The antidepressant debate and ethically defensible placebo-controlled trials

IRB: Ethics & Human Research, Nov-Dec, 2008 by Duff R. Waring

A basic tenet of biological psychiatry is that placebo-controlled, randomized drug trials under double blind have shown that antidepressants exert pharmacological efficacy--in other words, that they induce specific and measurable drug effects on research subjects. But there is ongoing dissention about whether pharmacological efficacy has been established reliably by these trials.

The antidepressant debate indicates a split in the expert clinical community about the size of the drug/placebo difference and whether the mechanisms that produce it stem from the subjects' expectation of benefit or from the intervention's pharmacological properties. A basic tenet of research ethics is that if standard effective treatments exist, then clinical equipoise requires their use as active controls against which novel, nonvalidated treatments are tested. Substituting a placebo for an effective treatment in the control arm of a trial does not maintain this duty, as it exposes research subjects to an intervention that is known to be inferior. Defenders of clinical equipoise argue that "conducting a trial is not an invitation to practice substandard medicine." (1) They have also argued that legal liability concerns might be raised about placebo-controlled trials of antidepressants. (2)

These ethical and legal concerns can be assessed with reference to cases in which clinical equipoise permits placebo-controlled trials. Given the recurring questions about whether the drug/placebo difference results from specific biological effects of antidepressants or from factors introduced by placebo response that have not been adequately excluded, I think a plausible argument can be made that clinical equipoise would permit the use of active placebo controls in the clinical trials that test them.

This argument can be abbreviated to three premises and a conclusion. The first premise is that placebo controls for drug trials are ethically defensible and scientifically justified when we have evidence leading to a reasonable concern that a standard treatment might only work because of expectancy enhanced by side effects--i.e., by enhanced placebo effects. The second premise is that this evidence must be endorsed by at least a respectable minority of the expert clinical community. Premise three asserts that a respectable minority of the expert clinical community endorses evidence leading to a reasonable concern that standard antidepressants might only work by enhanced placebo effects. If these premises are sound, then one can conclude that placebo controls for antidepressant trials are ethically defensible and scientifically justified.

Clinical Equipoise, Placebo Controls, and the Antidepressant Debate

Despite the controversy over their appropriate use in research, (3) Freedman allowed for five "broad classes of cases" in which placebo controls are both ethically justified and scientifically necessary. Two might be relevant to the antidepressant debate. These are cases in which the standard treatment for a condition has been shown to be no better than placebo and cases in which the standard treatment "has been called into question by new evidence warranting doubt concerning its presumed net therapeutic advantage." (4)

I do not think the antidepressant debate falls within the first case. While numerous studies have failed to demonstrate efficacy, many standard trials have shown at least a small but statistically significant difference between antidepressants and placebos. Whatever the size and cause of the greater antidepressant effect, we know an effect exists. (5) Therefore, the antidepressant debate is not driven by evidence-based claims that there is no difference between antidepressants and placebos. I suggest it is driven mainly by evidence-based claims that the average difference revealed in meta-analyses of standard trials is very small, and by methodological concerns that this difference is an artifact related to the breaking of the blind. In sum, "Even critics of antidepressants acknowledge that a genuine difference exists between antidepressants and placebos, with the debate focusing on how large this difference is and the mechanisms responsible for it." (6) An institutional review board (IRB) that adheres to clinical equipoise should not currently approve a placebo-controlled antidepressant trial under Freedman's first case.

However, I think a plausible argument could be made for an IRB approving a placebo-controlled antidepressant trial under the second case. It "involves those treatments that have fallen under suspicion because of previously unsuspected side effects, because of new information garnered about the disease process in question, or otherwise." (7) This case also covers "an interesting subset" of cases in which we suspect that a drug might only work "because of expectancy enhanced by side effects." We might have "new evidence" leading to a "reasonable concern ... that all of a drug's therapeutic benefit is a mechanism artifact, grounded in what were thought to be side effects of the drug in question." It would thus be "clinically reasonable to investigate and control" for these side effects on the grounds that "there are safer ways" of producing them. Active placebos would have a "favorable net therapeutic advantage." (8)