In the discussion section of their review, Leucht and colleagues commented on several controversial issues, including outcomes measures, duration of studies in a meta-analysis, and decrease of drug efficacy over the decades.
Psychiatry is often criticized, they wrote, for using “soft outcomes,” such as rating scales, whereas medicine uses “hard” outcomes, such as death or major events (eg, heart attack). Still, they wrote, there are examples in general medicine (eg, asthma, diabetes) for which intermediate outcomes may improve but mortal-ity increases, as well as other examples (esophagitis and migraine) for which the reductions of symptoms and suffering are regarded as primary outcomes.
“Therefore, reduction of disease severity (eg, degree of delusions and hallucinations in schizophrenia) and prevention of further episodes are primary outcomes, and it is not entirely appropriate to criticize psychiatry for using ‘soft’ outcomes. This said, there is considerable room for improvement in psychiatric outcome measures, and death or suicide should always be reported. The example of lithium(Drug information on lithium) shows that some psychiatric drugs may reduce suicide rates.”
Some of the most important outcomes take years to develop, and you can’t measure them with double-blind studies that are often only 6 to 8 weeks long, Davis added. “We have to look at other methodologies.”
Regarding the duration of studies, Leucht and associates noted that studies of many years’ duration would be necessary to obtain large differences in mortality, “but such studies are almost impossible to conduct for many reasons,” so shorter studies are performed, which show only small differences.
“In this context, many psychiatric drugs not only improve the acute episode but also prevent further episodes. Patients with severe, recurrent depression might have 20 episodes in their lifetime, which could be reduced by medication to 10,” they wrote.
The authors also acknowledged that earlier meta-analyses in psychiatry yielded higher effect sizes than recent meta-analyses. In a paper published last year, Davis and coworkers6 wrote that the antidepressant drug-placebo difference is larger in the more severely depressed subgroups and in older studies. They explained that in the early double-blind studies involving antidepressants, for example, there were severely ill and drug-naive patients referred to clinical trials by their physicians.
Davis said that many severely ill and suicidal patients are excluded from recent drug trials because of ethical concerns, that a lack of “fresh” (drug-naive) patients exists, and that there is an increase in advertisements offering free medications to clinical trial participants—all of which can influence the placebo response.
Also, pharmaceutical companies have, on occasion, suppressed data on negative trials, Davis said.
“The complaints against the drug companies hiding studies and heavily promoting drugs are often quite legitimate, but it doesn’t mean the drugs are worthless,” he said.
Results of all controlled studies—including failed studies—should be published, Davis believes. He points to some pharmaceutical companies that are sharing information on both published and unpublished studies. One example, he said, is a recent article of which he is a coauthor.7 The article is a reanalysis of the randomized placebo-controlled studies of fluoxetine(Drug information on fluoxetine) and venlafaxine that used complete longitudinal person-level data from a large set of published and unpublished studies. The reanalysis found that the drugs decreased suicidal thoughts and behavior for adult and geriatric patients and that the “protective effect was mediated by decreases in depressive symptoms with treatment.”
Davis stressed the need for everyone—physicians and patients alike—to examine the data on psychiatric drugs and efficacy and to understand the problems.
“It’s much harder,” he said, “to think through the issues than to come to snap judgments.”