This essay offers a critique of the Trump administration’s effort to ground its hostility toward gender-affirming care in science. The aim of this critique will be simple: to show precisely how and why the administration’s effort has failed. To most readers, the fact of such failure will hardly be surprising. To some, it may even seem pointless or counterproductive to devote critical attention to what is, after all, a post-hoc and bad-faith attempt at justification. Plainly, the administration’s desire to abolish gender-affirming care (and indeed, trans existence) was present before any scientific reason was sought for it, and it will remain even if the now-alleged scientific basis for it crumbles. Nonetheless, there are two reasons for engaging in detailed critique of the administration’s pursuit of a scientific justification. One is that existing criticisms of this search, offered by mainstream institutions such as the American Medical Association, have systematically underestimated the extent to which the administration’s search for scientific grounding has failed. Inadvertently, these criticisms have presented the administration as having come much closer to finding scientific grounds for its attacks on gender medicine than is the case; and this false presentation demands correction. A second reason for sustained critique is that the administration’s failure to ground its hostility toward gender-affirming care in science carries a broader lesson. Close consideration of this failure suggests that its root cause was not a specific mistake made by the administration in its pursuit of scientific justification, but the general inability of scientific methods to fully secure ethical judgments about medicine. This inability is crucial to reckon with, both in order to criticize the administration’s attacks on gender-affirming care most effectively, and in order to strengthen the public rhetoric in support of gender-affirming care. I will return to this last point briefly at the close of the essay; let us now consider, first, the administration’s yearlong pursuit of science, its alleged success, and the reasons for its manifest and total failure.

Throughout 2025, the Trump administration sought a scientific warrant for its crackdown on gender-affirming care. It wagered on the existence of objective proof that its attacks on gender-affirming care were legitimate, and set out to find it. In January 2025, Trump inaugurated the search with an executive order, commissioning a review of all scientific literatures that might contain the sought-after proof.[1] The Department of Health and Human Services (HHS) was tasked with assembling an expert committee to undertake the review, and with publishing their results within 90 days. The committee’s interim report duly appeared in just over three months, in early May, and its final report was published, following revisions, in late November. At a press conference on December 18, HHS Secretary Robert F. Kennedy Jr. summarized the report’s findings. They were, he announced, an unmitigated success. The committee had discovered “an overwhelming body of evidence that [gender-affirming] procedures hurt, not help, children,” and proved that “so-called gender-affirming care … is not medicine. It is malpractice.”[2] So the proof had apparently been found. What precisely did it consist in?

In a written declaration published the same day as the press conference, Kennedy went into more detail. According to this text, what the committee had produced, specifically, was evidence of an “unfavorable risk/benefit profile” for gender-affirming procedures.[3] That is, they had reviewed the relevant scientific literatures and found ample evidence of “harms” that can result from procedures such as mastectomy and puberty suppression. At the same time, they had found minimal or uncertain evidence that the intended benefits of these same procedures actually consistently result from them. Thus, they concluded that the probable harms of gender-affirming procedures “exceed” or “outweigh” their probable benefits; this is what it means for gender-affirming care to have an “unfavorable risk/benefit profile.” In lay terms, the committee had determined that gender-affirming procedures are more likely to yield harms than benefits. Such is the evidence, and the conclusion based on it, that Kennedy interprets as scientific proof that gender-affirming care is “not medicine” but “malpractice.”

