The administration released a proposed rule, for comment ot July 13, on DEI anda other policy issues in the awarding of US government grants.
Home page here. PDF, 108 pages, here. 91 Fed Reg 32198, May 29, 2026.
Law firm discussion, Jenner Block, here. Inside Higher Ed here. Center for American Progress here.
###
Concise summary of the 108-page proposed rule
The May 29, 2026 Federal Register proposed rule is a government-wide rewrite of the 2 CFR federal financial assistance framework, covering OMB plus many grant-making agencies, including HHS, NSF, USDA, DOE, DoD, EPA, Education, VA, and others. Its stated purposes are to improve transparency, accountability, and oversight of grants; clarify that OMB’s 2 CFR text has binding regulatory status; and reduce recipient burden. Comments are due July 13, 2026.
The central operational change is political/appointee-level pre-issuance review. Agencies would still run merit review, but senior appointees must review all discretionary awards for consistency with law, agency priorities, the “national interest,” and—where applicable—the President’s policy priorities. The rule says this pre-issuance review may itself support a decision not to fund an application.
The rule also gives agencies broader discretion to terminate or suspend awards when agency priorities change, including mid-award. Jenner & Block’s summary highlights this as a standardized authority to terminate grants due to changed “agency priorities,” potentially including changes in political leadership.
Substantively, the rule tries to codify several Administration priorities: barring use of federal awards to fund, promote, encourage, subsidize, or facilitate unlawful DEI/DEIA practices, “gender ideology,” or gender transition of minors under 19; discouraging disparate-impact liability theories; adding foreign-affiliation and national-security risk factors; encouraging “Gold Standard Science”; preferring lower indirect cost rates “all else being equal”; de-emphasizing institutional prestige; and adding viewpoint-neutrality rules for certain federally funded events.
Importantly, the proposal does not directly cap negotiated indirect cost rates at 15%, despite earlier expectations. However, it does signal a preference for lower indirect rates in discretionary awards and appears to restrict some publication-cost reimbursement unless required by statute or pre-approved case by case.
In plain English: this is not merely a technical grants-management rule. It converts the Uniform Guidance into a much more explicit policy-control instrument. Peer review, scientific merit, and traditional program criteria would remain present, but a new political/policy compatibility filter would sit on top of them.
Effect on healthcare and life sciences grants
For NIH, CDC, AHRQ, FDA cooperative agreements, HRSA, SAMHSA, ASPR, VA research, NSF bio/biomedical programs, and HHS-adjacent public health grants, the rule would matter in at least five ways.
1. Peer review becomes less dispositive
The rule does not abolish peer review, but it makes it more clearly advisory. A proposal could be scientifically excellent and still fail at the appointee-level review if it is viewed as inconsistent with agency priorities, the national interest, Administration policy, or the proposed DEI/disparate-impact provisions. For NIH-style grantmaking, that is a major structural shift: the “payline plus council plus programmatic balance” model would now be overlaid with a more explicit political/policy screen.
For life sciences, this could affect topics such as health disparities, community health, gender medicine, HIV prevention, maternal mortality, environmental justice, firearm injury prevention, immigrant health, vaccine communication, reproductive health, and studies framed around structural racism or social determinants of health. The effect would not necessarily be a formal ban on all such research; rather, it would create a review environment in which language, aims, recruitment criteria, dissemination plans, and intervention design would be scrutinized for policy alignment.
2. The greatest uncertainty is the boundary between unlawful preference and valid biomedical fact
The rule’s most important ambiguity for healthcare is the one you flagged: the difference between using race/ethnicity/sex as a preference in employment or program selection and studying objective epidemiologic or biologic differences across populations.
A workplace DEI example is relatively easy: a grant-funded university program gives a promotion, fellowship, slot, stipend, or leadership role to a less-qualified candidate because of race, ethnicity, or sex, or uses race as a selection criterion. The proposed rule is clearly aimed at that kind of practice; it specifically objects to racial preferences or intentional proxies for race in employment or program participation.
