Antiracist Medicine in Colorblind Courts

Govind Persad* November, 2024

This Article considers how health professionals’ efforts to combat racial health inequities interact with legal restrictions constraining their ability to consider race. In light of the Roberts Court’s recent invalidation of two university admissions programs, intensifying a “colorblind” judicial shift, the collision between antiracist medicine and colorblind law is a pressing concern. This Article anticipates the implications of this collision and explores how health professionals and systems can design programs that survive judicial examination.

In Part I, the Article examines the frameworks that will apply if antiracist medicine faces legal challenges. These include the Equal Protection Clause and federal statutes like Title VI and Section 1981. Part II systematically outlines three reasons why medicine might consider race—the enhancement of individual and population health through beneficent consideration, the mitigation of racial health disparities via egalitarian consideration, and the redress of historical injustices through reparative consideration. Additionally, it identifies five central contexts in which race has been considered: professional training, the professional-patient encounter, allocation of scarce resources, public health, and clinical research.

Part III delves into the critical question of how medicine should respond to courts’ colorblind turn. Because policies classifying patients or professionals by race will be subjected to strict scrutiny, their viability becomes a challenge. In contrast, policies aimed at antiracist goals while avoiding race-based classification will normally pass muster. It also discusses recent legislative, regulatory, and litigation developments questioning the legitimacy of antiracist medical objectives. Understanding the evolving legal landscape is crucial for medicine to effectively address racial health inequities, and this Article seeks to provide precisely that understanding.

Introduction

In August 2022, Ky’reelle Riley died after a severe asthma attack.¹ He was 13.

Ky’reelle’s devastating death exemplifies a tragic health disparity. Black children in the United States are almost eight times as likely as their white counterparts to die of asthma.² This inequity has multiple causes: Housing disparities disproportionately expose Black children to asthma triggers like mold.³ Environmental racism exposes them to particulate pollution from trucks and factories.⁴ The effects of “toxic stress, racism and discrimination” make them likelier to be born preterm or at low birth weight.⁵

Health systems’ responses to asthma are often also inadequate: Ky’reelle’s parents reported that “doctors ‘brushed off’ ” their “concerns that there might be something wrong.”⁶ Although they tried to meet with an asthma specialist for Ky’reelle, his parents were informed that “the earliest available appointment was in October”—two months after he died.⁷ Clinical research that could help develop treatments for asthma often fails to include racial and ethnic minority participants, despite the disproportionate burden of the disease.⁸

Health professionals and systems increasingly strive to counteract the structural racism that underpins these health disparities.⁹ Some have taken action to combat environmental racism and redlining; others have attempted to recruit health professionals from underrepresented groups¹⁰ or match those professionals with same-race patients. Yet others have attempted to improve representation in clinical trials.

Health professionals and systems understandably may not have law top of mind when taking action against racism—their priority is saving lives, not satisfying judges. But current legal precedent strictly limits the ability of governments, recipients of government funding, employers, and even private individuals to make decisions on the basis of race, including to mitigate racial injustice. The Supreme Court’s recent invalidation of two university admissions programs exemplifies this “colorblind” judicial turn.¹¹ Current precedent also constrains, though much less strictly, the pursuit of race-conscious goals like reducing racial health disparities even when done without racial classifications.

This Article anticipates what will happen when antiracist medicine collides with colorblind law and considers how health professionals and systems can design programs that pass legal muster. Given the complexity of doctrine in this area, the Article focuses on identifying what will pass muster under present precedent or plausible evolutions of that precedent.¹²

Part I describes the law that will be most relevant if antiracist medicine is challenged in court. Section I.A examines decisions interpreting the Fourteenth Amendment’s Equal Protection Clause. Section I.B examines federal statutes.

Part II identifies three reasons why medicine might take race into account: to improve individual and population health (beneficent consideration of race), to reduce racial health disparities (egalitarian consideration), and to redress historical injustice (reparative consideration). Sections II.A-II.E examine five areas within health systems where classifying patients or health professionals by race has been advocated as a means to these goals: professional training, the professional-patient encounter, scarce resource allocation, public health, and clinical research.

Part III discusses how medicine might respond to courts’ colorblind turn. Policies that classify patients or professionals by race will have to satisfy strict scrutiny, which will be difficult. In contrast, policies that pursue antiracist goals without classifying patients or providers by race will pass muster more easily.

In previous work, I have examined the legality of considering race in the allocation of scarce medical interventions.¹³ Other scholars have discussed specific issues, such as the legality of seeking same-race physicians.¹⁴ In this Article, by contrast, I systematically examine how the Supreme Court’s approach to race and equal protection constrains efforts to address racial health disparities.

I. Colorblind Courts

The legal presumption against “race-based state action . . . cannot be overridden except in the most extraordinary case”¹⁵—so said the Supreme Court in the summer of 2023. Applying this reasoning, six justices invalidated two university admissions processes that considered applicants’ race. The Court’s majority prominently quoted Justice Harlan’s Plessy v. Ferguson dissent: “Our Constitution is color-blind, and neither knows nor tolerates classes among citizens.”¹⁶ And, in rebutting dissenters, the majority reaffirmed its endorsement of colorblindness, concluding that “what one dissent denigrates as ‘rhetorical flourishes about colorblindness’ . . . are in fact . . . defining statements of law.”¹⁷

The Court’s explicit endorsement of colorblindness continues a trend that began in the 1980s: an increasing rejection of efforts to address racial inequities by race-based means. First, the Supreme Court overturned prior precedent and applied strict legal scrutiny to the use of racial classifications in public contracting.¹⁸ Later, it did the same for secondary school assignments.¹⁹ SFFA continues this pattern, invalidating previously permitted uses of racial classifications in higher education.²⁰

This next Section reviews and summarizes the law that currently applies to policies—including health policies—that consider race, either as a means or as an objective. The Section focuses on federal law. Although some states go beyond federal law and set additional limits on the consideration of race, these limits apply only to public employment, education, or contracting,²¹ and so may not apply to most health-related decisions.

A. Federal Constitutional Law

1. Individual Racial Classifications

In Adarand Constructors v. Pena, the Supreme Court held that “any person, of whatever race, has the right to demand that any governmental actor subject to the Constitution justify any racial classification subjecting that person to unequal treatment under the strictest judicial scrutiny.”²² Satisfying strict scrutiny requires that the racial classification further a compelling interest and be narrowly tailored to serve that interest:

Any exception to the Constitution’s demand for equal protection must survive a daunting two-step examination known in our cases as “strict scrutiny.” Under that standard we ask, first, whether the racial classification is used to “further compelling governmental interests.” Second, if so, we ask whether the government’s use of race is “narrowly tailored”—meaning “necessary”—to achieve that interest.²³

a. Compelling Interests

Because racial classifications must satisfy strict judicial scrutiny, a compelling interest is required for their use. According to the SFFA majority, the Supreme Court has explicitly recognized only three government interests as sufficiently compelling to justify the use of a racial classification:²⁴ achieving racial diversity in higher education;²⁵ “remediating specific, identified instances of past discrimination that violated the Constitution or a statute;”²⁶ and “avoiding imminent and serious risks to human safety in prisons, such as a race riot.”²⁷

Some believe SFFA rejected diversity as a compelling interest, but the more plausible reading is that the SFFA defendants failed to provide measurable evidence that their policies realized diversity’s educational benefits.²⁸ SFFA clearly did not overrule Fisher II, which recognized the need for a “critical mass” of minority students as a compelling diversity interest.²⁹

Lower courts have found additional interests sufficiently compelling to justify racial classifications. Such interests include racial diversity on a police force,³⁰ compliance with a consent decree,³¹ and crime prevention.³² Better health has also been described as a sufficiently compelling interest by both the Supreme Court and lower courts, albeit in dicta.³³