It is true, as Kennedy and the expert committee insist, that risk/benefit assessments are one common way of deciding whether a treatment is acceptable in modern medicine. It is not surprising, then, that most critics of the HHS committee’s work have not challenged their approach but only contested their reading of the available evidence. For instance, in their joint response to the report, the American Medical Association (AMA) and the American Academy of Pediatrics (AAP) reject its characterization of gender-affirming care as malpractice on the grounds that the report “misrepresent[s] the consensus of medical science” and presents “selective or politically motivated interpretations of data that ignore the totality of research and clinical outcomes.”[4] In other words, their criticism is that the report miscalculates the risk/benefit profile of gender-affirming procedures, by simply ignoring much of the evidence that shows these procedures’ benefits.[5] That may well be true. But this criticism leaves a more basic premise of the report unquestioned, and even reinforces it. The premise is that risk/benefit analysis, if properly conducted, is not just a common but an objective, scientific method for determining whether a medical procedure is benign or dangerous. And this is a consequential premise to leave intact, for it is what sustains the administration’s fantasy—or at least, its public insistence—that its attacks on gender-affirming care could be scientifically warranted.

But a moment’s reflection engenders some doubt about this premise. The method is apparently an objective weighing of known risks and benefits. But before the weighing takes place, who decides which outcomes are risks and which are benefits? How is the classification made if a given outcome is neither an absolute benefit nor an absolute harm, but a complex phenomenon, or ambiguous? Are any relevant outcomes not measurable and thus not accounted for at all? And what exactly does it mean to “weigh” outcomes that can be radically different in kind? What common measure could there be between, say, loss of bone density, on the one hand, and reduction of anxiety, on the other? If there is no common measure, how can these be compared objectively or scientifically?

These are not a skeptic’s questions. They are concerns that were first raised, and rigorously formulated, by the inventors of medical risk/benefit analysis themselves. They are, as well, indications of genuine limitations of the method, limitations that led its original proponents to present it precisely not as a source of scientific certainty but as a useful yet inevitably uncertain and fallible mode of judgment.

Consider here two classic texts on medical risk/benefit analysis. (The HHS report itself cites both as authorities on the topic.) The first is the Belmont report. Published in 1979 by a federal medical-ethics commission, this text set the terms of risk/benefit analysis as a method for medical decision-making. In particular, it develops and recommends this method as a tool for determining whether a given treatment is “beneficent,” or benign, and thus a legitimate medical procedure.[6] What is important for our purposes is the caveat that immediately follows the recommendation.

This caveat concerns the apparent objectivity of the method. “It is commonly said,” the authors note, “that benefits and risks must be ‘balanced’ and ‘in a favorable ratio.’”[7] Such terms suggest an objective calculation. But, the authors continue, “The metaphorical character of these terms draws attention to the difficulty of making precise judgments.” In fact, the authors understate the difficulty here. It is not just that the weighing, in risk/benefit analysis, is less precise in practice than we would expect. It is that literally speaking, risk/benefit analysis does not involve “weighing” at all; “weighing” (or “balancing”) is only a metaphor for a different, more uncertain form of judgment, for which the authors have no name. Whatever this form of judgment is, it is not objective calculation. And it cannot be, because although risk/benefit analysis is presented in quantitative language, medical risks and benefits are not objectively determinable quantities. The Belmont report makes clear that the relevant harms and benefits are not just measurable physiological states: “Many kinds of possible harms and benefits need to be taken into account. There are, for example, risks of psychological harm, physical harm, legal harm, social harm and economic harm and the corresponding benefits.” Plainly, many of these relevant harms and benefits will not be measurable at all, let alone objectively so. Hence, they will never, literally speaking, be objectively comparable. “Weighing” in risk/benefit analysis is thus always a partly non-objective method that only aspires to full objectivity. The authors write: “However, the idea of systematic, nonarbitrary analysis of risks and benefits should be emulated insofar as possible.” Objectivity is what is emulated, never achieved, by medical risk/benefit analysis; it is a regulative but unreachable ideal.

