Do Antidepressants Work? What Publication Bias Concealed

Abstract

Antidepressants are among the most prescribed drugs in the world, yet the evidence for their effectiveness has been seriously distorted by publication bias — the tendency for positive trials to be published and negative ones to disappear. When all registered trials are analyzed together, effect sizes are substantially smaller than published literature suggests, particularly for mild-to-moderate depression. The drugs do work for some patients, but the magnitude of benefit and the question of who benefits most remains poorly understood.

Background

Antidepressants — primarily selective serotonin reuptake inhibitors (SSRIs) like fluoxetine, sertraline, and citalopram — became dominant treatments for depression in the 1990s. By 2020, roughly one in eight adults in the United States reported taking antidepressant medication, making them among the most widely prescribed drug classes in the world. The case for their effectiveness rested on a large body of randomized controlled trial literature showing superiority over placebo.

The problem, identified in a landmark 2008 study, is that this body of literature was systematically incomplete.

The Evidence

Publication bias dramatically inflated perceived effectiveness. Irving Kirsch and colleagues submitted a Freedom of Information Act request to the FDA to obtain all clinical trials submitted for approval of six major antidepressants, including both published and unpublished studies. Their 2008 analysis in PLOS Medicine found that 94% of trials with positive results had been published, compared to only 14% of negative trials. When the full dataset was analyzed, the overall effect size dropped from 0.41 (a moderate effect) to 0.31 (below the 0.50 threshold the UK's NICE guidelines use to indicate a clinically meaningful difference). For patients with mild-to-moderate depression — the majority of people prescribed antidepressants — the drug-placebo difference was not clinically significant by these criteria.

A 2018 network meta-analysis found meaningful but modest effects. Cipriani and colleagues published a large meta-analysis in The Lancet covering 522 trials and 116,477 participants, examining 21 antidepressants. They found that all antidepressants were more effective than placebo, with odds ratios for response ranging from 1.37 to 2.13. This was presented as evidence that antidepressants work. Critics noted that the effect sizes, while statistically real, translated to modest real-world improvements, and that the placebo response rate (30–50%) was high enough to complicate interpretation.

Effect size increases substantially for severe depression. Both the Kirsch and Cipriani analyses agreed on one key finding: the drug-placebo gap is much larger for patients with severe depression than for those with mild or moderate symptoms. For patients with Hamilton Depression Rating Scale scores above 23 (indicating severe depression), antidepressants produce clinically meaningful benefits. The controversy is primarily about the much larger population of people with mild-to-moderate symptoms who are nonetheless prescribed these drugs.

Long-term evidence is weak. Most trials last 6–12 weeks — far shorter than the years or decades that many patients take antidepressants. Evidence on long-term outcomes, discontinuation effects, and whether the drugs prevent relapse as effectively as behavioral interventions is limited and mixed. The STAR*D trial, one of the largest real-world effectiveness studies, found that only about one-third of patients achieved remission on their first antidepressant, and that sequential treatment produced diminishing returns.

Placebo response is itself therapeutically significant. A consistent finding across psychiatric drug trials is that placebo responses are large and clinically meaningful. This does not mean placebos "don't work" — it means expectation, clinical attention, and the therapeutic relationship have genuine effects on depression outcomes. The appropriate comparison for antidepressant benefit may be active psychotherapy rather than inert placebo.

Counterarguments

The clinical significance threshold is contested. The 0.5 NICE threshold for the Hamilton scale has been criticized as arbitrary. Some researchers argue that even a 2–3 point difference on a 52-point scale represents meaningful symptom relief for individual patients, even if it appears small at the population level.

Antidepressants prevent recurrence. For patients with recurrent major depressive disorder, maintenance antidepressant therapy substantially reduces relapse rates. This benefit is separate from acute symptom reduction and is more consistently supported by evidence.

Withdrawal from antidepressants can be difficult. A relevant counterargument to the "modest effects" case is that many patients who want to discontinue antidepressants find it difficult — a practical problem that the publication bias debate has arguably underweighted.

What We Can Conclude

Publication bias concealed negative trials and inflated perceived antidepressant effectiveness for two decades. The best current evidence suggests antidepressants produce small but real effects on average, with substantially larger effects for severe depression and smaller or negligible effects for mild-to-moderate symptoms. For the majority of people prescribed antidepressants — those with mild-to-moderate depression — the evidence base is weaker than widely believed. Evidence-based psychotherapy (particularly cognitive behavioral therapy) has comparable effectiveness for mild-to-moderate depression with fewer side effects, and this option is underused.

References

• Kirsch, I., Deacon, B. J., Huedo-Medina, T. B., Scoboria, A., Moore, T. J., & Johnson, B. T. (2008). Initial severity and antidepressant benefits: A meta-analysis of data submitted to the Food and Drug Administration. PLOS Medicine, 5(2), e45.
• Cipriani, A., Furukawa, T. A., Salanti, G., et al. (2018). Comparative efficacy and acceptability of 21 antidepressant drugs for the acute treatment of adults with major depressive disorder. The Lancet, 391(10128), 1357–1366.
• Rush, A. J., Trivedi, M. H., Wisniewski, S. R., et al. (2006). Acute and longer-term outcomes in depressed outpatients requiring one or several treatment steps: A STAR*D report. American Journal of Psychiatry, 163(11), 1905–1917.
• Turner, E. H., Matthews, A. M., Linardatos, E., Tell, R. A., & Rosenthal, R. (2008). Selective publication of antidepressant trials and its influence on apparent efficacy. New England Journal of Medicine, 358(3), 252–260.
• Cuijpers, P., Karyotaki, E., Weitz, E., Andersson, G., Hollon, S. D., & van Straten, A. (2014). The effects of psychotherapies for major depression in adults on remission, recovery and improvement. Journal of Affective Disorders, 174, 400–410.