It was controversially concluded that significant antidepressant–placebo differences have not been established. Predictably, given the overwhelming evidence base supporting drug efficacy, the reaction against this paper was strong. McAllister-Williams states that the magnitude of therapeutic difference is the difference between drug and placebo, not absolute response to active drug and thus Kirsch’s study in fact supports the idea that antidepressants efficacy increases with depression severity [McAllister-Williams, 2008].

In addition, Matthew and Inhibitors,research,lifescience,medical Charney noted only group-level effects were addressed as analyses were based on differences in HDRS score between drug and placebo at Inhibitors,research,lifescience,medical the study endpoint [Matthew and Charney, 2009]. Huge variation at the patient level in antidepressant response is overlooked; indeed a patient-level meta-analysis found the magnitude of benefit of antidepressant treatment compared with placebo increased with the severity of depression [Fournier et al. 2010]. An interesting study by Fountoulakis and Mollera [Fountoulakis and Möller, 2011] has highlighted some important flaws in the calculations of Kirsch and colleagues’ meta-analysis

[Kirsch et al. 2008]. This recent study re-analysed data used in the meta-analysis and reported the correct drug–placebo difference to be 2.18 or 2.68, as opposed to 1.80 stated in the original study. In addition, Inhibitors,research,lifescience,medical Kirsch and colleagues failed to report the change in HAMD score was 3.15 or 3.47 points for venlafaxine and 3.12 or 3.22 Inhibitors,research,lifescience,medical for paroxetine, which are both above the National Institute for Clinical Excellence (NICE) threshold. Thus, it appears that reporting of results was selective and calculations were flawed, suggesting Kirsch and colleagues’ conclusions were unjustified. In addition, Kirsch and colleagues’ study suggests obtaining positive results against placebo is easy; however, failure of trialled antidepressants indicates this is not the case. MERCK and Co invested huge amounts of money into the drug aprepitant (an antagonist of the neurotransmitter G Protein antagonist substance P)

Inhibitors,research,lifescience,medical which failed to show advantages over placebo in phase III testing and was withdrawn. A final interesting point is that if antidepressants are not effective, why do patients respond to one antidepressant but not another? Following Kirsch and colleagues’ findings, differential improvement between antidepressants would not be expected. Resveratrol RCTs: gold standard or lead weight? Whilst Kirsch’s conclusion that antidepressants are not effective may go beyond the data, RCT methodological problems may mean antidepressant efficacy is overstated [Greenberg and Fisher, 1994]. One problem is unblinding, where patients know whether they are receiving antidepressants or placebo due to side-effects of psychoactive drugs. Thus, expectations associated with taking active medication may influence outcome, rather than true therapeutic effect [Toneatto and Sellers, 1992].