Side effects breaking the blind does not explain the antidepressant effects of the selective serotonin reuptake inhibitors

Other conference contribution, 2016

Introduction: Both in the scientific literature and in lay media the contention that antidepressant drugs exert no genuine pharmacological antidepressant effects at all has recently gained considerable attention and endorsement [1]. According to this theory, the reason why antidepressants generally outperform placebo in controlled trials is that they, unlike placebo, cause side effects. This disparity, it is argued, may compromise the integrity of the double-blind procedure as patients receiving a pharmacologically active treatment are more likely to experience side-effects and thus to conclude that they are indeed receiving an active compound and not placebo. This realization could then, putatively, augment an expectancy-based placebo effect. According to this “side-effects breaking the blind”-hypothesis, antidepressant drugs might hence be regarded as nothing more than “extra strength placebos” and their observed efficacy merely as a consequence of a flawed trial design.
Aims: The present study was undertaken to challenge this hypothesis by investigating if the presence of self-reported side effects may indeed explain the superiority of citalopram, sertraline and paroxetine over placebo in drug company-sponsored trials.
Methods: Pooled patient-level, post-hoc analyses of all company-sponsored, acute-phase, placebo-controlled trials using the Hamilton Depression Rating Scale (HDRS) and conducted to evaluate the effect of citalopram, paroxetine or sertraline in adult major depression were undertaken. The single item depressed mood, which has proven a more sensitive measure to detect an antidepressant signal than the conventional one [2], i.e. the sum score of all HDRS items (HDRS-17-sum), was regarded as the primary effect parameter whereas the full HDRS-17-sum was used as a secondary effect parameter. The primary analyses were conducted using ANCOVA models with the endpoint score on the outcome measure as the dependent variable and the baseline score on the corresponding measure as a covariate. Treatment (SSRI or placebo), trial, and side-effect status (0 side-effects or ⩾1 sideeffect) were included as fixed factors, as well as the interaction between treatment and side-effect status.
Results: With respect to the reduction in depressed mood, the effect size for the difference between active drug and placebo after 4 weeks of treatment was 0.29 (95% C.I. 0.16 − 0.43; p<0.001) in patients not experiencing any adverse events. For patients that did report at least one adverse event, the corresponding effect size was also 0.29 (95% C.I. 0.22 − 0.35; p<0.001). The results of the analyses using HDRS-17-sum as effect parameter were similar to those of the primary analysis. Sensitivity analyses comparing patients reporting any early side effects (i.e. during the first week of treatment) with those not reporting an early side effect also yielded similar outcomes.
Conclusions: SSRI outperforming placebo also in subjects not reporting side effects argues against the hypothesis that side effects breaking the blind is a major contributor to the antidepressant efficacy of the SSRIs.