The gap between medical risk/benefit analysis and a fully objective method becomes clearer in the discussion of this topic in Tom L. Beauchamp and James F. Childress’s Principles of Biomedical Ethics, widely considered a founding text of medical ethics. Beauchamp and Childress, too, forthrightly acknowledge that “a purely quantitative ideal” is not appropriate for medical risk/benefit analysis, and that we speak “metaphorically” if we say that “risks and benefits can … be ‘weighed’ and ‘balanced’ and shown to be ‘in a favorable ratio.’”[8]

Beauchamp and Childress go further and draw two practical consequences from this. The first is that medical risk/benefit analysis can devolve, if one is not careful, into what they call “intuitive weighing.”[9] The point is that since risks and benefits can never be objectively weighed, like literal quantities, the analyst who compares risks and benefits must always rely on some subjective standards in determining which risks carry the most “weight” or significance, whether they are “greater” or more concerning than the promised benefits, and so forth. These could be moral standards, or cultural standards, or the standards of a professional community. Ideally, the standards chosen will be broadly acceptable, but there is a danger that they will be merely “intuitive,” that is, reducible to the arbitrary, private preferences of the analyst. This is a danger that is always present in medical risk/benefit analysis, simply because it can never be a fully objective calculation. And if medical risk/benefit analysis does become mere intuitive weighing, then clearly it will have no value as a form of judgment, or no more value than a mere opinion. The danger can be mitigated, though, if the analyst is transparent about the non-objective standards that inform the weighing of risks and benefits—in Beauchamp and Childress’s words, if the analyst is “explicit in stating operative standards and principles.”[10]

One upshot of the non-objectivity of risk/benefit analysis is thus that it is always in danger of becoming merely subjective. A second is that in some contexts it cannot be used. Just after their warning about intuitive weighing, Beauchamp and Childress write: “Unfortunately, in some cases the risks and benefits may be fairly well known, and yet one may be incapable of determining whether the benefits outweigh the risks.” They do not exactly define this set of cases in which risk/benefit analysis is impossible but instead illustrate the point with an example. They ask us to imagine a baby born prematurely, who requires “intense resuscitation measures.” The procedure is begun, but after 14 minutes the baby has not responded. The ethical question now is whether the procedure remains “beneficent,” and thus whether it should continue. The problem is that, at this point, even if the treatment succeeds, there is a “high risk that a retarded [sic.] baby (at best) would result from further medical efforts.” On the other hand, the treatment promises the survival of a wanted child. In this constructed example, the likelihoods of cognitive disability and survival are also known. “But how one actually weighs the high risk of mental retardation [sic.] and the probable disappointment of the parents against the slender possibility of a healthy baby and the parents’ known desire to have the child is a matter clearly subject to differences of opinion, even if the probability of harm and the probability of benefit were well established and agreed upon by all.” The authors conclude that in such a case, risk/benefit analysis cannot meaningfully take place: in the next line, they write, “Risk/benefit assessments thus will not function as a panacea for decision-makers.”[11]

Here, Beauchamp and Childress set a hard limit on the utility of medical risk/benefit analysis. What exactly makes “weighing” unworkable in the premature baby example? It is that there are strong and unresolvable differences of opinion about how to value the various outcomes associated with the procedure. What is the “weight” of a so-called “harm” such as cognitive disability? Who decides whether the potential cognitive disability is a “harm” at all? (In the example, the analyst cannot be sure that the parents would be “disappointed” by that outcome; the reaction is cast as only “probable,” not assured.) What is the “weight” of an awaited child’s survival? The values at issue are so contested, their determination so dependent on differences that are plainly differences of opinion, and on standards that are plainly contentious subjective standards, that even the pretense (the “emulation”) of objective calculation cannot be sustained. Perhaps in a different social context, or at another historical time, these values will be less controversial, and the standards used to determine them will be less disputed, and then something more like an objective risk/benefit analysis can take place. But to the extent that the values involved are controversial, and are the subject of unresolvable differences of opinion, they can only be arbitrarily (subjectively) determined by the risk/benefit analyst, and not—even metaphorically speaking—scientifically measured and weighed.