In contrast, other interests cannot justify racial classifications. The Court has “repeatedly held that ameliorating societal discrimination does not constitute a compelling interest that justifies race-based state action.”³⁴ Lower courts have found that racial diversity in workplaces, other than police departments, is not a compelling interest.³⁵ Many governmental interests, however, remain in a grey area: The Supreme Court has neither endorsed nor rejected them as compelling, and lower courts have avoided opining on them for reasons of constitutional avoidance, the doctrine that counsels courts to avoid unnecessarily deciding constitutional questions.³⁶

b. Narrow Tailoring

Even if the government identifies a compelling interest supporting its use of a racial classification, it must also show that its use of that classification is narrowly tailored to realize the interest. Lower courts have adopted different approaches to assess narrow tailoring. Sometimes, courts examine narrow tailoring only after a compelling interest is shown.³⁷ At other times, courts consider narrow tailoring first to avoid deciding whether a governmental interest is compelling.³⁸ Courts may determine that a policy not only fails to advance a compelling interest but also fails narrow tailoring.³⁹

The Supreme Court has not precisely defined what narrow tailoring requires. But its precedents reveal two primary requirements that apply across contexts. First, race-neutral alternatives must have been found unworkable after serious, good-faith consideration.⁴⁰ “If ‘a nonracial approach could promote the substantial interest about as well and at tolerable administrative expense,’ ” then it must be chosen in preference to a racial classification.⁴¹

Second, the racial classification must precisely fit the compelling interest. It may not be overinclusive or underinclusive because “[r]acial classifications are simply too pernicious to permit any but the most exact connection between justification and classification.”⁴² The lumping together of multiple racial groups typically constitutes an unacceptably imperfect fit.⁴³

Two additional factors are also frequently discussed: whether the classification is time limited and whether it excessively burdens third parties. SFFA emphasized the need for a time limit,⁴⁴ as have government contracting cases.⁴⁵ Burden on third parties was similarly identified in both college admissions and contracting cases.⁴⁶ In fact, SFFA’s assertion that under “the Equal Protection Clause . . . race may never be used as a ‘negative’ ” can be read to prohibit any burden on third parties.⁴⁷

Other narrow tailoring factors appear either less important or confined to specific contexts. A program’s flexibility in using a racial classification counts in its favor,⁴⁸ but SFFA did not attempt to rebut the dissent’s references to the invalidated program’s flexibility.⁴⁹ This suggests that while inflexibility may doom a program, flexibility will not save it if it falters in other respects. Meanwhile, the “relationship of the numerical goals” of an affirmative action program “to the relevant labor market” is sometimes discussed but appears confined to the context of workplace affirmative action.⁵⁰

2. Intended or Foreseen Group-Level Racial Effects

Since Adarand, the Court has consistently subjected individual racial classifications to strict scrutiny.⁵¹ But decisionmaking often considers racial impact at a population level—rather than an individual level—without classifying anyone by their race. For instance, when comparing two policies, neither of which classifies anyone by race, decisionmakers sometimes select one over the other because of its predicted effects on intergroup racial disparities.⁵² At other times, decisionmakers may foresee that a policy preferable for other reasons will predictably benefit or burden people of some races more than others.

In Personnel Administrator v. Feeney, the Court established that policies’ population-level effects on protected groups are only legally suspect when those effects are both intended and adverse.⁵³ A discriminatory purpose exists when a “decisionmaker . . . selected or reaffirmed a particular course of action at least in part ‘because of,’ not merely ‘in spite of,’ its adverse effects upon an identifiable group.”⁵⁴ In contrast, merely foreseeing adverse effects on an identifiable group does not render a course of action per se discriminatory. However, “when the adverse consequences of a law upon an identifiable group” are so likely as to be “inevitable,” “a strong inference that the adverse effects were desired can reasonably be drawn.”⁵⁵ This inference, though, “is a working tool, not a synonym for proof”: If the adverse impact is inevitable but “an unavoidable consequence of a legislative policy that has in itself always been deemed to be legitimate,” that adverse impact does not render the policy discriminatory.⁵⁶ Feeney’s test recalled the classic philosophical distinction between intent and foresight.⁵⁷ Just as the ethics of war evaluates killing civilians as a side effect of bombing a military base differently from intentionally bombing a civilian site, the Feeney test judges a foreseeable and inevitable adverse impact differently from an intended one.⁵⁸

Notably, Feeney did not restrict policies with intended favorable effects.⁵⁹ Seeking to benefit an identifiable group is not discriminatory. The Third Circuit, however, has contratextually read Feeney to also encompass “beneficial or adverse effects ‘upon an identifiable group.’ ”⁶⁰ Similar atextual readings have been adopted elsewhere.⁶¹

Feeney freed governmental decisionmakers to take race into account at a population level. Decisionmakers redesigned an employment examination to diminish its disparate racial impact.⁶² They created school assignment policies that aimed to diversify the racial makeup of schools.⁶³ They established hiring preferences for socially disadvantaged populations not specifically defined by race.⁶⁴ They revised policies for access to subsidized apartments to ensure they were “made available to all applicants regardless of race” rather than favoring prior community residents.⁶⁵ A majority of the Supreme Court has reaffirmed the permissibility of policies motivated by beneficial racial effects, holding that “local housing authorities may choose to foster diversity and combat racial isolation with race-neutral tools.”⁶⁶ These precedents provide governmental decisionmakers with flexibility recognized by lower courts.⁶⁷

Both Feeney and Inclusive Communities sharply distinguished individual racial classifications, which prompt strict scrutiny, from intended effects on the racial makeup of a broader population, which do not. But despite these precedents, both the Supreme Court and lower courts inconsistently uphold the distinction between permissible means and permissible ends. Occasionally, they have applied strict scrutiny to policies that neither classify by race nor intend adverse group-level racial outcomes.

The case that most muddles the means/ends distinction is Ricci v. DiStefano, which invalidated a municipality’s plan to rerun an administered employment examination in order to avoid a racially disparate outcome.⁶⁸ Although rerunning the employment examination sought to diversify the racial makeup of a population, it did not classify any examinee by race. Yet the Court puzzlingly evaluated the decision to rerun the examination using a test developed in Croson, a Supreme Court case evaluating individual racial classifications, rather than using the tests applicable to nonclassificatory policies.⁶⁹ Per the Ricci Court, Croson “held that certain government actions to remedy past racial discrimination—actions that are themselves based on race—are constitutional only when there is a ‘strong basis in evidence’ that the remedial actions were necessary.”⁷⁰ But the “government actions” at issue in both Croson and in Wygant, the case from which Croson drew its “strong basis in evidence” test—were not merely motivated by or “based on” race. They were explicit, individual racial classifications.⁷¹ Nothing in Croson requires a “strong basis in evidence” for nonclassificatory policies, even if they aim to remedy past racial discrimination or pursue other group-level racial goals.⁷² Indeed, Croson itself endorses “the use of race-neutral means to increase minority business participation in city contracting.”⁷³

The same doctrinal muddling of racial classifications and intended outcomes infects other parts of the Ricci opinion. The Court claimed that rerunning the examination would “amount[] to the sort of racial preference that Congress has disclaimed,”⁷⁴ even though neither the original nor the revised examination considered the examinee’s race. Further, it claimed that rerunning the examination would be “antithetical to the notion of a workplace where individuals are guaranteed equal opportunity regardless of race,”⁷⁵ even though all examinees would receive the same examination regardless of race.

Ricci justified its use of the Croson test by claiming that because a municipality “cannot rescore a test based on the candidates’ race . . . then it follows a fortiori that it may not take the greater step of discarding the test altogether to achieve a more desirable racial distribution of promotion-eligible candidates.”⁷⁶ This a fortiori is an ungrounded ipse dixit. Rescoring a test based on individual candidates’ race would treat examinees differently based on an individual racial classification and thus clearly trigger strict scrutiny. Discarding the test altogether, in contrast, would not be a “greater step.” Rather, it is an individually race-neutral way of achieving “a more desirable racial distribution”⁷⁷—just like the race-neutral efforts to improve minority business participation that Croson recommended.

The Court also drew on an ungrounded distinction between examination design and redesign. It claimed that once “employers have made clear their selection criteria, they may not then invalidate the test results, thus upsetting an employee’s legitimate expectation not to be judged on the basis of race.”⁷⁸ But invalidating the test results for all employees would not judge anyone on the basis of their race. Rather, it would be akin to the individually race-neutral decisions to revise established school assignment policies that the Court and lower courts have recognized as acceptable.⁷⁹ Any entitlement to the status quo testing regime is contractual or quasi-contractual, not Constitutional.