But a biomedical fact pattern is different. Statements such as “Black men have higher prostate cancer incidence and mortality,” “Ashkenazi Jewish ancestry is associated with higher prevalence of certain BRCA variants,” “sickle cell disease is more common in people with sub-Saharan African ancestry,” or “maternal mortality differs by race/ethnicity” are not, by themselves, preferential treatment. They are descriptive epidemiology, population genetics, risk stratification, or public health surveillance. A rational healthcare system must be able to describe such facts, or it cannot target screening, prevention, trial enrollment, outreach, or quality improvement.
The danger is not that the text necessarily forbids every such statement. The danger is over-compliance and ideological flattening. A zealous reviewer, institutional lawyer, or program officer might treat any race-conscious language as suspect, even when the grant is not allocating benefits by race but measuring, explaining, or reducing a documented disease burden. That would be a serious category error: “race used as a hiring or participation preference” is not the same thing as “race/ethnicity measured as a covariate, risk marker, exposure proxy, or epidemiologic descriptor.”
3. Health disparities research becomes more linguistically and administratively fragile
The proposed rule’s rhetoric treats much prior DEI-related grant activity as wasteful, divisive, or unrelated to statutory public purposes; the preamble cites NSF, PEPFAR, and other examples as evidence that federal grants were used to promote ideological agendas rather than core program goals.
That framing will likely chill some healthcare research even where the underlying science is mainstream. Investigators may need to reframe proposals around disease burden, access, outcomes, mechanism, prevention, diagnostic accuracy, quality, safety, or cost-effectiveness rather than “equity” as a freestanding ideological term. For example:
“Improve equitable access to prostate cancer screening among Black men” may be reviewed differently than “Evaluate whether a risk-stratified screening intervention reduces late-stage prostate cancer diagnosis in a population with elevated incidence and mortality.”
The second sentence is harder to attack as DEI because it is anchored in disease epidemiology and measurable outcomes. Same substantive project; lower ideological surface area.
4. Public health infrastructure grants may be especially exposed
The CAP report argues that DEI-related rollbacks and grant terminations have already affected public health research, surveillance, maternal mortality monitoring, environmental monitoring, community health worker programs, and workforce pipelines. It warns that losing disparities research and targeted community-health infrastructure could weaken outbreak detection, emergency response, and prevention.
Even allowing for CAP’s progressive framing, the operational point is real: much public health work naturally involves population subgroups. Surveillance asks who is affected, where, and why. If agencies or grantees become afraid to name subgroups—Black men with prostate cancer, rural mothers with obstetric risk, tribal communities with diabetes, low-income children with lead exposure—the public health function degrades. You cannot manage what you cannot measure, and you cannot target an intervention if you are not allowed to identify the target population.
5. Institutions will change grant-writing behavior
Healthcare and life sciences institutions will likely respond with a new compliance style:
They will avoid using race, ethnicity, sex, gender identity, or socioeconomic status as eligibility preferences unless clearly authorized and legally reviewed.
They will preserve demographic variables where scientifically necessary, but justify them as epidemiologic, clinical, statistical, or health-services variables.
They will distinguish recruitment goals for scientific validity from quotas or preferences. For example, “ensure adequate representation to estimate subgroup effects” is stronger than “prioritize enrollment of historically marginalized groups” unless the latter is tightly tied to the scientific question.
They will document statutory and programmatic fit: cancer burden, maternal mortality, infectious disease control, diagnostic accuracy, workforce shortage, rural access, national preparedness.
They will scrutinize community advisory boards, stipends, training slots, pilot awards, and outreach programs to ensure that selection criteria are not framed as race-based preferences.
They will also prepare for more mid-award risk: a funded project that was acceptable at award could become vulnerable if priorities change, leadership changes, or terminology is later interpreted as noncompliant.
