Commentary on “Ethics as if Your Life Depended on It”

Introduction

Pearce (2026)'s provocative contribution asks us to reconsider our approach to the ethics oversight of medical research and medicines. Before new drugs can be administered, they must succeed in clinical trials, which are long, expensive, and require detailed and complex ethics review. While this review system reduces the harm done by unsafe and ineffective drugs, it also reduces the speed at which safe and effective drugs can be developed and disseminated, and some analyses suggest that current standards greatly overemphasize the prevention of harm from novel drugs over the prevention of harm from the status quo disease burden. In response, Pearce proposes lessening restrictions on drug access while increasing transparency and democratizing governance, on the grounds that this would both respect autonomy and rebalance risk-related tradeoffs to a more socially optimal position.

Pearce's central message is compelling. We should be attentive to the harms of untreated disease under the status quo, and when we succeed through departing from standard practice, as in Operation Warp Speed, we should consider whether there are lessons to be incorporated going forward. We should also take the perspectives of patients and participants seriously, and something has gone sideways when the ethical value of respect for persons is used to justify decisions that run counter to both the “respected” person's stated preferences and wellbeing.

Yet I also have reservations about whether, and in what respects, democratization is the answer to these problems. In considering tradeoffs between the risks of existing health threats and the risks from novel medical products, Pearce asks: “if your life were on the line, what would you want the ethical rules to be?” He then answers: “You would want the choice to be yours” (Pearce, 2026). However, my answer would be more qualified: I would want some, but not all, elements of that choice to be mine. Additionally, there are relevant considerations aside from what I would want. In this commentary, I first explain this difference in perspective, then return to address Pearce's specific proposed reforms.

Objectionable Paternalism

Here is an element of democratization I fully agree with: I want my own values, rather than the values of an IRB member or regulator, to be controlling with respect to appeals to my own good. Consider, for illustration, the following quotes drawn from individuals in hospice who were living with HIV and expressed a strong interest in HIV-related research participation: “I feel like these last few weeks are wasted;” “I wish I could do something else to help;” “At least I could be doing something” (Gianella et al., 2017, p. 3).

On a natural reading, these individuals are expressing a sincere judgment that research participation would be the best way to spend the end of their lives. Yet individuals in frail health are commonly excluded from research participation on the grounds of excessive risk. These individuals, naturally, also expressed that they would not appreciate being excluded on these grounds.

To disregard that view and forbid their participation would be paternalistic. Competent adults should generally be allowed to make decisions about their own good. This principle can be grounded in a framework of rights and dignity: competent adults have authority over themselves, and it is a form of disrespect and subordination to overrule their judgment of their own good. It can also be grounded in more practical considerations: competent adults know their own lives and situations better than anyone else, and they care more about their own good than others do, so they tend to do better jobs at running their lives than anyone else would.

Both rationales are available in this case. It seems disrespectful to disregard these patients’ own views about what would be a good way to spend the ends of their lives. It also seems substantively incorrect. Regardless of one's views of what makes for a good life, it is easy to see how research participation, for all its risks, could be the genuinely better choice. For those inclined toward subjective views of wellbeing, that a person values and desires participation is itself an important determinant of it being good for them. For those inclined to objective views, meaning and achievement are both plausibly important objective contributors to one's good, and participation in research can be both meaningful, and an achievement.

Unobjectionable Paternalism

So, when it comes to choices that affect me, I do not appreciate a distant committee or board ignoring my values and substituting their own. But there are also other kinds of case. I do sometimes appreciate them ignoring my factual appraisals and substituting their own.

Consider an example from John Stuart Mill (Mill, 2011/1859). He imagined seeing a person unknowingly walk out onto an unsafe bridge, and maintained, plausibly, that it is right to intervene forcefully to prevent that person from putting themselves in harm's way. Under sensible assumptions, it is not this person's life plan to die on a rickety bridge. Rather, they are most likely unaware of the danger and would both want you to intervene and thank you afterward. If anything, physically restricting them furthers rather than hinders their autonomy, as it better matches what happens to them (not drowning) to their desires for what happens to them (not wanting to drown).

Sometimes restrictions on this model are known as “soft” rather than “hard” paternalism (Dworkin, 2020). Less important than the terminology is the fact that a restriction does not always reduce autonomy, particularly when it is of the kind that would be welcome by those subject to them.

Consider the example of consumer safety measures for non-medical products. Automobiles, for instance, are subject to safety and performance requirements. We could imagine these being done away with. Instead, we could have a regime of information transparency. I could be provided with engineering specifications, blueprints, materials sourcing plans, test course results, and more. These could be presented to me in infographics, with teach-back opportunities, and according to general best practices. I could then decide which vehicular risks I was or was not willing to accept. Perhaps some car enthusiasts with odd taste and high levels of expertise would in fact prefer this.

I would not prefer this, as I am skeptical of whether I would be able to do a good job in evaluating the safety risks of my vehicle. Even if I could, given enough study and diligence, I do not want to. I suspect this is a more common psychological profile than that of the enthusiast.