Ultimately, Ricci’s holding is doctrinally inapposite to its facts. The Court purported to hold only that “before an employer can engage in intentional discrimination for the asserted purpose of avoiding or remedying an unintentional disparate impact, the employer must have a strong basis in evidence to believe it will be subject to disparate-impact liability if it fails to take the race-conscious, discriminatory action.”⁸⁰ But rerunning the examination was not intentional discrimination or a “discriminatory action.” Its goal was not to adversely affect any racial group but, rather, to ensure that no group was unfairly excluded from opportunity, and the means proposed to achieve that goal—revising and rerunning the examination—would not have treated any examinee differently on the basis of their individual race.

Despite applying Croson to evaluate the policy at issue, the Court decided Ricci on statutory rather than constitutional grounds. One explanation for the confusion in Ricci may arise from its abridgment of Title VII’s language. Title VII establishes that it is unlawful for an employer “to fail or refuse to hire or to discharge any individual, or otherwise to discriminate against any individual with respect to his compensation, terms, conditions, or privileges of employment, because of such individual’s race.”⁸¹ But the Ricci court summarized this statutory text as “because of . . . race.”⁸² The ellipsis concealed an important distinction: Title VII prohibits treating an employee differently, or making adverse employment decisions about an employee, because of that employee’s race. It does not, however, prohibit making individually race-neutral employment decisions with the racial makeup of the workforce as a consideration.

The Ricci Court also suggested that, even if the city did have a strong basis in evidence for its decision to rerun the examination, that decision might trigger strict scrutiny under the Equal Protection Clause. It observed that its “holding does not address the constitutionality of the measures taken here in purported compliance with Title VII,” nor establish “that meeting the strong-basis-in-evidence standard would satisfy the Equal Protection Clause in a future case.”⁸³ The Court also remarked that “we need not decide whether a legitimate fear of disparate impact is ever sufficient to justify discriminatory treatment under the Constitution.”⁸⁴ Again, however, the policy invalidated in Ricci should not constitute discriminatory treatment under the Constitution; it neither classified any candidate by race nor aimed to impose an adverse impact on any racial group.

In my view, the most defensible understanding of Ricci’s holding is what Richard Primus called the “visible-victims” interpretation of the case.⁸⁵ Even though the white firefighters who initiated the Ricci litigation were never treated differently on the basis of their race, their lost opportunity was sufficiently closely tied to their racial identity to treat them akin to employees who are disadvantaged by individual racial classifications.⁸⁶

Justice Scalia’s Ricci concurrence further obscured the distinction between racial classifications and race-conscious, nonclassificatory policies. In his concurrence, he suggests that “requiring employers to evaluate the racial outcomes of their policies, and to make decisions based on (because of) those racial outcomes,” “is, as the Court explains, discriminatory.”⁸⁷ He cites Feeney for this proposition.⁸⁸ But Feeney does not say it is discriminatory to make decisions because of the racial outcomes that will result. Feeney says it is discriminatory to make decisions because of the adverse racial outcomes that will result.⁸⁹ Under Feeney, Croson, and Inclusive Communities, decisions made in part because of their beneficial racial outcomes—including reduced racial health disparities—are not thereby discriminatory nor do they trigger strict scrutiny. Strict scrutiny applies only if a policy aims to inflict an adverse impact on a racial group or if it treats individuals differently based on their race.

Scalia’s concurrence also provides a hypothetical designed to conflate race-conscious policy design with racial classifications. He asked, “Would a private employer not be guilty of unlawful discrimination if he refrained from establishing a racial hiring quota but intentionally designed his hiring practices to achieve the same end? Surely he would.”⁹⁰ Indeed such an employer would violate the law, but the violation would stem solely from the impermissible end—the racial hiring quota. Scalia’s imagined hiring practice classified no individual by race and so would not trigger strict scrutiny on that ground: Strict scrutiny only applies and is violated because of the quota. The relevant question is instead: Does a private employer unlawfully discriminate if he intentionally designs his hiring practices to achieve a permissible goal such as racial diversity or reduced racial isolation, while refraining from considering any applicant’s race? Feeney and other cases say he does not.

3. Strict Scrutiny Without Racial Classifications

Although Ricci’s conclusion is factually confused and doctrinally inconsistent with the mainstream of Supreme Court jurisprudence, there are better-established bases for applying strict scrutiny to governmental actions that intend racial effects but do not classify by race.

a. Intended Adverse Effects

First, as Feeney recognizes, policies that intend adverse effects on a racial group face strict scrutiny. When a policy is allocating a limited resource—such as educational places, employment opportunities, or scarce medical treatments—the line between intending to benefit one racial group and intending to adversely affect another can be delicate.⁹¹

b. Racial Balancing

Second, the Court has looked askance at governmental efforts to “racially balance” government-operated programs. Racial balancing involves either seeking a specified proportion of different racial groups for its own sake,⁹² or having a program’s racial composition mirror the population from which participants are drawn.⁹³

Precisely when racial balancing is proscribed remains uncertain. Because the Supreme Court has only rejected racial balancing as unacceptable where it was pursued via an individual racial classification, its precedents may merely deny that racial balancing is a sufficiently compelling interest to justify a racial classification, not reject racial balancing as inherently objectionable.⁹⁴ Some federal appellate courts have adopted this position.⁹⁵ However, the Court has elsewhere suggested that racial balancing not only fails to be a compelling interest but is a categorically unacceptable governmental aim.⁹⁶ Some appellate opinions and other judicial statements have thus indicated that even individually race-neutral programs are on sounder legal footing when they eschew racial balancing, or are unacceptable if they aim at racial balancing.⁹⁷

c. Pretexts and Proxies for Racial Classifications

Third, a “classification that is ostensibly neutral but is an obvious pretext for racial discrimination” prompts the same strict scrutiny as a facial racial classification.⁹⁸ The charge of pretext has been leveled against classifications based on Native American status, ancestry, economic status, and place of residence.

Under current precedent, policies that treat Native Americans differently based on connection to a recognized tribe are political rather than racial classifications that prompt only rational basis scrutiny.⁹⁹ Litigants have unsuccessfully argued, however, that Native American status should instead be treated as a racial classification that prompts strict scrutiny. Most recently, in the adoption case Brackeen v. Haaland, the Supreme Court bypassed an equal protection challenge to a policy considering Native American status for lack of standing.¹⁰⁰ Justice Kavanaugh, however, concurred to assert that in his view:

[T]he equal protection issue is serious. . . . [A] child in foster care or adoption proceedings may in some cases be denied a particular placement because of the child’s race . . . . And a prospective foster or adoptive parent may in some cases be denied the opportunity to foster or adopt a child because of the prospective parent’s race.¹⁰¹

In Justice Kavanaugh’s view, these scenarios “raise significant questions under bedrock equal protection principles.”¹⁰² Distinguished experts on Native American law, however, disagree with the justice.¹⁰³

It is unclear how other members of the Supreme Court would view a future case squarely challenging, on equal protection grounds, a policy—including a medical one—that considers Native American status. In their Brackeen concurrence, Justices Gorsuch, Jackson, and Sotomayor endorse the “bedrock principle that Indian status is a ‘political rather than racial’ classification.”¹⁰⁴ Justice Kagan has endorsed this approach elsewhere.¹⁰⁵ In contrast, Justices Thomas, Roberts, and Alito have previously indicated willingness to entertain equal protection challenges to policies that consider Native American status.¹⁰⁶ Although Justice Barrett authored Brackeen, she has not either there or elsewhere indicated her stance on how Native American status should be viewed.

In contrast to classifications based on Native American status, classifications based on ancestry generally prompt strict scrutiny because they are regarded as equivalent to, or proxies for, racial classifications.¹⁰⁷ In rare cases, however, classifications based on ancestry may satisfy strict scrutiny.¹⁰⁸

Allegations that classifications other than ancestry or Native American status constitute pretexts for racial discrimination have been unavailing. For instance, basing school placements on economic status,¹⁰⁹ or the demographics of the area where one resides,¹¹⁰ does not trigger strict scrutiny. Two separate opinions have asserted that considering neighborhood racial demographics should trigger strict scrutiny,¹¹¹ but this approach has not commanded a majority of any court.