The key analytic distinction: “DEI as preference” versus “DEI-adjacent objective science”
The defensible line, in my view, is this:
Problematic under the proposed rule:
A federally funded program says, explicitly or functionally, “we will select, hire, promote, fund, admit, or advantage person X over person Y because of race, ethnicity, sex, or an intentional proxy for race.” That is the workplace/promotion concern: a more qualified white male, Asian applicant, or any other person is passed over because demographic balancing is treated as a selection objective.
Not inherently problematic:
A federally funded cancer epidemiology study says, “Black men have higher prostate cancer incidence and mortality; therefore, this study will measure risk, access, tumor biology, screening patterns, stage at diagnosis, treatment differences, and outcomes in Black men and comparison populations.” That is not a racial preference. It is a scientific statement about observed disease distribution and a research design built around a public health question.
Gray zone:
A grant says, “we will provide services only to members of group X,” or “we will select trainees from group X,” or “we will prioritize group X applicants,” even if motivated by a disease disparity. That may still be defensible in some contexts, but it is much more vulnerable unless the eligibility criterion is tightly connected to the statutory purpose, clinical risk, geography, disease status, or a valid research design rather than identity preference as such.
The practical drafting rule for life sciences grants should be: use demographic categories as variables, risk markers, stratification factors, recruitment-validity tools, or disease-burden descriptors—not as moralized allocation preferences. In cancer, cardiology, infectious disease, maternal health, pharmacogenomics, and public health surveillance, suppressing objective subgroup facts would be scientifically irresponsible. But institutions will need to be much more precise in explaining why those subgroup facts are necessary to answer the biomedical question.
Bottom line
For healthcare and life sciences, the proposed rule’s largest effect is not just “anti-DEI.” It is a shift from expert-driven grant selection toward policy-supervised grant selection, with DEI, disparate-impact theory, gender-related topics, foreign affiliations, indirect costs, institutional prestige, and “Gold Standard Science” all becoming explicit review dimensions. The immediate result will be caution, rewriting, and legal review. The longer-term risk is that valid population health science gets chilled because reviewers conflate identity-based preferences with objective disease epidemiology. The best defense is rigorous, neutral, disease-centered drafting: describe disparities as measurable facts, tie them to statutory health purposes, and avoid selection mechanisms that look like demographic preference rather than scientific necessity.
###
###
BIDIL review in NEJM, June 3, 2026. Here.
###
The NEJM piece is useful because BiDil is almost a perfect stress test for the question we were discussing: when is “race” a crude but sometimes pragmatically useful clinical descriptor, and when does it become a misleading biological fiction or a commercial/political instrument?
What the NEJM BiDil article says
The article is an NEJM “Intention to Treat” podcast transcript titled “BiDil — The Story of the Black Pill.” It asks how a fixed-dose combination of hydralazine plus isosorbide dinitrate came to be approved and marketed specifically for Black patients with heart failure.
The sympathetic version of the story is represented by Dr. Keith Ferdinand, the Tulane cardiologist. He describes a real, severe clinical problem: Black Americans have very high rates of cardiovascular disease, heart failure, stroke, hypertension, and cardiovascular mortality. In his account, the point was not to create a mystical “Black drug,” but to address a population of patients historically excluded from major cardiovascular trials and suffering excess morbidity and mortality. In the A-HeFT trial, conducted in self-identified Black patients, BiDil reportedly reduced mortality by 43%, which naturally made many Black cardiologists and advocacy groups view it as a potential tool for reducing a real health disparity.
The critical version of the story is represented especially by Dorothy Roberts and David Jones. They argue that BiDil converted a social and clinical category—self-identified Black patients in the United States—into a biological drug category. Roberts’ core objection is that race should not be used as a proxy for genetics, because racial categories are not biologic-genetic categories in the way the marketing implied. Jones adds the historical and regulatory oddity: the drug was originally tested largely in White patients, failed to win broad approval, then was rescued through a race-specific claim after a subgroup signal and a dedicated trial in self-identified Black patients.
