Factors contributing to the increasing placebo response in antidepressant trials.

Abstract

In their interesting paper, Khan and Brown (1) sought to summarize the factors contributing to the trend of continuously decreasing drug-placebo differences in antidepressant randomized controlled trials (RCTs) (2). This trend conveys the impression that the newer marketed antidepressants are less efficacious than the older ones, or even that the older and well-established antidepressants have lost efficacy over time. Thus, recognizing the methodological reasons for the decline in antidepressant-placebo separation within RCTs is highly relevant in order to appraise the clinical value of an antidepressant for psychiatric routine care. This is especially meaningful with regard to newly developed and newly introduced antidepressants. In our opinion, the constant increase in placebo response over the last decades is the main factor accounting for the mitigation of the drug-placebo differences in antidepressant RCTs. Due to the larger symptom improvement of study participants randomized to placebo, it is much more difficult for an antidepressant to outperform placebo at a statistically significant level. This could be the beginning of an ominous cascade. The diminishing drug-placebo contrast leads to a higher probability of so-called inconclusive or even negative trials. As a consequence, the risk for a newly developed compound to fail the market approval because of negative phase III studies increases. As such late failure is associated with enormous costs for the pharmaceutical industry, this could result in a slowdown of research efforts for new antidepressant agents. An often cited reason for the continuously increasing placebo response is the fact that placebo administration represents in itself a non-specific treatment (3). Study participants in the placebo groups of RCTs receive enormous clinical attention. Indeed, parallel to the raising placebo response in RCTs, the requirements concerning the accomplishment of RCTs have become more stringent, for instance through the need of an increasingly closer monitoring of participants (4). The intensive contact with the clinical staff can produce positive effects in terms of symptom improvement, particularly in not severely ill participants, who are nowadays increasingly included in RCTs, because the enrollment of more severely ill patients is often not possible due to ethical concerns (a phenomenon called “baseline inflation”). Another element contributing to the magnitude of placebo response in RCTs is the hope of participants in the placebo groups to receive an active, efficacious treatment (a phenomenon called “hope induction” or “expectation bias”). A number of systematic evaluations corroborate this assumption: the more study arms an RCT comprised (i.e., the lower the probability to receive placebo), the higher was the placebo response. On the other hand, the highest antidepressant response was found in direct drug comparisons (head-to-head trials), where participants were certain to receive active drug treatment (5). Furthermore, in many antidepressant RCTs, the enrolled subjects are not representative of clinical practice, in which a number of depressive patients suffer from severe comorbidities or suicidal ideation. Exactly those severely ill patients who are excluded from RCTs might particularly benefit from antidepressant pharmacotherapy. However, it must be noted that, in some analyses, a higher symptom severity at baseline was associated with a higher placebo response (6), a phenomenon which is not fully understood as yet. There is a large body of evidence suggesting that clinical studies carried out in the U.S. are characterized by a larger placebo response compared to non-U.S. trials. This phenomenon could be observed, to provide a recent example, in RCTs of vortioxetine (7). The inclusion of so-called professional research participants in U.S. studies may account for this finding. These subjects are mainly recruited by advertisements, and their motivation is often the prospect for free medication or other financial compensation. Therefore, they often aim to please the investigators in order to be invited again for participation in a clinical study. It appears meaningful to emphasize that the drug-placebo contrast in long-term, relapse-prevention studies is usually higher compared to acute-phase trials (8). Interestingly, in a double-blind, long-term citalopram RCT, responders to acute-phase treatment were randomized to either placebo or continuing citalopram, while placebo responders of the acute phase continued the placebo administration under double-blind conditions. Both placebo responders and citalopram responders receiving placebo in the long-term study exhibited a higher relapse rate compared to the participants in the citalopram continuation arm (9), suggesting that the underlying biological abnormality is not sensitive to placebo. In summary, the reasons contributing to the raising placebo response over time and the subsequent decreasing drug-placebo separation should be critically considered in the interpretation of antidepressant clinical trial results. Expectation bias and increased clinical attention are often called “unspecific effects” of placebo administration, and these effects are not present in routine clinical care (10). Therefore, it can be assumed that the effectiveness of an antidepressant in the routine care is higher than the antidepressant-placebo difference of RCTs indicates. The increase over time of placebo response should also be considered in meta-analyses (11), in which data are pooled from trials carried out in different periods.