Where Does Evidence from New Trials for Schizophrenia Fit with the Existing Evidence: A Case of the Emperor's New Clothes?

Abstract

Advent of "atypical" antipsychotics has spawned new trials in the recent years and the number of such trial reports has been increasing exponentially. As clinicians we have been led to believe that "atypicals" are better than "typicals" despite the odd dissenting voice in academic and clinical circles. This has been largely ignored until the publication of two landmark, independent, pragmatic trials, Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) and Cost Utility of the Latest Antipsychotic Drugs in Schizophrenia Study (CUtLASS), which proved that thoughtfully chosen "typical" antipsychotics were as good as the newer "atypicals." We pooled "leaving the study early data" from Cochrane Reviews that existed before CATIE and CUtLASS and added data from CATIE and CUtLASS to the pool for a "before and after" comparison. Addition of CATIE and CUtLASS data only led to narrowing of the already existing confidence intervals, merely increasing precision, and decreasing the risk of Type II error. Perhaps surprisingly, CATIE and CUtLASS when pooled with the already existing data showed us that we had chosen to turn a blind eye to findings that already existed. This leads clinicians to question as to whether, in future, we need to feel less guilty about crying out early on that the emperor has no clothes on.

Figure 1

Forrest plot showing comparison of cumulative risk ratios for the outcome of “leaving the study early” before and after CATIE and CUtLASS.