The commercial angle is central. The transcript describes how the underlying components were old generic drugs. The “race-specific” claim helped create a new patent and market identity, but insurers could look at BiDil and say, in effect: why pay a premium for a branded fixed-dose combination when the same two drugs can be dispensed generically for far less? Sales disappointed, NitroMed faded, and no major company has apparently repeated the BiDil-style race-specific drug strategy.
The deeper lesson: BiDil was both reasonable and wrong
What makes BiDil so fascinating is that both sides had a point.
Ferdinand’s position is morally and clinically serious. Black patients were underrepresented in trials. Heart failure was and is a major burden. A trial in self-identified Black patients produced a striking result. It would have been perverse to say, “We refuse to study this because race is socially constructed.” Social construction does not mean social irrelevance. Race in America can correlate with neighborhood, access, diet, stress, environmental exposure, clinician bias, insurance, wealth, delayed diagnosis, and treatment patterns.
But Roberts is also right that a beneficial result in self-identified Black patients does not prove that “Black biology” caused the result. The category “Black” may have captured many things: ancestry, nitric oxide biology, environmental exposure, diet, renal physiology, comorbidity, severity of illness, background therapy, access to care, trial selection, or unmeasured social determinants. The failure was not studying Black patients; the failure was letting the drug become branded as though race itself were the mechanistic explanation.
The best sentence in the piece, conceptually, is Ferdinand’s later distinction: he would have been comfortable with a label saying the evidence was based on a study of self-identified African Americans, but he was uneasy with the implication that this was exclusively a “Black drug.” That is exactly the distinction that medicine needs more often.
Race as evidence descriptor versus race as mechanism
BiDil shows three different uses of race that are often conflated.
First, race as a trial-population descriptor: “This evidence comes from a trial of self-identified Black patients.” That is empirically true and useful.
Second, race as a clinical risk marker: “This population has higher observed heart failure burden and mortality.” Also true and clinically important, though it requires explanation.
Third, race as a biological mechanism: “This drug works because Black bodies are intrinsically different in a race-specific way.” That is the hazardous leap. It may sometimes be that ancestry-linked genetic variation matters for drug response, but then the answer should be ancestry, genotype, biomarker, renal physiology, enzyme activity, receptor biology, or another measurable variable—not folk-racial labeling.
The BiDil episode therefore argues not for color-blind medicine, but for better-variable medicine. Do not erase race from datasets. But do not stop at race. Collect socioeconomic status, geography, educational status, ancestry when relevant, environmental exposures, discrimination, access, insurance, comorbidities, genotype, biomarkers, and treatment history. Race can be a clue; it should rarely be the endpoint of explanation.
Connection to the Federal Register grants/DEI proposal
This is where BiDil links directly to the 108-page Federal Register proposal. The proposed grants rule is hostile to federal funding that promotes unlawful DEI preferences, disparate-impact theories, or race-based selection criteria, and it gives senior appointees a larger role in judging whether discretionary awards align with agency priorities and the national interest.
A superficial reader could take the BiDil story and say, “See? Race-based medicine is bad; therefore grants studying Black cardiovascular health, Black prostate cancer mortality, maternal mortality by race, or race-stratified outcomes are DEI and should be disfavored.” That would be the wrong lesson.
The better lesson is subtler:
A grant that says, “We will give opportunities, jobs, training slots, or program benefits preferentially by race” is in the danger zone under the proposed rule.
A grant that says, “Self-identified Black patients have higher heart failure mortality; we will study whether this reflects access, severity at diagnosis, comorbidities, environmental exposures, treatment patterns, biomarkers, or genotype” is not the same thing. That is not racial preference. It is scientific investigation of an observed clinical disparity.
BiDil is actually a warning to both camps. It warns progressive health-disparities researchers not to be sloppy with race, not to treat race as magic, and not to let advocacy or commercialization outrun mechanism. But it also warns anti-DEI regulators not to flatten all race-conscious biomedical work into “DEI.” If they do, they will suppress exactly the kind of research needed to move from crude racial categories toward better mechanistic variables.