SFFA has also provided a new basis for pretext challenges by warning that “universities may not simply establish through application essays or other means the regime we hold unlawful today.”¹¹² This language may prompt litigation to challenge classifications—such as those based on the content of essays—that are believed to serve as proxies for racial classifications of applicants.

d. Predominant Racial Motives

Fourth, some cases hold or suggest that strict scrutiny is triggered when a policy’s predominant aim concerns race, even absent any individual racial classification or any intent to adversely affect any racial group or to racially balance. This “predominance” test originates in districting cases, where the Court applies strict scrutiny if race is the predominant factor used in drawing district lines, but recognizes that some consideration of race is acceptable.¹¹³ Courts have also regarded whether race was the predominant motivation for a decision as relevant outside the districting context.¹¹⁴

The predominance test presents an alternative basis for the Court’s Ricci holding. Ricci explained that “[t]he problem, of course, is that after the tests were completed, the raw racial results became the predominant rationale for the City’s refusal to certify the results.”¹¹⁵ This explanation would reach a more modest result not dependent on conflating facially neutral, race-conscious decisions with racially classificatory ones. On this approach, race-conscious aims would only trigger strict scrutiny if they predominated over others.

B. Federal Statutory Law

In addition to the Equal Protection Clause, three federal statutes also constrain the consideration of individuals’ race.

The first is Title VI of the Civil Rights Act of 1964. This statute provides that “[n]o person in the United States shall, on the ground of race, color, or national origin, be excluded from participation in, be denied the benefits of, or be subjected to discrimination under any program or activity receiving Federal financial assistance.”¹¹⁶ Despite its text, governing precedent reads Title VI’s requirements as identical with the Equal Protection Clause’s.¹¹⁷ Accordingly, Title VI merely subjects all recipients of federal financial assistance to the precedents described in Section I.A.

In his SFFA concurrence, Justice Gorsuch proposed reading Title VI as more restrictive of race-based decisionmaking than the Equal Protection Clause. In his view, “[u]nder Title VI, it is always unlawful to discriminate among persons even in part because of race, color, or national origin,” even if such discrimination would satisfy strict scrutiny.¹¹⁸ Justice Gorsuch’s concurrence also recognizes, however, that SFFA in any event brings the Equal Protection Clause closer to his understanding of Title VI. His concurrence takes the position that Title VI proscribes only discrimination against individuals because of their individual race, not race-conscious decisionmaking more generally.¹¹⁹ And it states that SFFA “returns this Court to the traditional rule that the Equal Protection Clause forbids the use of race in distinguishing between persons unless strict scrutiny’s demanding standards can be met.”¹²⁰ The daylight between the SFFA majority and Justice Gorsuch’s concurrence therefore only illuminates a narrow sliver of cases: ones where a racial classification used by a recipient of federal financial assistance would satisfy strict scrutiny but nevertheless disadvantages some individuals on the basis of their race.

The second relevant statute is Section 1557 of the Affordable Care Act. This section simply extends Title VI’s requirements to various health system actors, such as providers and insurers. Courts now agree that it provides no new content or enforcement mechanism for Title VI.¹²¹ Therefore, liability under Section 1557 will depend on how Title VI is interpreted.

Third, Section 1981 of the 1866 Civil Rights Act textually guarantees that:

All persons within the jurisdiction of the United States shall have the same right in every State and Territory to make and enforce contracts, to sue, be parties, give evidence, and to the full and equal benefit of all laws and proceedings for the security of persons and property as is enjoyed by white citizens, and shall be subject to like punishment, pains, penalties, taxes, licenses, and exactions of every kind, and to no other.¹²²

Section 1981 also defines the right to make and enforce contracts as a right to the “enjoyment of all benefits, privileges, terms, and conditions of the contractual relationship.”¹²³ Notwithstanding the Act’s text, however, the Supreme Court concluded that “the prohibition against discrimination in § 1981 is co-extensive with the Equal Protection Clause.”¹²⁴

Published caselaw applying Section 1981’s nondiscrimination provisions to the facial use of racial classifications is limited, with the leading appellate case declining to apply an Equal Protection Clause analysis.¹²⁵ Justice Gorsuch’s concurrence in SFFA suggests that he might also read Section 1981 differently from the Equal Protection Clause, due to the enactments’ different text.¹²⁶ Textually, Section 1981 promises citizens of all races the same rights as are “enjoyed by white citizens,” which would not necessarily prohibit racial classifications or race-conscious decisionmaking.¹²⁷

Current precedent leaves ambiguous how extensively Section 1981 constrains the use of race by private actors receiving no federal funding.¹²⁸ Precedent in this area, however, will likely develop rapidly. Following SFFA , litigants have brought Section 1981 suits against multiple private businesses and organizations, including in the health sector, that consider applicants’ race.¹²⁹ Some such challenges have met with success. For instance, a group represented by the same attorneys who argued SFFA secured a preliminary injunction against a private grant program open only to Black applicants.¹³⁰ And they did so over a dissent that questioned whether such a use of Section 1981 was consistent with congressional intent and also observed that defendants’ aim to “convey the importance of Black women-owned business” was protected by the First Amendment.¹³¹ A merits panel later upheld the preliminary injunction.¹³² A similar Section 1981 challenge has been brought against the National Association of Emergency Medical Technicians’ operation of a scholarship for “students of color,” but a temporary restraining order was denied due to lack of standing because the scholarship application window had not yet opened.¹³³ Other recent Section 1981 cases, including ones involving health programs, have led to the removal of considerations of race, potentially indicating defendants’ concern that courts might not permit private parties to consider individual recipients’ race when disbursing funds.¹³⁴ Development of Section 1981 precedent concerning private actors will be practically important in health contexts because some private-sector health system actors are not covered by Section 1557, Title VI, or the Equal Protection Clause,¹³⁵ and because Section 1981 can allow for quicker litigation.¹³⁶

II. Antiracist Medicine

Part I presaged an impending collision between an increasingly colorblind Supreme Court and a storied profession moving in the opposite direction.¹³⁷ The American Medical Association (AMA), the “largest professional association of physicians and medical students in the United States,”¹³⁸ has sought to become an “anti-racist, diverse, multicultural organization.”¹³⁹ In its equity strategic plan, the AMA criticizes “color- and identity-evasive actions and policies” for magnifying inequities.¹⁴⁰ It seeks to ensure that the “race [and] ethnicity . . . of health care investors and innovators mirrors that of our nation”¹⁴¹—the exact sort of “proportional representation” SFFA castigated as unconstitutional racial balancing.¹⁴² Even as the Supreme Court was embracing color-blindness in SFFA, the AMA’s House of Delegates endorsed “racism-conscious, reparative, community-engaged interventions at the health system, organized medical society, local, and federal levels which seek to identify, evaluate, and address the health, economic, and other consequences of structural racism in medicine.”¹⁴³

The AMA and other medical societies also partnered with legal experts to endorse race-conscious approaches. Forty-six medical organizations, including the AMA, filed amicus briefs supporting the SFFA defendants.¹⁴⁴ Separately, the AMA’s litigation center filed amicus briefs that “explain[ed] the importance of physicians considering race and ethnicity when determining a treatment plan for COVID-19 patients” in recent litigation over COVID-19 treatment allocation.¹⁴⁵

Medicine and health systems might consider race for three primary reasons. Each has distinct legal implications. First, they might consider race with the goal of improving individual and overall population health, irrespective of the beneficiaries’ racial identity. Call this a beneficent justification. Second, they might seek to reduce health disparities between racial groups—a racially egalitarian justification. Last, they might strive to redress historical racial injustice—a racially reparative justification.

To illustrate these justifications, compare three different reasons why health systems might track, and seek to reduce, racial disparities. A health system might “identify ways to reduce or eliminate racial and ethnic disparities that contribute to infant mortality.”¹⁴⁶ This is a beneficent justification: The policy’s aim is to reduce overall infant mortality, and reducing racial disparities is merely a means to that end. A health system might also seek to “reduce racial disparities in infant mortality.”¹⁴⁷ This is a racially egalitarian justification: The aim is to reduce mortality disparities across racial groups.

Beneficent and racially egalitarian goals frequently coincide. Texas’ Health and Safety Code includes a program aiming to “reduce racial disparities in the rates of and the overall rates of fetal and infant mortality.”¹⁴⁸ Reducing racial disparities in the rate of infant mortality is an egalitarian aim, whereas reducing the overall rate itself is a beneficent aim. Reducing the rate of infant mortality in the hardest-hit racial groups advances both beneficent and egalitarian aims.

