Are new drugs for schizophrenia better than old ones?

摘要

Stefan Leucht and colleagues (Jan 3, p 31) did an impressive meta-analysis of 150 head-to-head trials of antipsychotic drugs for schizophrenia. They note that so-called second-generation drugs do not deserve this label, but they report in their Summary that four of these drugs were better than first-generation antipsychotics. I believe it is impossible to know whether or not this is true. First, the comparator was haloperidol, which is known to be a poor drug, in 95% of the trials. Second, selective publication of positive results is common.The paper Leucht and colleagues quote showed that the effect of antidepressant drugs was more than twice as large in published placebo-controlled trials than in similar trials that were only available on the US Food and Drug Administration website. A similar study found a preponderance of per-protocol analyses in published reports, rather than the more conservative intention-to-treat analyses. For head-to-head comparisons, selective reporting might be an even bigger problem than for placebo-controlled trials, since positive findings-although spurious-can be worth billions of dollars in sales.