There has been debate about the frequency and severity of antidepressant withdrawal effects.

We set out to appraise and reanalyze an influential systematic review by Henssler and colleagues that concluded that withdrawal effects are not particularly common and rarely severe. We repeated the meta-analysis, including only studies where data were derived from systematic measures of withdrawal symptoms.

Most data in the Henssler review are derived from pharmaceutical industry–sponsored efficacy studies in which withdrawal was a minor consideration. Shortcomings of the review include the use of spontaneously reported adverse events to estimate withdrawal symptoms, potential misclassification of withdrawal symptoms as relapse, inclusion of data from retrospective case-note studies, short duration of prior antidepressant use, short observation periods, the overlooking of differences between placebo and drug withdrawal effects, and the use of questionable proxies for severe withdrawal. There were also discrepancies and uncertainties in some figures used. In our reanalysis, we included only the five studies that used a systematic and relevant method to assess the incidence of any withdrawal symptom. Prior treatment was short-term (12 weeks or less) in all but one of these. The pooled percentage was 55% (95% confidence interval, CI, 31% to 81%; N = 601) without subtracting nocebo effects, with high heterogeneity.

Henssler’s review is based on unreliable data and does not provide an adequate basis for the evaluation of antidepressant withdrawal effects. Further good-quality research on antidepressant withdrawal is required.