Last, a health system could seek to specifically reduce infant mortality in racial groups who have experienced higher rates of infant mortality due to historical injustice. No health system has explicitly framed efforts to reduce racial disparities in infant mortality as remedial or reparative. Some scholars, however, have suggested remedial rationales for mitigating other racial health disparities.¹⁴⁹ Because historical injustice often produces present-day disadvantage, and because assisting the disadvantaged also typically advances overall health, reparative aims will also often overlap in practice with beneficent and egalitarian ones.¹⁵⁰

While beneficent, racially egalitarian, and racially reparative aims can overlap and be simultaneously realized by the same policy, they raise distinct legal questions. Beneficent justifications raise no legal issue in themselves, and generally constitute compelling interests. Rather, law becomes involved when racial classifications are used to achieve beneficent ends. In contrast, the pursuit of racially egalitarian or reparative aims presents a greater risk of impermissible motive, even if racial classifications are not used.

This Part examines five contexts in which medicine and health systems consider the race of the health professionals who provide medical treatment or the patients who receive it: the recruitment and training of health professionals (Section II.A), the professional-patient encounter (Section II.B), the allocation of scarce medical treatments (Section II.C), public health (Section II.D), and medical research (Section II.E).

A. Recruiting and Training Professionals

Consideration of individual race in the selection and training of health professionals has been defended on beneficent, egalitarian, and reparative grounds. Citing amicus briefs, Justice Sotomayor’s SFFA dissent asserted that “increasing the number of students from underrepresented backgrounds who join ‘the ranks of medical professionals’ improves ‘healthcare access and health outcomes in medically underserved communities,’ ” and that “all physicians become better practitioners when they learn in a racially diverse environment.”¹⁵¹ Likewise, Justice Jackson’s dissent argued that “the diversity that UNC pursues for the betterment of its students and society is not a trendy slogan[,] [i]t saves lives[,]” and explained that “Black physicians are more likely to accurately assess Black patients’ pain tolerance and treat them accordingly.”¹⁵² Improving health outcomes among Black patients both narrows racial health disparities and improves overall population health.

Justice Jackson’s dissent also offered reparative arguments. She observed that “[t]he race-based gaps that first developed centuries ago are echoes from the past that still exist today,” and attributes “race-based wealth, health, and opportunity gaps” to unjust policies including slavery, Jim Crow segregation, and exclusion of Black Americans from the Homestead Act’s benefits.¹⁵³ Among the health gaps identified by her dissent are disparities in deaths during childbirth and in overall life expectancy.¹⁵⁴ Addressing these health gaps would advance beneficent and egalitarian goals as well as reparative ones, since both overall health and the health of Black Americans would improve.

The SFFA majority neither discussed the health implications of its holding nor attempts to rebut the dissents’ empirical assertions. Perhaps these health claims were not compelling interests. But, more likely, the university defendants’ policies were viewed as insufficiently tailored to serve those interests. In his concurrence, Justice Thomas denied that “Justice Jackson’s statistics regarding a correlation between levels of health, wealth, and well-being between selected racial groups prove anything.”¹⁵⁵ Also in concurrence, Justice Gorsuch, meanwhile, chided Justice Sotomayor for “venturing beyond the trial records to discuss data about . . . healthcare.”¹⁵⁶

Individual racial classifications have been used in admissions and financial aid processes for health professionals. For instance, the Congressional Black Caucus sponsors a scholarship open to “African American and Black scholars pursuing an undergraduate, graduate or doctoral degree leading to a career as a healthcare professional.”¹⁵⁷ Some universities invite underrepresented secondary school students to learn about career opportunities in medicine.¹⁵⁸ Many academic medical centers have used admissions policies akin to those struck down in SFFA.¹⁵⁹

Like other admissions and financial aid programs that classify candidates by race, these programs in the health professions have faced intensifying legal challenges, which will further intensify after SFFA. Requirements that medical boards responsible for licensing and disciplining trainees include members of racial minority groups have likewise faced legal attacks.¹⁶⁰ Even before SFFA, scholarships open only to students of a specific race were successfully challenged in court as not narrowly tailored to realize diversity.¹⁶¹ In many cases, universities and organizations have elected to drop racial classifications rather than face litigation.¹⁶²

Occasionally, medical training itself incorporates racial classifications. Physicians at the University of California, San Francisco (UCSF) recently published a Perspective in the influential New England Journal of Medicine describing and advocating the use of “racial affinity group caucuses” (RAGCs) as part of training.¹⁶³ These RAGCs “are facilitated sessions involving participants grouped according to self-identified racial or ethnic identity to support integration of antiracism curricula into clinical practice.”¹⁶⁴ UCSF offered “RAGCs for three racial and ethnic identity groups: Black or African American, all people of color, and White,” and ensured that RAGC “facilitators share[d] their group’s racial identity.”¹⁶⁵ The authors view RAGCs as valuable “throughout medical school and in residency programs as an opportunity for learners to contextualize antiracism curricula, bring greater depth and meaning to such curricula’s lessons, build supportive communities, and expand knowledge and skill in antiracist doctoring,” and suggest that they “provide an improved sense of belonging and a safe space where BIPOC learners can reclaim their voices.”¹⁶⁶

RAGCs were rapidly challenged. An activist complained to the Department of Education’s Office for Civil Rights that “the University’s RAGCs [were] treating individuals differently by segregating and separating on the basis of race and color” and constituted “illegal race-based discrimination in violation of Title VI.”¹⁶⁷ Do No Harm’s leaders claim their complaint “prompt[ed] the university to cancel its racial affinity groups before the investigation could conclude.”¹⁶⁸ They also claimed similar success in their parallel “complaint against Harvard’s T.H. Chan School of Public Health.”¹⁶⁹

Under present precedent, universities may have been wise to cancel the RAGCs. They classified students by race, which would trigger strict scrutiny and therefore require narrow tailoring to realize a compelling interest, such as patient health.¹⁷⁰ This illuminates a puzzle in compelling interest analysis: A new program may struggle to show that it is supported by a compelling interest, because it will lack supporting data at first.¹⁷¹ The universities would also have needed to show that the racial classifications used to constitute the RAGCs were narrowly tailored to serve the asserted interest. To do so, they would have needed to show serious, good-faith consideration of race-neutral alternatives (such as different caucusing groups for students with different levels of familiarity with antiracist doctoring, or for students interested in addressing specific types of racial disparities).¹⁷² They would also have needed to show that the classifications were not overbroad, vague, or underinclusive. RAGCs for “all people of color”¹⁷³ might have been particularly challenging to defend with respect to this requirement.¹⁷⁴

B. The Provider-Patient Encounter

As health professionals move through and ultimately complete training, they begin providing health care services. The clinical encounter between health professionals and patients sometimes involves decisions made in reliance on racial classifications. This Subpart discusses three such examples: professional-patient racial concordance, racial classifications in clinical practice guidelines, and hospital admissions policies that classify patients by race.

1. Professional-Patient Racial Concordance

Racial concordance occurs when patients are treated by health professionals of the same race. Concordance may improve overall patient outcomes as well as reducing racial disparities, thereby serving both beneficent and egalitarian ends.¹⁷⁵ Many state health system task forces endorse racial concordance.¹⁷⁶

Like other race-aware aims, racial concordance can be pursued at a “macro” or “meso” level, by increasing the makeup of professionals of specific races in a jurisdiction or health system. It can also be pursued at the “micro” level of the clinical encounter itself. At the level of the clinical encounter, concordance requests by patients are widespread and frequently honored.¹⁷⁷

a. Micro-Level Concordance

How would honoring patients’ racial concordance requests at the clinical level fare under current precedent? The practice treats professionals differently on the basis of their race and thus would face strict scrutiny under the Equal Protection Clause if pursued by a state actor such as a public or state university hospital. Kimani Paul-Emile argues that “improving patient care, countering racial bias, reducing race-based health disparities, and saving lives should qualify as a compelling interest that justifies bowing to patients’ racial preferences.”¹⁷⁸ It is doubtful under current precedent that reducing racial health disparities or countering racial bias constitute compelling interests for purposes of a strict scrutiny analysis, unless the disparity or bias identifiably stems from the prior conduct of the state actor receiving the concordance request.¹⁷⁹ Under current precedent, in contrast, improving patient care and saving lives should constitute a compelling interest.¹⁸⁰ Showing that concordance improves care and saves lives, however, will likely involve a battle of experts.¹⁸¹

Narrow tailoring may be more challenging. It should be easy to show that the classification is not over- or underinclusive, since it is a simple race match. Challenges could arise, however, if the racial classifications used to fulfill concordance requests do not match up with the classifications used to prove the compelling benefits of concordance—for instance, if Black patients are matched with physicians of color more generally, rather than with Black physicians.¹⁸² Proving that race-neutral alternatives to concordance have been seriously considered in good faith, meanwhile, will be more challenging. Even organizations supporting concordance recognize that there are workable alternatives.¹⁸³ Relatedly, particularly after SFFA, it may be necessary to add a durational limit to these programs.