Miller and Wertheimer (2007) have developed a view of research regulations along these lines: limits on allowable research risk benefit individuals who receive low-fuss protection from metaphorically walking out onto risky bridges (joining studies with high net risk), and while this may restrict a few idiosyncratic, high-health-literacy types, that's an unavoidable consequence of general rule-setting.

In my view, Miller and Wertheimer's general framing is compelling, but there is no reason to assume, as they do, that rules must be stated simply and at a high level of generality (“no more than this level of net risk”). Instead, we can adopt a more flexible approach (“no more than this level of net risk for a phase 3 efficacy trial, but phase 1 research can have higher risks if the following procedures are adopted…”) (Steel, 2020). Indeed, while we do not let manufacturers sell unsafe cars, we do let enthusiasts build, trade, and drive vehicles however they’d like so long as they do not take them onto public roads. This arrangement admirably assists those with a standard preference profile in buying cars that will suit them with a minimum of fuss while still allowing the enthusiasts their options.

A fully participant-centered approach would enable those hospice patients quoted earlier to participate in the research that makes sense for their particular situation and value scheme, including research which might be seriously risky without corresponding prospects for medical benefit, while still allowing a different (and more common) type of patient to enroll in, e.g., a large COVID vaccine trial while taking it on trust that the experimental intervention is reasonable.

Effects on Others

But there are also relevant considerations aside from what centers (or otherwise helps) current participants, including effects on future participants and the broader public. What is good for the former and latter may come apart. For instance, if an investigational drug is being tested in a trial using dose escalation, this may mean that a prospective participant enrolling at the start may need to accept a dose that is likely to be subtherapeutic, and they may rationally prefer off-trial access instead if given the choice. However, allowing off-trial access may undermine recruitment and prevent the timely generation of safety and efficacy data, prolonging clinical uncertainty and increasing overall social costs.

There is also the matter of paying. If patients are given access to unapproved drugs on the terms of their choosing, but must pay out of pocket, then, given prevailing prices, this may be a more symbolic than practical enhancement to their freedom. But if public or private insurance pays for these treatments, then those costs are passed on to the broader public, and this creates another kind of conflict. As Pearce correctly emphasizes, it may be in a particular very sick individual's interests to try anything that might conceivably help. But third parties may reasonably be unwilling to shoulder a considerable financial burden unless the evidence meets a stronger standard than that of conceivable benefit.

Returning to the Recommendations

How do these remarks relate to Pearce's specific recommendations? Those are: 1) tiered consent based on disease severity, 2) adaptive trial designs with patient governance, 3) mandatory open-source transparency, 4) post-market surveillance, and 5) independent safety monitoring with patient representation.

Recommendations 1 and 3 involve, at least in part, democratization through transparency in support of informed decision-making by individual potential participants. This approach is challenged by the suggestions that individuals may reasonably prefer to be governed by regulations that resolve technically demanding questions for them in ways that minimize difficulty and prevent mistakes. Pearce anticipates a version of this concern; he helpfully clarifies that the proposal is not for everyone to become a biostatistician, and he cites empirical evidence that alternative consent procedures can produce better understanding than traditional ones. All the same, showing a relatively better understanding than some other procedure is not equivalent to showing a level of understanding that is adequate in an absolute sense, nor does it show that participants actually want to take on significant and independent responsibility for judging this evidence.

Recommendations 2 and 5 involve, at least in part, democratization through increased patient, participant, and public governance. If all these parties had the same interests, then it would not matter much precisely who was included, and if only one party's interests mattered, it would be clear that they are the ones who need to be included. But, as mentioned, there are unfortunately distinct, opposed, and apparently reasonable interests at play. Given that different procedures for resolving them may thus yield very different results, it is important to understand who should be engaged and why, and there is unfortunately limited empirical evidence assessing outcomes of different engagement strategies and little normative clarity on what to prioritize (Rhodes & Ostertag, 2026; Steel, 2026). And a separate, further potential problem, one I take to be fully in keeping with the spirit of Pearce's paper, is that additional requirements for engagement run the risk of themselves becoming yet another bureaucratic hurdle extending research timelines and driving up costs.

Nothing in the forgoing casts any doubt on the general recommendation to approve new drugs faster and then use post-market surveillance; nor to use adaptive trial designs; nor to the general appeal to recalibrate the Type 1 and 2 error rates standardly tolerated in clinical research; nor to many other specific elements of Pearce's recommendations. Indeed, as mentioned at the outset, I am extremely sympathetic to the allegation that the current system over-optimizes the prevention of visible iatrogenic harms over the prevention of less visible but much, much larger harms from the status quo disease burden.

Pearce's paper points out, frustratingly, that arguments against this kind of error have existed for nearly half a century without translating into practical reform. Democratization may be a means to finally achieving change: it can be reasonable to respond to persistent failure by proposing that someone else (participants, the public) should be in charge. Against this, the present commentary also proposes some counter-reasons as to why it could be desirable to retain a significant role for review processes, ethics guidelines, and regulations that express an expert assessment of risks, benefits, and the balancing of distinct social interests, but while using different and better judgments as the basis of those regulations.

Admittedly, that it would be nice for regulatory judgments to be of a higher quality (and faster) is not a new idea and there is no obvious means to achieving it. The positive spin would be that perhaps someone will discover a non-obvious means.