I recently had occasion to think fondly of a member of our PRIM&R community whom we lost in April: Alan Wertheimer, PhD. Alan was a senior research scholar in the Department of Bioethics at the National Institutes of Health, professor emeritus of political science at the University of Vermont, and a long-time PRIM&R faculty and conference planning committee member. He was also a cherished friend and mentor to me. He passed away April 10.
Ever the scholar and truth-seeker, Alan never hesitated to let me know—collegially, of course—if he disagreed with something I said. For instance, he wrote me a note about my Ampersand post on informed consent last year, and in it, reminded me that it’s actually a matter of controversy within the bioethics community whether lack of comprehension on the part of a potential research subject invalidates that person’s informed consent. At the time, I had considered including that dimension of the debate around consent in my post, but ultimately decided not to, in the service of keeping the post brief and focused on the central point of not using consent as a transitive verb.
But as I was reflecting on this exchange recently, I went back and re-read the Kennedy Institute of Ethics Journal article Alan and his colleague Frank Miller wrote in 2011, in which they argued for a model of consent (which they called the “Fair Transaction” model) that does not make comprehension a requirement for valid consent. It’s a brilliant paper—though in the end, I’m not sure I agree with their view (more on that later)—and when reading it, I was reminded just how significant are Alan’s contributions to research ethics scholarship. So I thought I’d use this post to both pay tribute to Alan, whom I miss very much, and shed some light on this controversy, by taking a brief look at Alan and Frank’s conception of informed consent.
According to the traditional model of informed consent (often called the autonomous authorization (AA) model), substantial comprehension on the part of a subject is a requirement for valid informed consent. But that model, the most comprehensive account of which can be found in Ruth Faden and Tom Beauchamp’s seminal book, A History and Theory of Informed Consent (Oxford University Press, 1986), runs into a real problem, namely, the therapeutic misconception (TM). TM is the phenomenon whereby clinical research subjects fail to understand the difference between research and ordinary treatment (including, for example, that research follows a fixed protocol which precludes making personalized decisions about care), and therefore mistakenly attribute therapeutic intent to research procedures (Lidz C. and Appelbaum, P. “The Therapeutic Misconception: problems and solutions .” Medical Care, 2002). There is evidence that a substantial percentage of clinical research subjects manifest TM (see Appelbaum, P. and Lidz, C. “The Therapeutic Misconception,” Oxford Textbook of Clinical Research Ethic, Oxford University Press, 2008))
Since TM constitutes a lack of substantial understanding, on the AA model, it would seem to follow that if a subject consents to research participation while under the TM, his or her consent is invalid and the research unethical. In their paper, Miller and Wertheimer point out that most people who subscribe to the traditional view of consent nevertheless reject that conclusion in practice—that is, they don’t acknowledge that TM invalidates informed consent or that a considerable portion of current clinical research is unethical. According to Miller and Wertheimer, this inconsistency between what the model entails and what people who subscribe to the model are in practice willing to conclude suggests a flaw in the AA model of consent. So they suggest an alternative.
According to Miller and Wertheimer, requiring substantial comprehension for valid consent in all cases can actually be a failure of respect for persons. To show this, they ask us to consider two cases: first, a case where subjects are randomized in a trial comparing two standard treatments for depression that are expected to offer similar risks and benefits, and where there has been clear disclosure by investigators about the nature of the research, but some of the subjects fail to appreciate that they will be randomized to one of the two interventions, believing instead that the doctor will select the intervention that is best for them. They argue, “Despite the subjects’ defective comprehension, it seems disrespectful not to recognize the validity of their less than autonomous consent, given their choice to participate after receiving adequate information about the trial and given the personally favorable risk-benefit ratio of trial participation” (207). In other words, Miller and Wertheimer suggest that though these subjects’ consent may not be fully autonomous (because it is given without full comprehension), it is nevertheless freely given, and to discount such freely given consent by saying they nevertheless cannot participate in the trial seems disrespectful. Furthermore, they argue, it seems unfair not to let these subjects participate, in light of the prospect of benefit and lack of predictable harm from participation.
If, on the other hand, we are talking about a placebo-controlled trial of an experimental anti-depressant in the same population, then things look very different. In this second case, Miller and Wertheimer argue, the experimental intervention involves much greater uncertainty about risks and benefits, and randomization to placebo means that some subjects will get something that departs dramatically from standard treatment for depression. (I leave aside, as they do, the fact that some would argue that conducting a placebo-controlled trials in such a case is unethical). Manifesting a TM in this case, Miller and Wertheimer argue, does invalidate consent, because subjects would be failing to appreciate the significant possible disadvantage to them of not receiving active treatment by being randomized to placebo. That is, here the failure to understand the difference between research and treatment constitutes a failure to understand the much less favorable risk-benefit profile of the research, and, in particular, that there is a “known potential for disadvantage” if randomized to placebo (210).
