Posted on April 18, 2019 at 4:52 PM
Editorials appear in cooperation with the American Journal of Bioethics. This article and its associated pieces can be found here.
by Jerry Menikoff
How can we appropriately ensure that people who enroll in clinical research understand what they are getting into? The recent revisions to the primary set of U.S. regulations for protecting research subjects (the Common Rule) attempt to better achieve that goal by, in part, providing more useful information to prospective subjects. In this issue of AJOB, Stephanie Morain and her colleagues address another important aspect of informed consent to research: Who should be obtaining consent? They note that although a variety of ethical guidelines warn against having physician-investigators obtain consent from someone who is already their own patient, U.S. regulations do not actually prohibit this. Indeed, as their article states, guidance from the Department of Health and Human Services’ Office for Human Research Protections specifically provides that institutional review boards (IRBs) may use discretion in determining when to allow this.
In an effort to advance the conversation about the ethics of this practice (what they refer to as “dual-role consent”), their article seeks to push back against the widespread condemnation of this practice by the ethics community. Suggesting that that there should be an appropriately nuanced approach to this issue certainly seems reasonable. As a result of their article, there are now a host of suggestions for ways to deal with this issue coming from the open peer commentaries that accompany the target article.
While Morain and her colleagues are to be congratulated for giving this issue greater prominence and encouraging further discussion, one aspect of their specific conclusions on how to draw the line between acceptable and unacceptable uses of dual-role consent merits further analysis. They note that as “the key features of research participation—including risks … more closely approximate those of usual care,” it should become more acceptable “for physician-investigators to seek consent from their own patients.” They contrast this with circumstances “when a trial deviates in important ways from what is typical of clinical care.” These considerations lead them to derive one of their two conditions for when dual-role consent should be permitted: that “research participation would not substantially disadvantage the patient as compared with clinical care.” They further note that in such studies, “patients are unlikely to have any preference between the two arms.”
On its face, this seems like a reasonable distinction. What they fail to highlight, however, is that the line they are drawing—one that apparently looks to the extent to which the arms in a trial might be versions of current approaches to clinical care—is one that has been raised, and extensively discussed, as part of another ongoing debate that has been taking place over the past several years. Moreover, based upon the references they cite and the vagueness of the line they attempt to draw, readers of this article who are not already intimately familiar with the prior “usual care” debate might implement the proposed standard in a less than ideal way. Indeed, it might be the case that relatively few clinical trials are actually going to meet the standard they propose, yet that conclusion is not readily ascertainable from reading the article.
And far more importantly, such readers might apply the line that Morain and colleagues draw not only to issues with regard to dual-role consent, but also to a far wider range of issues, including the determination of what constitutes appropriate informed consent in “usual care” research. This could end up undermining the ethics of informed consent in such studies, an outcome that would be unfortunate.
In the limited space allotted, it is difficult for me to do justice to the array of issues that have been raised with regard to “usual care” research, including the risks it poses to subjects. But by focusing on two of the papers that Morain and colleagues refer to in performing their analysis, it is possible to highlight one of the most relevant aspects of this debate.
The two papers in question are those cited as Cho et al. and Kraft et al.. Both of the papers described empirical results obtained by researchers studying what is referred to as “research on medical practices,” for which they use the acronym ROMP. The two papers state that their work was motivated in response to the ongoing debate about the risk level of clinical trials that randomize subjects between different versions of “standard” care. In order to provide empirical data on that issue, the researchers had people view three videos about research, and then asked them a variety of questions, including ones about the risks of the research and how consent should be obtained. In both papers, one of the scenarios people were asked about involved a hypothetical randomized study of two unnamed commonly used drugs to treat hypertension. Participants were told that “data was lacking” with regard to the relative effectiveness of the two drugs. They were also told that both drugs “cause mild side effects that are similar in frequency.”
What is particularly noteworthy about the hypothetical scenario that was presented to the participants is that they were provided with not a single factor on which a person (and not just a patient, but even a doctor who would be choosing which drug to prescribe) could justify a preference for one drug over the other. (That conclusion appeared to be reinforced by one frame that may have appeared in one of the videos, currently highlighted on the home page of their website, which shows three doctors, each with a huge smile on his or her face, holding one of the three antihypertensive medications, and each saying—with exclamation points—that the medicine is good! Any such choice would be pure guesswork—unencumbered by the thought process, as the “Car Talk” brothers famously said. Thus, when asked about such a study, it is actually quite reasonable for someone to conclude it does not involve much risk, and that there isn’t a need for much in the way of informed consent.
