The statistical comparison of clinical trials.

Being able to compare clinical trials using statistically proven analyses is essential, but certain protocols must be followed if such a comparison is to be made. The difficulty of making meaningful statistical comparisons is illustrated in 5 clinical trials comparing atypical antipsychotics. Certain judgments can be made on the basis of internal and external validity, for example, but there are many other areas--effect size, for example--in which it is impossible to make any statistical comparisons across these trials because they were not conducted in a uniform fashion. In the final analysis, it is doubtful that the differences between atypical antipsychotics covered in these trials are greater than those due to chance.