Practical takeaway for grants
A life sciences grant after this proposed rule should probably avoid language that makes race sound like destiny or uses identity as a freestanding allocation principle. But it should absolutely be able to say:
“Prior studies show excess heart failure mortality in self-identified Black patients. This project will evaluate the clinical, social, environmental, treatment-access, and molecular contributors to that excess risk, with the goal of improving diagnosis and treatment for all patients.”
That is the post-BiDil sweet spot: race acknowledged, not reified; disparities studied, not moralized; biology pursued, not presumed.
###
###
The Rule: "Institutional Prestige" Comments
The proposal says that less weight should be given to "institutional prestige." But where is there explicit weight? Grants aren't rated as 1 to 7 score for "prestige."
#
Yes — that is exactly right. The proposal seems to be attacking a real-world phenomenon, but not one that is usually explicit in the NIH scoring rubric.
The proposed text says that, “to the extent institutional affiliation is considered,” agencies should prioritize rigorous, reproducible scholarship over “historical reputation or perceived prestige.” It also says senior appointees must conduct pre-issuance review and that peer review remains advisory rather than binding. Jenner & Block reasonably paraphrases this as an “anti-prestige” provision, but the actual regulatory wording is conditional: if institutional affiliation is considered, prestige should not dominate.
For NIH, you are correct that reviewers do not normally score “Harvardness” or “Stanfordness” as a formal criterion. The traditional NIH scored criteria are things like significance, investigator, innovation, approach, and environment. “Environment” is the closest fit, but it is not supposed to be a prestige score. It asks whether the scientific environment, institutional support, equipment, patient population, cores, collaborators, and resources are adequate for the proposed work. A famous institution may help indirectly because it often has extraordinary cores, biostatistics, trial infrastructure, mouse facilities, sequencing, informatics, or senior collaborators. But that is not the same as saying, “This is Johns Hopkins, therefore +1.”
So the proposal is probably responding to implicit prestige effects, not a formal NIH box-checking practice. Those effects can appear in several subtle ways:
A reviewer sees a famous institution and assumes competence, feasibility, compliance capacity, or access to resources.
A famous PI or institution gets the benefit of the doubt on a risky approach.
A less famous institution may have to prove feasibility more explicitly.
A grant from an elite center may look “safer” to program staff when paylines are tight.
Big institutions may have professional grant-writing infrastructure that makes applications cleaner, more preliminary-data-rich, and more responsive to reviewer expectations.
That is sociology of science, not an explicit scoring rule.
The complication is that “prestige” and “capacity” are hard to disentangle. If a cancer center has a specialized biobank, a 50-person clinical trials office, CLIA sequencing, an NCI-designated infrastructure, and access to 2,000 relevant patients per year, those are legitimate environment factors. Calling that “prestige bias” would be wrong. On the other hand, if reviewers assume rigor because the institution is famous, without examining reproducibility, data quality, trial operations, prior performance, or actual resources, then the proposed rule has a fair target.
The best critique of the proposal is therefore: it identifies a plausible implicit bias but overstates the degree to which federal grant review formally privileges prestige. NIH peer review already has a legitimate “environment” criterion, and the answer should not be to pretend institutions are interchangeable. The right question is whether the institution has the specific capabilities needed for the project, not whether it has a famous name.
A clean comment might say something like:
OMB should distinguish between impermissible reliance on institutional reputation and permissible evaluation of institutional capacity. In biomedical research, institutional environment is often directly relevant to feasibility, patient access, biospecimen availability, specialized equipment, regulatory infrastructure, data security, animal facilities, biostatistics, and clinical trial execution. Reviewers should not reward historical prestige as such, but they must be able to consider concrete institutional resources that materially affect whether the proposed work can be completed rigorously and safely.
That distinction would make the anti-prestige idea sensible rather than performative.