It seems to me that, by Beauchamp and Childress’s criteria, gender-affirming care is precisely a case beyond the reach of risk/benefit analysis, at least at this social-historical moment. It is not a procedure whose outcomes can, at present, be even apparently objectively weighed. To illustrate this point, let me offer a second hypothetical scenario, built on the model of Beauchamp and Childress’s. (To be clear, the point of this scenario will be only to show that risk/benefit analysis breaks down in an evaluation of gender-affirming care, just as it breaks down in the premature baby scenario. It is not meant to realistically depict what a deliberation among the parties I gather in it would look like.) Suppose, for a moment, that Kennedy, the HHS committee, and representatives from the AMA and the AAP are in conversation. They have gathered to review the science about, and then to judge the acceptability of, a particular gender-affirming procedure such as puberty suppression. Imagine that they all read the scientific literature and somehow manage to agree on the following: Treatment outcome studies clearly show, on the one hand, that the use of hormone blockers to delay puberty causes infertility later in life for most patients who take them; the studies also clearly show, on the other hand, that in many but not all cases, puberty suppression substantially improves the mental health of those same patients, by restoring or newly supporting their sense of bodily integrity. Now, even if the participants were to agree on these facts about risks and benefits, as soon as they began to compare the risks and benefits, and to judge whether the treatment is more beneficial than it is harmful or risky, any attempt at an objective “calculation” of risks and benefits would break down, or appear invalid. To paraphrase Beauchamp and Childress, at this point in the scenario: how one actually weighs the risk of infertility and the “probable disappointment” of the patient against the possibility of psychological improvement and the patient’s known desire for a better supported sense of somatic integrity would be a matter clearly subject to differences of opinion, even if the probability of harm and the probability of benefit were well established and agreed upon by all. The “weights” of the outcomes involved are simply too contested, and the standards that one would use to decide them too actively disputed as well, for anything even resembling a quantitative, scientific weighing of risks and benefits to take place. There could only be, at best, an attempt at “intuitive weighing,” and the judgment that resulted from it would have merely the weight of a private opinion.

There is a narrower and a broader lesson to be drawn from all this. The narrower lesson is that the Trump administration has doubly failed in its search for a scientific warrant for its crackdown on gender-affirming care. Not only has it failed to produce a convincing warrant. It has not even found a method that could, in principle, produce such a warrant. (This second failure is obscured by point-by-point rejoinders to the HHS report’s calculation, although this is not to say that such rejoinders are not otherwise valuable.) For even under the best circumstances, a risk/benefit analysis never actually delivers a scientific proof but only emulates one; and in a case like gender-affirming care, it cannot even perform a convincing emulation. Having seen this, we can say that what the Trump administration has managed to accomplish, in its search for a scientific warrant, is not even a weaponization of science for its attacks on gender-affirming care. Rather, it has weaponized what is, on its own terms, a non-scientific method (when applied to a case like gender-affirming care) as the warrant for these attacks. It seems as if the administration’s commissioned report has actually proved the opposite of what it was intended to demonstrate: that the basis of the administration’s crusade against gender-affirming care is precisely not science, but mere subjective judgment—only opinion, intuition, and animus.

The broader lesson to be drawn from this consideration of the HHS report is not about the Trump administration, but about the possible role of science in our debates about gender-affirming care. This lesson is less a strict factual inference than a speculation. I think that the limitations of risk/benefit analysis considered above are likely to be general limitations—that is, the limitations of any method that purports to be a scientific (or near-scientific) approach to producing ethical judgments. Any such method, I think, will be enabled at crucial points by contestable subjective standards (moral, cultural, professional), and hence will be unworkable, or at least will lose its veneer of scientificity, whenever the subjective standards that enable it are the subject of heightened attention and violent controversy. If this is right, then a disappointing conclusion follows: Quasi-scientific approaches to ethics will be the least helpful in precisely those hard cases where one might want their help the most, namely, cases in which one must produce, or publicly defend, an ethical judgment concerning a medical procedure, but in a context of extremely strong disagreement about the values and standards that should inform the ethical judgment (such as about the goodness or badness, the propriety or impropriety, of gender-affirming care). A desire for the help of science would make sense, but it may be that science simply cannot help us, in precisely these hardest cases. (Science is thus no panacea for decision-makers, as Beauchamp and Childress might say.) There would then be a difficult but unavoidable challenge, not just for the Trump administration but for all of us: namely, to find another way of defending our ethical judgments about gender-affirming care besides, and more effective than, the tempting but futile way of emulating science.