Most seriously for narrow tailoring, concordance can impose genuine burdens on third parties. Racial classifications confining employees to interacting with same-race clients have been found unlawfully discriminatory under Title VII.¹⁸⁴ In the medical context, the Seventh Circuit held that “Title VII allows an employer to respect a preference for same-sex health providers, but not same-race providers,”¹⁸⁵ and has rejected the view that a state agency providing counseling services “was entitled to consider the preferences of its clientele . . . for a counselor of the same race” when making hiring, promotion, or retention decisions.¹⁸⁶ Lower courts have reached similar conclusions.¹⁸⁷ In particular, concordance may diminish work opportunities for medical professionals whose racial group is overrepresented compared to the patient population. Terminating such employees to pursue concordance is likely illegal. Meanwhile, professionals of races underrepresented compared to the patient population may find their work opportunities constrained to treating same-race patients, especially if those same-race patients highly value concordance.¹⁸⁸

Because honoring concordance requests is an employment action, it must also satisfy Title VII (for employers with more than 15 employees),¹⁸⁹ and Section 1981 (for all employers).¹⁹⁰ These statutes provide an additional protection for government employees and serve as the sole federal protections for nongovernmental employees against racial discrimination.¹⁹¹ Title VII has been read to permit remedial affirmative action in hiring and is more lenient with respect to such affirmative action than the Equal Protection Clause.¹⁹² But Title VII categorically prohibits the nonremedial consideration of race in employment.¹⁹³ For this reason, some experts who favor fulfilling racial concordance requests argue that Title VII would need to be amended first.¹⁹⁴ Case law has similarly invalidated the non-remedial consideration of race in employment settings under Section 1981.¹⁹⁵

In several states restricting affirmative action, concordance requests also face the hurdle that the “state shall not . . . discriminate against . . . any individual or group on the basis of race . . . in the operation of public employment.”¹⁹⁶ This language is relatively new and has produced only limited state law precedent in many states.¹⁹⁷ State medical employers, such as county or state university hospitals and medical centers, would need to satisfy this restriction.

Because concordance requests are initiated by private individuals rather than employers, employers could also argue that acting on a private party’s racially classificatory request does not constitute a racial classification. In Monroe v. City of Charlottesville, the Fourth Circuit concluded that when a suspect “description included the fact that the suspect was African-American, but the officers were not the source of that portrayal,” no governmental racial classification existed to trigger strict scrutiny.¹⁹⁸ The court’s reasoning is unpersuasive. When officers seek out suspects on the basis of the description, they engage in government action in reliance on a racial classification, just as a legislature that enshrined a privately drafted distinction into law would create a classification.¹⁹⁹ Nevertheless, this precedent provides one potential foothold for the use of concordance.

b. Meso- and Macro-Level Concordance

At the population level, decisionmakers can pursue individually race-neutral approaches enabling patients to be treated by a same-race professional. Such approaches might include recruiting professionals from specific geographic areas or from training settings such as Historically Black Colleges and Universities or Hispanic-Serving Institutions. They might also include recruiting professionals who state, or otherwise evidence, a commitment to treating patients from specific racial groups or backgrounds.

These individually race-neutral approaches would be unlikely to trigger strict scrutiny under current federal precedent. They should also satisfy state laws that stringently restrict the use of individual racial classifications.

Patients, meanwhile, remain free to request providers on whatever basis they prefer, so long as no provider is forced by their employer or by government to accede to a racial classification. Some grey areas exist—for instance, whether platforms for finding a provider may permit patients to search for providers by race.²⁰⁰ But the patient’s internal motivation for selecting a provider remains beyond the reach, and likely the proper ken, of the law.

2. Racial Classifications in Clinical Practice Guidelines

Guidelines for medical practice sometimes direct providers to consider their patients’ race. Some guidelines such as these have been questioned and even rescinded.²⁰¹ But many racial classifications remain in governmental guidance and the recommendations of eminent medical bodies.

As an example, FDA-approved drug labels include racial classifications. The label for lisinopril, an antihypertension drug, states that it has “an effect on blood pressure that is less in Black patients than in non-Blacks.”²⁰² FDA guidance for rosuvastatin, a cholesterol-lowering drug, states that a “5 mg dose . . . should be considered as the start dose for Asian patients and any increase in dose should take into consideration the increased drug exposure in this patient population.”²⁰³ Eltrombopag, a drug for aplastic anemia and thrombocytopenia, is to be offered at a lower dose for “patients of East Asian ancestry.”²⁰⁴ The labeling for telbivudine, a hepatitis B drug, notes that its “safety and efficacy . . . have not been evaluated in Black/African American or Hispanic patients.”²⁰⁵

State and local governments also include racial classifications in medical guidance. Some states recommend earlier testicular and prostate cancer screening for adults with “African-American ancestry.”²⁰⁶ California’s Department of Public Health recommends that providers “consider screening Black/African American women up to age 30” for certain sexually transmitted infections.²⁰⁷ The New York City Department of Health recommends that physicians discuss cancer screening “especially for Black patients because of high cancer mortality rates among Black New Yorkers.”²⁰⁸ The Bureau of Prisons screens for hypertension more frequently in Black patients.²⁰⁹ Some insurers likewise recommend that for colorectal cancer screening, if you are “African American, American Indian or Alaska Native, start at age 45.”²¹⁰

These differential recommendations involve the use of individual racial classifications to make decisions, and so facially implicate the Supreme Court’s precedents requiring strict scrutiny of racial classifications.²¹¹ Some courts, however, have read these precedents to apply only where the classification is used to allocate benefits or burdens.²¹² In the areas where courts have applied strict scrutiny to the use of racial classifications—educational admissions, employment, and eligibility for grants and contracts—individuals are competing. Using racial classifications to improve access to opportunity for some candidates diminishes, at least in the immediate term, access for others. In contrast, using racial classifications to optimize decisions about when to begin screening for a condition, or how to dose a drug, does not have the same effect on those who are classified.

The absence of competitive effects reveals a flaw in one argument against the use of racial classifications. Justice Thomas insists that it “is not even theoretically possible to ‘help’ a certain racial group without causing harm to members of other racial groups,”²¹³ and that “it should be obvious that every racial classification helps, in a narrow sense, some races and hurts others.”²¹⁴ But screening African American men at age 45 for colon cancer may be a net benefit for them, whereas screening white men at age 45 may be a net harm given differences in colon cancer risk.²¹⁵ Considering one patient’s race when it is predictive of outcomes is not zero-sum and need not even impose any burdens on patients of other races.

The Supreme Court’s current precedents subjecting all racial classifications to strict scrutiny, however, are not grounded in Justice Thomas’ zero-sum vision of race. Rather, the principle at their core is more plausibly dignitary: There is something distinctively objectionable about suffering a bad outcome because of how others classified you by race, as opposed to how others classified you in some other respect, such as geography, age, or health.²¹⁶ This dignitary approach might require strict scrutiny even of benefit-maximizing racial classifications. In dissent in Bush v. Vera, a redistricting case, Justice Stevens criticized the majority’s approach to racial classifications as so absurd that it would “prohibit the Public Health Service from targeting African-American communities in an effort to increase awareness regarding sickle-cell anemia.”²¹⁷ The plurality opinion responded by claiming that “we subject racial classifications to strict scrutiny precisely because that scrutiny is necessary to determine whether they are benign—as Justice Stevens’ hypothetical of a targeted outreach program to protect victims of sickle cell anemia would, no doubt, be.”²¹⁸ In light of Bush v. Vera and lower courts’ assessment of similar issues, it appears that the use of racial classifications in diagnosis and screening must satisfy strict scrutiny if challenged in court.

