In its December 15 report, the Presidential Commission for the Study of Bioethical Issues (the Commission) makes 14 recommendations to improve the current system of research protections.
One of the most nuanced and interesting parts of the Commission’s 193-page report is the section on ensuring ethical study design. It is relatively uncontroversial that scientifically flawed studies are unethical—they expose subjects to risks without any corresponding benefit to society. (This is not of course to say that determining what is a scientifically sound study is always straightforward.) The really difficult ethical questions about study design emerge when rigorous scientific design threatens to come at the expense of adequately protecting or respecting human subjects.
Regarding ethical study design, the Commission favors a “middle ground” position between the “placebo orthodoxy”—the view that comparing an experimental drug ensures scientific validity, and that it is therefore ethical to use placebos as controls, even when alternative therapies exist—and the “active-control orthodoxy”—the view that placebos are never ethical if an alternative therapy exists, and that, therefore, the best available alternative must be used as the control. In brief, the Commission suggests that if certain criteria are met, some research designs can be ethically justified if control arm subjects receive less than the best-proven treatment.
The Commission offers three specific criteria for evaluating the methods and risks of study design and determining if the study is ethically justified. One criterion is that the design be adequate to yield usable results, which, in turn, places differing methodological constraints on trials, depending on the type of trial one is conducting (e.g. placebo-controlled or equivalency design).
A second criterion involves looking at local treatment standards to determine an appropriate control arm. The Commission is careful to say that this does not mean that if, for instance, the local standard of care is no care at all, it follows that an ethically appropriate control arm for a study in that location is placebo. But neither should the standard of care against which an experimental intervention is tested always be the “best-proven” treatment, regardless of where the study takes place—after all, “best-proven” treatments may not be the best for a particular population. Rather, “standard of care” is a normative, but also contextualized notion—it is determined by local features such as underlying health status and behaviors of the population, as well as the medical and logistical infrastructure.
The third criterion is that the risks to which subjects are exposed—which may be higher in trials in which, for methodological reasons, subjects are receiving less than the optimal standard of care—be minimized. This means that protocols for studies that pose additional or greater risks to subjects by departing from current best-proven treatments as controls should include not only explicit justification for that design, but also mechanisms for closely monitoring those subjects as well as for addressing serious symptoms and withdrawing subjects who experience adverse events. But it also means that a study meeting the other two criteria might nevertheless be determined to be ethically problematic on this third ground. That is, if by receiving a less then optimal standard of care a subject is exposed to substantially increased risk of mortality and morbidity, or severe discomfort, then a study in which that standard of care is the control is not ethically justified.
Just before it summarizes its discussion of the three criteria in recommendation 12, the Commission suggests that “IRBs should scrutinize the elements of study design as a distinct focus of their review.” Presumably, then, what the Commission is suggesting is that determining whether and when these criteria have been adequately met should be part and parcel of an IRB’s discussion and deliberation about a research protocol.
What do you think of these as criteria for IRB review? Do they represent an expansion of review criteria, or are IRBs already, implicitly or explicitly, using such standards? Is it feasible for IRBs to apply these criteria? Would IRBs have adequate information to apply these criteria? I look forward to hearing your thoughts.