). For comparison we also analyzed haloperidol which was ranked as number five with a success ratio of 1.26 (CI (1.10-1.44)). The posterior probabilities that clozapine was better than respectively olanzapine, risperidone, aripiprazole and haloperidol were 0.78, 0.81, 0.98 and 1.The posterior probabilities that olanzapine was better than respectively aripiprazole and haloperidol were 0.97 and 1, and that risperidone was better than respectively aripiprazole and haloperidol were 0.96 and 1. The posterior probability that aripiprazole was better than haloperidol was 0.95. Finally, the posterior probability that clozapine, olanzapine and risperidone were the best three was 0.94.
Conclusion: According to our metaanalysis clozapine, olanzapine and risperidone are significantly better than aripiprazole and haloperidol. Clozapine is clearly ranked before olanzapine and risperidone (although not being significantly better). In addition, aripiprazole is significantly better than haloperidol. Among the group of atypical neuroleptics aripiprazole is the least effective treatment, which corresponds with its effect being a partial dopaminergic agonist as well as an antagonist.
Candidate for the Lee B. Lusted Student Prize Competition