What is fascinating about these ROMP articles, and the hypothetical clinical trial they asked people about, is that they do not actually directly address the type of trials that had generated the controversy about how to obtain consent in “standard care” research. In effect, they changed the topic of discussion, without even acknowledging that fact, by instead focusing on a particular type of hypothetical trial—one where there are no suspected differences between the two “usual” treatments being compared, and thus no grounds on which to reasonably choose between them. (A similar approach—picking exactly that type of hypothetical trial as the example—was taken by the New England Journal of Medicine in an editorial that had the same goal as the ROMP researchers, namely, to demonstrate that such trials involve little risk to subjects.
To better appreciate this conclusion, merely contrast this type of hypothetical study with the very real study named SUPPORT that triggered much of the ongoing debate about “usual care” research. In SUPPORT, premature infants were randomized to two levels of oxygen—high versus low—to see whether receiving low oxygen would reduce the likelihood that the infants might develop an often severe visual problem called retinopathy of prematurity. The study was also designed to determine whether the low-oxygen group would receive that benefit without also experiencing a greater likelihood of permanent brain damage or an increased risk of death.
Prior to beginning that clinical trial (according to some versions of the consent form used in the trial), many such infants were provided a level of oxygen that was in the middle of the range of oxygen levels that constituted the usual range of standard care. By participating in the trial, half of those infants would be moved to a lower oxygen level—thus raising a concern about possibly increased risks of brain damage or increased mortality. The other half of the infants would be moved to a higher oxygen level than they would otherwise have gotten—thus raising a concern about a possible increased risk of serious injury to their eyes, including blindness.
Obviously, this type of study, unlike the hypothetical “no suspected differences” hypertension example cited earlier, raises a number of specific possible risks that might motivate a physician to recommend, or a parent to choose, one versus another level of oxygen for their child. What if prospective subjects were shown, instead of a video featuring smiling doctors who found all of the medications to be good!, one doctor highlighting his proposal to use a low oxygen level, while outlining the possible risks versus benefits of that approach. A second physician would do the same for the high oxygen level, while a third would be proposing to treat the infant at an oxygen level in the middle of the range, and highlighting the possible risks and benefits of that approach. Many readers of this piece would likely agree that, after seeing such a video, quite a few of the parents would indeed have a preference among the various options.
The reality is that there may be very few clinical trials being proposed that are like the hypothetical “no suspected differences” hypertension study. In the real world, where there are limited resources and trials can cost tens of millions of dollars, it makes sense to actually compare two treatments where there is already a suspicion of a possible specific difference between them in terms of efficacy or side effects or both. Furthermore, for the same reasons, the suspected differences in such trials may often relate to something having to do with important health consequences, and thus an outcome about which a prospective subject will often have a preference.
It is noteworthy that even the one real trial that Morain and colleagues chose as a good example of where dual-role consent is appropriate—meaning that prospective subjects should presumably be unlikely to have any preference between the two arms—doesn’t seem to very clearly meet that test. The trial involved randomizing subjects between receiving peripheral-blood stem cells versus bone marrow from an unrelated donor. There had already been randomized trials demonstrating that, for high-risk patients who had a matched sibling as a donor, the use of peripheral-blood stem cells led to improved survival—with the consequence that use of those cells had become the norm (and was then being used in 76% of such transplants). It certainly seems more than plausible that many patients, for good reason, would be very interested in making sure they got unbiased information from a disinterested party before deciding whether to participate in the trial (and therefore possibly being denied the treatment that more than three-quarters of doctors were then recommending).
Another distinct point that highlights the rarity of the “no substantial disadvantage compared with usual care” trial is that for many of the types of medical issues for which there is uncertainty about how best to treat a patient, it may not be that easy to even determine what constitutes “usual” care. To the extent that a trial gets that wrong—and assigns people to what is claimed to be usual care when it actually isn’t—that in itself is exposing subjects to a “misalignment” risk that can be substantial. In this regard, it is noteworthy that even for the SUPPORT trial, there is now evidence suggesting that one of the arms in that trial did not even reflect what care doctors were actually providing prior to the start of the trial.
So, by all means let us try to draw better and more nuanced lines in determining when dual-role consent should or should not be allowed. But at the same time, this can become an excellent opportunity to educate people about drawing similarly good lines in determining when trials do indeed alter usual care sufficiently such that, if the relevant information is provided, many patients might actually have meaningful preferences about participating versus obtaining care outside the trial.