The association between placebo arm inclusion and adverse event rates in antidepressant randomized controlled trials: An examination of the Nocebo Effect

Abstract

BackgroundAntidepressant efficacy is influenced by patient expectations and, in randomized controlled trials (RCTs), the probability of receiving a placebo. It is unclear whether tolerability demonstrates a similar pattern. This study aimed to determine whether study design influences adverse event (AE) rates in antidepressant trials for subjects receiving active treatment or placebo. MethodsRCTs comparing one antidepressant to another antidepressant, placebo, or both in major depressive disorder (MDD) (1996-2018) were retrieved from Medline and PsycINFO. Clinicaltrials.gov was searched for unpublished trials. Of 1,997 studies screened, 77 trials were included. Studies were classified as drug-drug, drug-drug-placebo, or drug-placebo based on design and overall number of subjects experiencing any AE was recorded. Subgroup meta-analysis of proportions and meta-regression techniques were used to compare AE rates across study designs in patients receiving active antidepressant treatment and placebo. ResultsAmong the actively treated, AE rates were lower in drug-drug trials (58.5%) compared to drug-drug-placebo (75.7%) and drug-placebo (76.4%) (the model reported coefficients for percent differences between AE rates of different study designs were B=17.0, p<0.001 and B=17.8, p<0.001, respectively). AE rates in patients receiving placebo were not different between study designs. LimitationsThe present study is limited by the diverse range of study populations, variability in reporting of AEs, and specific antidepressants employed in the included trials. ConclusionsThe inclusion of a placebo arm in the study design was unexpectedly associated with higher rates of AEs among patients receiving active medication in antidepressant trials. This observation has important implications for interpretation of trial tolerability findings.