5 Questions: John Ioannidis discusses large meta-analysis of antidepressants

Some 350 million people globally are currently diagnosed with depression, making it the world’s leading psychiatric disorder. As many physicians and patients opt to use drug-based therapy, pharmaceutical companies have churned out a variety of antidepressants.

An international team of researchers specializing in fields such as psychiatry, public health and statistics conducted an in-depth analysis of more than 500 clinical trials that tested any of 21 different antidepressants. The group found that all 21 drugs were modestly more effective than a placebo — a validation for the overall efficacy of pharmacological treatment of depression.

The results of the meta-analysis, which were published online Feb. 21 in The Lancet, bring new evidence and insights to drug-based depression therapies, and could provide a valuable resource for doctors and patients sorting out antidepressant options.

John Ioannidis, MD, DSc, professor of medicine and co-director of the Meta-Research Innovation Center at Stanford, is an author of the study, which he said offers the best currently available, evidence-based guide for antidepressant treatments.

Recently, writer Hanae Armitage asked Ioannidis, an expert on improving the reliability and reproducibility of scientific research, to discuss the results of the analysis and what it means for navigating the pharmacological treatment of depression.

Q: What new light does this study shine on antidepressant treatment and antidepressants generally?

Ioannidis: Antidepressants have been one of the most widely used, misused and debated classes of drugs. Major depression is a huge problem, and having the best information available to make decisions on how to treat patients is totally essential. Relevant evidence has typically been highly fragmented and biased, which has not helped resolve the misuse of these medications or settle the debates about them. By putting together cleaned, standardized data from more than 116,000 patients and 522 randomized trials, and by using the highest possible standards of analysis, we can offer some concrete evidence to inform the proper use of 21 different antidepressants.

Q: What would you say are the biggest takeaways of this analysis from the perspective of a clinician? What about from the perspective of a patient with depression?

Ioannidis: Some key lessons are, first, that for acute depression in adults, antidepressants are effective, modestly so. Second, when a new drug is tested against an older one, there is sometimes bias favoring the newer drug — and the newest drugs should not necessarily be the top choice. Third, bias does not completely account for the efficacy of antidepressants. They do have a role in treating major depression. All 21 antidepressants that we assessed are better at treating depression than placebo, although the benefit is, on average, quite modest. Moreover, almost all of them are more acceptable, meaning that they are better tolerated, than placebo. (Acceptability is a measure that typically combines the impact of toxicity and the perceived efficacy.)

Finally, different agents have different profiles of efficacy and acceptability. Our study shows the relative merits and harms of each of the 21 antidepressants compared against one another. Some of them seem to be better, or more acceptable, although the differences between them are less pronounced when we compare them against placebo.

A patient can discuss this information with his or her physician and make a choice to start a specific antidepressant, if it is indicated, with some particular evidence-based expectations in mind. Of course, different patients may still respond differently, and this is difficult to predict ahead of time. But at least there is sensible and congruent information that informs patients on the average response that he or she can expect to have if taking one of these antidepressants.

Q: How can doctors use this information to better navigate treatment for their patients?

Ioannidis: Doctors may find it difficult or confusing to sift through the results of more than 500 trials in the literature, many of which may only be published in a fragmentary fashion or not at all. The network analysis offers a condensed version of the evidence, and they can quickly see where a specific drug is mapping itself in terms of efficacy, acceptability and other outcomes of interest. Hopefully, this will make decision-making easier and more appropriate.

Q: What other classes of drugs do you think could benefit from this type of analysis?

Ioannidis: For most types of drug treatments and most diseases, evidence is fragmented across a large number of mostly small and inconclusive trials, and there is substantial evidence that many of these trials are either left unpublished or have selective reporting of their outcomes. Large network meta-analyses that examine the entire body of evidence of all available drug treatments for various diseases, that make a systematic effort to unearth all the data and clean the results, would be very useful to perform across all medical specialties. There is an increasing number of papers that do perform network meta-analyses, but most of them look at snapshots of the evidence, consider some but not all the possible available treatments and make little or no effort to get unpublished results unearthed. 

Q: Do you think the results of this study carry implications for drug developers moving forward?

Ioannidis: Even though depression is very common, and thus the market for effective and well-tolerated drugs is very large, there has been a slowing of research and development of new drugs in this field. Our analysis clearly shows that there is a lot of room for improvement — both in efficacy and in acceptability compared with the currently available treatments. Thus, focusing on developing new drugs would be welcome, and our analysis can provide a yardstick of what the currently available drugs can achieve for comparison. It also shows that differences between available drugs are modest, therefore new research is needed to find drugs that use very different modes of action and may thus achieve better results.