Meta-analyses overview

The meta-analyses review, Davis said, is particularly important for primary care and other physicians who may “think that psychiatric drugs are not efficacious, may not prescribe them, and may discourage their patients from taking them. Such perceptions and actions,” he noted, “can cause great harm to patients.”

Davis added he has tried for years in his lectures to make psychiatrists aware that the effect sizes of the psychiatric drugs are in “the ballpark with most of the internal medicine drugs” and that “most medical drugs were not the breakthroughs they [psychiatrists] thought they were.”

“With this review,” he told Psychiatric Times, “we finally got it done.”

In a commentary published in BMC Medicine, Seemuller and colleagues³ from the Department of Psychiatry and Psychotherapy at the Ludwig-Maximilian University of Munich described the review as “a milestone in destigmatizing psychiatry and its pharmacological treatments.” They described Leucht as “an experienced member of the Cochrane collaboration” who is very familiar with the pitfalls of meta-analyses.

Similarly, Davis is highly experienced with meta-analyses. “I wrote the first ones in psychiatry in 1975 and 1976, even before they were called meta-analyses,” he said.^4,5

For their article, Leucht and colleagues searched Medline and the Cochrane Library for systematic reviews on the efficacy of drugs compared with placebo and then systematically presented the effect sizes for primary efficacy outcomes. They included 94 meta-analyses of 48 drugs in 20 medical diseases (eg, cardiovascular disease, hypertension, rheumatoid arthritis, chronic asthma, type 2 diabetes mellitus, and hepatitis C) and 33 meta-analyses of 16 drugs in 8 psychiatric disorders (eg, schizophrenia, bipolar disorder, MDD, obsessive-compulsive disorder, ADHD, and Alzheimer disease). They excluded meta-analyses of subgroup studies and, if available, chose reviews of classes of drugs rather than single drugs.

“To be up to date, we also chose more recent studies,” he said. “And when there were several meta-analyses on the same topic, we looked to see if they agreed or not; if they disagreed, we called the authors to find out why. So there is extensive information in the fine print.”

While the review paper “covers all our important findings,” Davis explained, the team made available some 55 pages of Tables and Figures online at bjp.rcpsych.org “for individuals interested in all the data that lie behind the analysis.”

According to the research team, an effect size of 0.2 is considered significant but low, while an effect size of 0.8 or above is considered high. The median of all effect sizes was 0.40.

There was a lot of variability in effect size for medical conditions, Davis said. For example, there was a high effect size (1.39) for proton pump inhibitors to treat reflux esophagitis and a high effect size (2.27) for interferon to treat chronic hepatitis C. But many commonly used general medicine drugs, such as statins and aspirin in cardiovascular disease and stroke, had small effect sizes (0.12 for aspirin for secondary prevention of cardiovascular events and 0.15 for statins for cardiovascular events).

“As a generalization, the effect sizes of psychiatric drugs are right in the middle of most of the drugs used in internal medicine,” Davis said.

Antidepressants used as “maintenance treatment” to prevent a relapse of MDD had an effect size of 0.64; antipsychotics used to prevent relapse in schizophrenia had an effect size of 0.92. Less pronounced was the effect size of 0.26 for cholinesterase inhibitors for dementia. In between were atypical antipsychotics and haloperidol, with an effect size of 0.44 for acute mania in bipolar disorder.