Abbreviations used in this article ADR adverse drug reaction ES effect size RCT randomized controlled trial TAU treatment as usual Can J Psychiatry 2008;53(7):430-432 It is widely accepted that the most powerful research design to assess the efficacy and effectiveness of a novel intervention is the RCT, in which one group of eligible patients is randomly assigned to receive the new drug and the other group receives something else. The question is, What should constitute that comparator condition? In situations where there is no effective treatment or where the existing treatment has many undesirable side effects, the answer is relatively straightforward: the comparison group should be a placebo. A much more difficult situation arises when an effective treatment does exist, such as selective serotonin reuptake inhibitors for depression or the atypical antipsychotics for schizophrenia. More than a decade ago, Rothman and Michels1 argued very cogently that under such circumstances, it is a violation of the Declaration of Helsinki to have a placebo arm in the trial, as that would deny patients available treatment, and that the comparison should be against TAU. However, the sheer number of letters to the editor in response to that article is an indication that the issue is neither clear-cut nor settled. At first glance, Rothman and Michels' points are very compelling. If a drug already exists, how can we refuse to give it to some patients? From a scientific point of view, why would we even consider using a placebo? Aside from the pecuniary reason of pharmaceutical companies wanting to introduce their me too version of a drug, it would seem reasonable that the correct scientific question should not be, Does this new drug also work for this condition? but rather, Does this new drug work better or have fewer ADRs than the existing treatment(s)? My position is that this argument would be unassailable if this were le meilleur des mondes possibles. However, despite Dr Pangloss' unwavering belief, this is not the best of all possible worlds. In light of that, placebo-controlled trials are a necessary evil, even when effective drugs exist. My arguments are 2-fold: first, the results of studies with 2 active drugs may be ambiguous and impossible to interpret; and second, more people are potentially left untreated or harmed in trials with active controls, compared with placebo controls. Regarding the first point, there are 2 possible outcomes in any clinical trial: the study finds a statistically significant (and ideally, clinically important) difference between the drugs in one direction or the other, or it does not. If the first situation is obtained, then there is no ambiguity: the new drug was either superior or inferior in comparison to TAU, and should or should not be adopted. The problem arises when the study finds no significant difference. This can occur for 1 of 2 reasons: both drugs were equally effective, or in this trial, both drugs were equally ineffective. However, if TAU has been shown to work in earlier studies, how can it occur that it was ineffective in the current trial? This is not merely a theoretical question. One metaanalysis found that, among 52 trials, only 48% of antidepressant arms showed superiority over a placebo; and the same proportion was seen in 13 trials of anxiolytics.2 Let us go through the steps of executing an RCT with a new drug. First, patients are assessed to determine whether they meet the inclusion criteria. second, those who satisfy the criteria are randomly assigned to one of the 2 groups. Third, therapeutic dosages of both drugs are given in a doubleblinded manner to all patients for a time sufficient for them to show an effect. Fourth, relevant outcomes are measured in an unbiased and blinded manner for all patients who entered the trial. Finally, the data are accurately and competently analyzed to see if the new drug is superior, or equivalent in effect but has fewer ADRs. …
Search from 250 M+ research papersSearch from 250 M+ research papers
Open Access
Jul 8, 2022
Max Goodman + 1