 

[1] “Executive Order 14187 of January 28, 2025: Protecting Children from Chemical and Surgical Mutilation,” Federal Register, Feb. 3, 2025, https://www.federalregister.gov/documents/2025/02/03/2025-02194/protecting-children-from-chemical-and-surgical-mutilation.

[2] “Health and Human Services Secretary Kennedy News Conference on Child Health Care Policy,” Press Conference held on December 18, 2025, video recording available at https://www.c-span.org/program/news-conference/heath-and-human-services-secretary-kennedy-news-conference-on-child-health-care-policy/670663.

[3] U.S. Department of Health and Human Services, “Declaration of the Secretary of the Department of Health and Human Services, Re: Safety, Effectiveness, and Professional Standards of Care for Sex-Rejecting Procedures on Children and Adolescents,” Dec. 18, 2025. This declaration, released in December 2025, was vacated by a federal judge in April 2026. Although the government claimed that the declaration was merely a statement of opinion, which would have been legal, the judge ruled that it was an official agency action that attempted to override individual states’ authority to set medical standards without following proper legal procedures. Thus, the declaration has since been taken down from the HHS website. Yet this declaration still represents the HHS Department’s interpretation of the report that it commissioned, and it remains the best available account of what the Trump administration takes to be the scientific warrant for its attacks on gender-affirming care. An archived version of the declaration is available at: https://web.archive.org/web/20260422230725/https://www.hhs.gov/sites/default/files/declaration-pediatric-sex-rejecting-procedures.pdf.

[4] David Aizuss and Susan J. Kressly, “AMA and AAP Joint Statement on Evidence-Based Health Care,” American Academy of Pediatrics, Nov. 19, 2025, https://www.aap.org/en/news-room/news-releases/aap/2025/ama-and-aap-joint-statement-on-evidence-based-health-care.

[5] The report justifies its neglect of these studies by characterizing them as “low-quality evidence,” and thus as unworthy of inclusion in its literature reviews. The critics, of course, contest this characterization. The AMA and AAP text is a brief position statement, and does not offer a detailed critique of the report’s handling of evidence; for a fuller critique that arrives at the same conclusion as the AMA/AAP statement, see this discussion of the HHS report by a group of prominent pediatricians and bioethicists: Ian D. Wolfe et al., “Proposed Ban on Medicaid Funding for Hospitals Providing Gender-Affirming Care for Minors is Deeply Unethical,” STAT News, Jan. 30, 2026, https://www.statnews.com/2026/01/30/gender-affirming-care-why-experts-oppose-proposed-ban/.

[6] In fact, the Belmont report is about experimental treatments, offered as part of medical research, in particular. But its principles have been generalized to deal with the ethics of medical treatments in general, including by the seminal work by Beauchamp and Childress cited below, and the report itself continues to be an important reference for the field of medical ethics as a whole.

[7] National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research, The Belmont Report: Ethical Principles and Guidelines for the Protection of Human Subjects of Research (U.S. Department of Health and Human Services, 1979), https://www.hhs.gov/ohrp/regulations-and-policy/belmont-report/read-the-belmont-report/index.html.

[8] Tom L. Beauchamp and James F. Childress, Principles of Biomedical Ethics (Oxford University Press, 1979), 150, 148.

[9] Beauchamp and Childress, Principles of Biomedical Ethics, 149.

[10] Beauchamp and Childress, Principles of Biomedical Ethics, 150.

[11] Beauchamp and Childress, Principles of Biomedical Ethics, 151.