Is the Vera plurality correct that a racially targeted sickle-cell anemia screening program would readily satisfy strict scrutiny? The targeted outreach program would clearly serve a compelling interest (health and human life). The narrow tailoring inquiry, however, would require showing that race-neutral alternatives had been seriously considered in good faith and found unworkable, that the use of race as a proxy for sickle-cell anemia was not over- or underinclusive, that the program was durationally limited, and that the rights of third parties were respected. It would be fairly easy to show that third parties are unharmed by the program. But it might not be as simple as the plurality believed to show that the racial classification was not over- or underinclusive, or that race-neutral alternatives would not be workable. Kimani Paul-Emile, for instance, argues that the sickle-cell trait does not “express race in a biologically legitimate way.”²¹⁹ As she notes, “[P]hysicians believed that sickle cell disease was exclusively African, but it can also be found in other areas where malaria is endemic, such as Greece, Saudi Arabia, Turkey, and Iran.”²²⁰ Indeed, “[T]he highest incidence of the gene mutation—a rate double that for African Americans—can be found in a population in Greece,”²²¹ demonstrating that identifying sickle-cell trait with Black patients is underinclusive. Meanwhile, “Kenyans have a low frequency of sickle cell,”²²² demonstrating overinclusiveness. Paul-Emile concludes that “perceived race or an individual’s racial self-identification” does not “reliably indicate anything meaningful about her or his genes.”²²³ Paul-Emile’s explanation suggests an appealing race-neutral alternative: Rather than relying on self-identified race, medical decisionmakers might conduct outreach based on geographic ancestry.

Medical decisionmakers could also conduct genetic testing to directly determine whether a patient in fact has sickle-cell trait, rather than relying on proxies. Genetic testing—which would itself require outreach—would be impractical for the sort of outreach program debated in Vera. But precedent indicates that a race-neutral alternative must be pursued even at some net cost: If “a nonracial approach . . . could promote the substantial interest about as well and at tolerable administrative expense,” government may not use racial classifications.²²⁴ The frequently high stakes of medical decisionmaking, however, make defining “about as well” challenging. For instance, if an approach eschewing racial classifications is expected to save fewer statistical lives or years of life, it is not clear whether it works “about as well,” or meaningfully less well.

3. Racial Classifications in Hospital Admissions

In contrast to the use of racial classifications in screening and diagnosis recommendations, their use in some areas of medical practice may genuinely place the interests of different patients into conflict even absent the sorts of absolute scarcity discussed in the next Subpart. A recent program, the Healing ARC, piloted by Boston’s Brigham and Women’s Hospital and the University of Virginia Medical Center exemplifies this potential tension. The authors describe the program’s goal as being to “operationalize equity and provide redress to the communities of color that we have inequitably served for decades.”²²⁵ The program design was modeled on a “framework of acknowledgment, redress, and closure (ARC) from leading reparations scholars,” and was refined through Wisdom Councils comprising “members recruited by racial equity organizers from local Black and Latinx communities based on health equity interest and experience.”²²⁶

The Wisdom Council process generated the following clinical protocol: When a patient in the emergency room requires admission to the hospital for a likely diagnosis of heart failure, and the admitting provider selects the General Medical Service, the computerized Clinical Decision Support System would respond differently depending on whether the patient being admitted is white or “Black/Latinx.” If the patient is Black or Latinx, the system alerts the provider that the “[p]atient is from a racial or ethnic group with historically inequitable access to the Cardiology service; consider changing admission to Cardiology unless extreme census or overriding clinical reasons for General Medical Service.”²²⁷ (The protocol does not explain what happens if the patient is from a racial group other than White, Black, or Latinx, or has not been classified by race.) If the admitting physician wishes to continue admitting a Black or Latinx patient to the General Medical Service despite the alert, they must take time to enter a “free-text explanation” for their choice.²²⁸

The protocol designers discuss political backlash against the Healing ARC protocol, as well as technical limitations due to incomplete data.²²⁹ They nowhere, however, discuss potential legal obstacles or the role of law in crafting the protocol. No author of the protocol held a law degree.²³⁰

Perhaps reflecting the lack of legal input, the Healing ARC protocol contains multiple legal vulnerabilities. First, the authors assert the “impossibility of achieving equity by overlaying race-blind solutions onto a structurally racist system.”²³¹ They overlook, however, the important distinction between race-conscious approaches that address structural racism but do not classify individual patients by race and race-conscious approaches that assign benefits and burdens based on individuals’ self-identified race. Only the latter prompts strict legal scrutiny. Second, while the authors discuss and quickly reject a few alternatives that they describe as “race-blind,” such as checklists, they do not discuss a highly plausible alternative that avoids racial classifications: reducing or eliminating the role of patient requests. The authors attribute white patients’ greater representation on the cardiology service to the fact that “white patients . . . were perceived by providers to show greater preference for admission to the cardiology service than Black and Latinx patients . . . and to self-advocate more strongly when they did so.”²³² To mitigate outcome disparities, admissions decisions could be screened off from self-advocacy. Third, the protocol lumps Black and Latinx patients together, ignoring within-group and across-group differences, and entirely ignores other minority racial groups. Fourth, the use of racial classifications to constitute the Wisdom Councils imperils those councils’ legal legitimacy.²³³ Fifth, rather than merely informing physicians about options, the requirement to enter a free-text description creates a concrete financial incentive to direct patients to a particular medical service based on their race, given the cost and scarcity of physicians’ time.

Last, by emphasizing the reparative basis for the protocol to the exclusion of beneficent and egalitarian goals, the protocol invites the response that reparative measures must be targeted to compensate the victims of past intentional discrimination by the hospitals themselves. The protocol identifies a statistical disparity in admissions, but attributes it to differences in perceived self-advocacy. Given this explanation, it is unclear whether the statistical disparity constitutes proof of past intentional discrimination, as needed to satisfy strict scrutiny.²³⁴ This problem is buttressed by the fact that the patients whose admissions are influenced by the protocol may not be the same patients who experienced unequal admissions under prior policies.²³⁵ In contrast, courts have suggested that considering patients’ race in order to pursue better health outcomes for those patients will satisfy strict scrutiny.²³⁶ And while using patients’ race in order to address present disparities is more controversial, narrowing racial health disparities is widely recognized by state and federal decisionmakers as a legitimate aim. In contrast to these beneficent and egalitarian goals, attempting to redress historical injustice by nudging physicians to admit patients to the cardiology service based on their race stands apart from legally endorsed or countenanced uses of racial classifications in medicine.

Health systems housed within state universities, like the University of Virginia’s, should also be mindful of state policies restricting the consideration of race in university settings. In Texas, for instance, state higher education institutions may not, except as required by federal law, “give preference on the basis of race . . . color, ethnicity, or national origin to . . . a participant in any function of the institution.”²³⁷ Such a restriction would proscribe the Healing ARC’s use of race. Similar legislation has passed in Utah,²³⁸ and is being considered in other states.²³⁹

C. Allocation Under Scarcity

Unlike ordinary diagnosis and treatment decisions, choices about how to allocate scarce medical resources place different patients’ interests into direct conflict. In this respect, scarce resource allocation policies are more akin to the university and secondary-school admissions policies that the Supreme Court has described as “zero-sum.”²⁴⁰

Immediately prior to the SFFA decision, the COVID-19 pandemic necessitated policies for the allocation of scarce medical resources, including vaccines and antivirals. A few of these policies used individual race as an eligibility criterion. Some such policies were challenged via demand letters or administrative complaints. Most were later revised to remove the use of racial classifications.²⁴¹

New Hampshire’s vaccine allocation plan, which allowed residents to “qualify for an equity-plan vaccine by meeting one of ten criteria, including identifying as a racial or ethnic minority,”²⁴² was challenged in court. The plaintiff, however, lacked standing because residents of his area “regardless of their race or ethnicity . . . [did] not qualify for a vaccine through the equity plan,” and accordingly the plaintiff could not show that “but for the allegedly impermissible criteria, he would [have been] eligible to apply for a vaccine through the equity plan.”²⁴³ Although the merits were never reached, in their briefing, the governmental defendants were careful to emphasize that the equity plan’s aim was beneficent.²⁴⁴

