Regulators of medical research face the daunting task of vetting protocols, drugs, and devices to ensure that they are appropriate, meaning that the benefit/risk is reasonable.

Understandably, there has been considerable focus on better defining both the benefit and the risk here, an effort that has paralleled the focus on "personalized medicine" - the right medicine for the right patient at the right time.

The basic idea here is that if you can define more precisely (hence the term "precision medicine") exactly which patients are or are not going to benefit, you significantly increase the benefit/risk ratio, as patients unlikely to respond aren't exposed to a drug, and you increase the likely benefit achieved by patients ultimately receiving the product.

Segmenting patients can also be done using phenotypic characteristics - assessing HER2 expression level to determine whether to administer Herceptin use is a common example, but you can also think about a range of physiological or even behavioral parameters as well.

On the face of it, this would all seem to be good news - and to an extent, it is: defining with greater precision the likely benefit and risk faced by a patient potentially enrolled in a clinical study or potentially using a particular product should lead to more informed decision making.

The challenge, however, is that this conceptualization misses the fact that patients differ not only in their genetic and phenotypic characteristics, but also in the way they perceive risk/benefit, as described by Michelle Meyer, a Harvard legal scholar who offers in a recent paper perhaps the most eloquent and insightful perspective on what she terms "the heterogeneity problem."  Focusing on IRBs, she writes,

"...even if all prospective participants could expect the same costs and benefits from participating in a study, they are very likely to differ in their willingness to assume those risks in pursuit of those benefits.  Yet IRBs must assign a single risk-benefit profile to each study, and then determine, for all prospective participants, whether that risk benefit profile is ‘reasonable.'"

Bingo.

It's important to recognize that the key issue here is one of de-averaging, of needing to recognize that a consensus assessment may not accurately reflect the divergent views of individual participants.  More concerning, as Meyer astutely observes,

"By employing a broad understanding of research risk and a narrow understanding of research benefit, IRBs tend to characterize the risk-benefit profiles of proposed studies in ways that reflect the preferences and other circumstances of often-idiosyncratic, hypothetical eggshell participants.  Erring on the side of restricting research opportunities to which individuals may be invited is benign only if we assume that individuals can only be harmed, and never benefited, by participating in the research, a claim that is refuted by much of the empirical evidence discussed in this Article."

Bingo, again.

Furthermore, while Meyer's comments here are aimed at IRBs, they could just as easily apply to federal regulators of medical products as well - something familiar to anyone who's ever watched an advisory committee meeting.  More often than not, individual committee members presented with the same exact information will come to very different conclusions regarding whether a product's benefit/risk profile is "reasonable."

Without question, patients would also have a range of views here, and the real challenge going forward is refining our approach to regulation to more effectively preserve patient choice.  If a drug is shot down, that choice disappears.

The goal here must be what I call "personalized regulation" - an approach that preserves patient choice - at least to a greater extent than is available today.

Some might invoke a variation of what might be called compassionate paternalism, and point out that since individuals often make bad choices (a point convincingly made by Kahneman and others), it's important that wise regulators protect patients and physicians from themselves.

This is reasonable, but only to a point.  It is misleading to assume implicitly that if a patient has a different view of risk/benefit than a committee's consensus, the individual's view must somehow reflect deranged thinking on the part of the patient.

To the contrary, regulation must endeavor to more respectfully, expansively, and creatively accommodate the rational preferences of diverse patients in order to avoid leaving so much opportunity on the table.  If we can agree on the need for such personalized regulation, we are more likely to find a way to achieve it.