Miller and Wertheimer thus suggest that by placing comprehension above all else, the traditional model of consent is inappropriately insensitive to other considerations that are relevant to respecting persons and their well-being, such as advancing their interests and values, and respecting their preferences—including, for instance, preferences not to be excluded from research that has a personally favorable risk-benefit ratio. The model is thus unfair to subjects. Miller and Wertheimer argue that determining what level of understanding is required for informed consent to be valid should be sensitive to the risk-benefit profile of the research in question. For low-risk research, there is nothing ethically wrong in accepting as valid a subject’s less-than-fully-autonomous consent, though there should always be good faith efforts to communicate information about the research in language the potential subject can understand, and other regulatory safeguards and research protections mechanisms, such as regular ethics review of risks and benefits, must, of course be in place.
Interestingly, Miller and Wertheimer also suggest that the traditional model of consent is unfair to investigators. “The lengths to which investigators should be expected to go in assessing the quality of informed consent should be reasonable in view of the risk-benefit profiles of different studies” (210). While they agree that, in general, investigators should assess the capacity of subjects to consent and enroll subjects only when they are “reasonably confident that they have adequate comprehension,” they think that insisting on rigorous testing of comprehension in low-risk studies is unnecessarily costly and unreasonable, and thus unfair to researchers.
And so, Miller and Wertheimer propose instead a Fair Transaction model of consent, which does not require substantial comprehension in all cases of valid research consent, as the AA model does, but is instead, sensitive to context. What fairness requires, with respect to comprehension, will vary depending on the risk-benefit profiles of the research in question. They write, “It is fair to both subjects and to investigators to require more stringent efforts to promote and to test comprehension when the negative consequences to subjects from trial participation are potentially more significant as compared with standard medical care,” but unreasonable to require “exacting scrutiny of consent to low-risk, life-saving treatment.” And with respect to the TM, specifically, they argue that there need be only minimal concern about correcting it in trials that compare similar treatments in conditions that are like standard practice (i.e., pragmatic trials), and more concern in trials that pose potentially important disadvantages to subjects (212).
Thus, according to Miller and Wertheimer, and contrary to much standard thinking, “Autonomy is not the decisive consideration in the validity of consent” (212). Rather, consent is valid “when subjects are treated fairly in making decisions about research participation consistent with their preferences and values” (217).
So what to make of Miller and Wertheimer’s critique of the traditional view of informed consent, and of their alternative? Though I find their arguments thought-provoking, I wonder whether their account misses some of the point of seeking informed consent for participation in research by focusing so much on the relevance of the risk-benefit profile of the research in question. That is, I’m not convinced that the level of risk subjects are about to undertake in entering a research study should be the arbiter of how much concern we have about whether they understand the difference between research and clinical care. Rather, respect for research subjects would seem to require that we always be concerned that people who are about to enroll in research understand what it means to participate in research—an enterprise whose goal is the creation of generalizable knowledge rather than achieving a personalized benefit, and whose procedures are therefore governed by a standard protocol, rather than personalized decisions about care. That is not to say that the informed consent process need look exactly the same in high and low risk cases. But what seems objectionable to me is the idea that if the risks are low enough, we need not be too worried whether people understand that they are research subjects, with all that that entails.
What do you think?
I was distracted, while reading these comments, by the unspoken assumption that understanding or not understanding the content of an informed consent process is somehow in the purview of the subject. Whether or not a person understands what is happening depends, we seem to assume, on something in the person who is supposed to be getting informed.
I don’t think that is the case at all. In most of the consent processes I have reviewed, the chances that a prospective subject actually understands or is even able to understand the information provided are somehow between low and non-existent. The reading level needs to be that of People magazine; instead, it is above that of The Economist. 15 second Superbowl ads and 140 character tweets are the norm for actual communication; yet we expect someone to grasp the content of 18, 24, even 32 page consent forms.
I hope the current national effort to improve the HRP/IRB system will result in a consent process that actually works. Lacking that, I don’t think a discussion of how much a research subject needs to know can be anything more than the intellectual equivalent of a discussion of who is likely to win the NFL Playoffs.