Thursday, February 3, 2011

Equipoise: No Help for Ethical Deliberation

This post might seem a bit off the wall for this blog, but I'll defend it by noting that we discuss ethical issues at the medicine-Pharma interface, and so unavoidably touch on the ethics of drug research more generally. The article I will discuss here (subscription required) makes some excellent points that have not previously been a part of that ethical discussion.

Frank Miller, a philosopher-bioethicist at NIH, and Steve Joffe, pediatric oncologist with the Dana-Farber Cancer Institute, write on the question of "equipoise." The standard view today of the ethics of randomized clinical trials takes off from the apparent dilemma of a physician enrolling subjects in a randomized experiment. Traditional medical ethics says that the physician's duty is to advise the patient with regard to therapy, according to the doc's best clinical judgment. This seems inconsistent with allowing a patient of yours to get a treatment, or not, at the toss of a coin. The solution, according to ethical thinkers back in the 1980s, was that a randomized trial is ethical only in a state of clinical equipoise--when the relevant community of physicians considers it a toss-up as to which of the two treatment arms is superior.

First thing to note is that the equipoise rule would prohibit the vast majority of placebo-controlled trials. Yet the FDA routinely requires placebo-controlled trials for most of the meds they approve. So tons of supposedly unethical trials are conducted to win FDA approval. Is the FDA unethical, or does this signal a flaw in the equipopise theory?

A few years back, Frank Miller and I argued that it's the equipoise end of the stick that has the problems. (Not that the FDA doesn't have problems, but that's another post.) We relied at that time heavily on a set of arguments that claim that clinical medicine and clinical research are two quite different enterprises and require different ethical rules. To argue that the clinician trial investigator has the same duty to the research subject as the physician has to the patient mixes up those different sets of rules.

Miller and Joffe here expand that argument considerably by arguing that it's not just a matter of conducting research according to the ethics of the individual physician-patient relationship, erroneous as that may be. There's another issue at stake. The equipoise argument is based on the idea that the only "consumer" of the research findings is the individual doc trying to make treatment decisions for individual patients. If that doc is in a fog about which of two treatments is superior for any given patient, then it's presumably ethical to do a randomized trial. But as soon as that doc becomes unfogged, and has a clear preference for one treatment over the other, then scrap the trial. (Ending trials early, before they allow definitive conclusions, is one deleterious consequence of equipoise that Miller and Joffe list.) Or in the case of clinical equipoise, stop the trial when the relevant community of docs becomes unfogged to a sufficient degree (whatever that is, yet another problem with equipoise).

This view, say Miller and Joffe, is fundamentally incomplete. There are other important consumers of this research, namely policymakers, including those often-derided folks at the FDA. And policymakers don't have the luxury, or the burden, of dealing with patients one at a time. These folks are duty bound to adopt a population perspective, asking what the impact of a drug (for example) is on the entire patient population. To get such data, randomized trials are usually needed (not always), and the trial may have to go on often beyond the stage at which individual physicians have reasonable preferences for one treatment over the other. The policy need for these data is legitimate and ethically sound, and so it's shortsighted to invoke "equipoise" as a reason to rule out those trials, or to stop them too early.

Bottom line--ethics is important in research and the rights of subjects must be adequately protected, but the concept of equipoise is quite a wrong-headed way to go about that task.

Miller FG, Joffe S. Equipoise and the dilemma of randomized clinical trials. New England Journal of Medicine 364:476-80, Feb. 3, 2011.

Miller FG, Brody H. A critique of clinical equipoise: therapeutic misconception in the ethics of clinical trials. Hastings Center Report 33(3):19-28, 2003.


Nancy Wilson said...

The authors’ failure to provide an operational definition for equipoise weakens their argument, in my opinion. The more pressing problem here is the lack of evidence-based literature on the concept of equipoise.

I like Christine Grady's take on equipoise in Principles and Practice of Clinical Research (2002) on page 23:

“It is important to recognize that there is a distinction but not necessarily a contradiction between a state of equipoise, necessary for the conduct of an RCT, and offering a treatment recommendation for an individual patient. Not knowing which of two or more interventions is superior in terms of long-range outcomes for a group of patients does not preclude judgments about what is best for a particular patient at a particular time. It may be possible to make a recommendation for an individual in favor of one of two unproven interventions based on the individual’s unique symptoms, side effects, values, preferences, etc. The greatest tension in this regard may be in the case of a clinician, responsible for the care of patient, who also is the investigator of a study in which the patient is a subject. Being aware of this tension, clearly informing the patient, relying on other members of the team, or in some cases, separating the roles of clinician and investigator may be necessary so that the patient’s needs are not overlooked.”

anti snore said...

. It may be possible to make a recommendation for an individual in favor of one of two unproven interventions based on the individual’s unique symptoms, side effects, values, preferences, etc.

Howard Brody said...

Nancy Wilson may have inferred from my terse comments that it was the authors of the paper I am reviewing here, Miller and Joffe, who failed to provide an operational definition of 'equipoise.' I intended rather to convey that Miller and Joffe make the claim in their paper that it is the advocates of the equipoise criterion who have failed to offer any workable, practical definition of when equipoise holds and when it doesn't. That failure is one of the reasons to reject equipoise as a workable ethical principle in research. I also agree that Christine Grady's comment is an insightful one, and makes clear the important ethical diference between the respective roles of treating physician and trial investigator.

Nancy Wilson said...

Hello, Howard.

I inferred from the essay by Miller and Joffe that the authors themselves drew conclusions about the conduct of clinical research using the term equipoise--as in "equipoise-driven consequences," without adequately defining the term.

Whose responsibility is it to define the concept of equipoise, if not academicians and NIH bioethicists?

Alex John London (Two Dogmas of Research Ethics and the Integrative Approach to Human-Subjects Research) proposes an integrative model for clinical research that “respects the fundamental moral equality of all community members...." More here:

snore stop said...

This site was great, It may be possible to make a recommendation for an individual in favor of one of two unproven interventions based on the individual’s unique symptoms, side effects, values, preferences, etc.