Later, multiple plaintiffs challenged New York’s prioritization guidance for antiviral medications in court. Two challenges were dismissed for lack of standing because the plaintiffs did not show that they faced an imminent injury due to the operation of the allocation policy.²⁴⁵ Another was dismissed by agreement.²⁴⁶

Although the New York cases never reached the merits, they prompted members of both the Second Circuit panel and the Supreme Court to preview merits analyses. In a footnote, Second Circuit Judge Cabranes stated his belief that “government ‘guidance’ effectively directing health-care providers to prioritize the treatment of patients based on race or ethnicity may indeed present portentous legal issues if challenged by plaintiffs with standing.”²⁴⁷ Subsequently, while agreeing with the denial of certiorari, Justice Alito (joined by Justice Thomas) opined at length on the substance of the New York plan. In his view, “whether the Equal Protection Clause permits governments to use race or ethnicity as a proxy for health risk and therefore ‘prioritize the treatment of patients’ on that basis” is an “issue of ongoing importance.”²⁴⁸ He then stated, citing SFFA, that the Equal Protection Clause requires narrow tailoring of any governmental action “seeking to allocate benefits or burdens based on race or ethnicity,” and that “government actors may not provide or withhold services based on race or ethnicity as a response to generalized discrimination or as a convenient or rough proxy for another trait that the government believes to be ‘characteristic’ of a racial or ethnic group.”²⁴⁹ Given these limitations on governmental consideration of race, in Justice Alito’s view “New York’s general reference to ‘longstanding systemic health and social inequities’ would not have sufficed to allow the State to deny a person medical treatment simply because that person is viewed by the State as being a member of the wrong racial or ethnic group.”²⁵⁰ Rather, in his view, if “any government again resorts to racial or ethnic classifications to ration medical treatment, there would be a very strong case for prompt review by this Court.”²⁵¹ That only Justice Thomas joined Justice Alito’s statement, however, suggests that his perspective may not be shared by a majority of the Court.

Had the merits been reached, New York would have struggled to satisfy strict scrutiny because its policy did not examine plausible race-neutral alternatives and used an undifferentiated “BIPOC” priority that lumped together patients of different races.²⁵² The governmental defendants asserted that “no race-neutral alternative . . . would account for the medically proven fact that non-white race or Hispanic ethnicity is an independent risk factor for severe COVID-19 illness.”²⁵³ But they did not explain why “non-white race”—as opposed to some more specific category—was the appropriate category for purposes of epidemiological analysis. COVID-19 outcomes across Black, Hispanic, Asian-American, and Native American communities have been substantially different.²⁵⁴ The defendants also failed to show that New York seriously considered approaches used in other jurisdictions that did not rely on racial classifications.²⁵⁵ They asserted that “the fact that other States choose not to include certain risk factors in medical guidance does not preclude New York from making an independent judgment on the issue,”²⁵⁶ but they did not explain why New York arrived at the independent judgment that individually race-neutral alternatives used in other states were unworkable. To pass strict scrutiny, parties using racial classifications must provide a detailed explanation of why race-neutral alternatives would not suffice.²⁵⁷

Similarly, amici including the AMA, American Public Health Association, and Infectious Diseases Society of America filed a brief recounting numerous studies demonstrating correlations between race and more severe COVID-19 outcomes at various points during the pandemic, but citing no cases establishing the acceptability of using racial classifications.²⁵⁸ The brief identifies multiple drivers of poorer COVID-19 outcomes, such as living in segregated communities, lacking access to health care, subjection to poor working conditions, and exposure to racism.²⁵⁹ But it does not explain why prioritizing patients based on a binary BIPOC/non-BIPOC racial classification is the only workable way of identifying the patients at highest risk, compared to alternatives that focus on exposure to health and social inequities. As the amici recognize, their brief “should not be read as suggesting that race in and of itself is a genetic or biologic factor that causes medical risk.” Rather, they assert that “it is longstanding systemic health and social inequities that have contributed to increased prevalence of severe illness and death.”²⁶⁰ Prioritization approaches based on exposure to inequities, rather than self-identified race, were used in other jurisdictions.²⁶¹ Using self-identified BIPOC status as a proxy for exposure to health risk factors lumps together disparately situated individuals based on a racial classification.

New York’s COVID-19 antiviral allocation policy was distinctively vulnerable from a legal perspective because it used individual recipients’ race as a priority factor. In contrast, recent reforms to organ transplantation policies, as well as several COVID-19 vaccine and antiviral allocation policies,²⁶² have pursued beneficent and racially egalitarian goals without classifying individuals by race.²⁶³ As an example, recent reforms to the processes used to determine eligibility for kidney transplants have changed the equations used to assess kidney function. Kidney function is now determined using calculations that exclude transplant candidates’ race as an input variable.²⁶⁴ Some prior calculations, however, used race as an input variable.²⁶⁵ Unlike using race to allocate scarce organs, removing candidate race as an input variable presents no obvious legal problem, even if the objective is to mitigate racial disparities. Notably, even critics of affirmative action like the George Mason law professor Ilya Somin recognize that “[t]he new . . . formula for estimating kidney functioning is actually race-neutral, whereas the old one explicitly took race into account, in a way that penalized black transplant candidates.”²⁶⁶

In addition to removing race as an input variable, policymakers have also proposed reducing the waitlist priority former donors receive. It is unclear whether this proposal would advance either beneficent or egalitarian goals, given Somin’s point that such a policy would “predictably reduce the incentive to donate kidneys in the first place, thereby further exacerbating the organ shortage.”²⁶⁷ Nonetheless, Somin is mistaken to believe that reducing priority for donors is “clearly motivated by a desire to benefit one racial group (African-Americans) relative to others.”²⁶⁸ The policy’s goal is instead to improve African-Americans’ absolute health outcomes.

Additionally, Somin is mistaken to believe that “racially motivated policies are subject to special legal scrutiny and are presumptively unconstitutional unless the government can prove they would have adopted the same policy even in the absence of racial motivation.”²⁶⁹ Under present Supreme Court precedent, if a policy neither classifies individuals by race nor is adopted “because of” its adverse effects upon an identifiable group, it is not presumptively unconstitutional, even if it aims to narrow racial disparities. That precedent may change, as Section III.B will discuss. But it has not changed yet.

D. Public Health

Both beneficent and egalitarian uses of racial data are frequent in public health. Public health, however, starts in a stronger legal position than medicine because it makes decisions at the level of populations rather than individuals. Unlike decisions about individuals made in reliance on their race, use of racial demographics at the census block, zip code, or locality level has not prompted strict scrutiny in education cases.²⁷⁰ Nor does the pursuit of racial disparity reduction as a public health aim lead to strict scrutiny.²⁷¹

Occasionally, even expert commentators appear to confuse the use of race to classify individuals with its use at the population level. For instance, the New York Times quoted distinguished constitutional law scholar Erwin Chemerinsky as stating that because “race-based classifications must be subjected to strict scrutiny to ensure that the government has exhausted race-neutral remedies,” the “administration stands a better chance of keeping its environmental justice policies intact without being explicit.”²⁷² But strict scrutiny only applies when policies classify individuals by race. If they consider racial demographics at the level of a neighborhood or community, strict scrutiny does not apply. While Chemerinsky asserts that “the Supreme Court is very hostile to any attempt to use race as a basis for giving benefits,”²⁷³ its current precedents clearly distinguish individual racial classifications from community-level demographic information.

SFFA does caution against using other information to attempt to proxy for individual race. It warns that “universities may not simply establish through . . . other means the regime we hold unlawful today,” because “the prohibition against racial discrimination is ‘levelled at the thing, not the name.’ ”²⁷⁴ But the use of population-level demographic information, including information about community racial demographics, does not classify individuals by race in order to determine their access to a benefit. The use of individual racial classifications as an eligibility criterion is what SFFA held unconstitutional, not pursuit of racial diversity or efforts to narrow racial disparities. Public health efforts do not attempt to proxy for individual race: Rather, they use community-level information about racial demographics as an input into public health policies. Such policies might include where to place pop-up or mobile health services, or where to site parks, schools, or recreation centers.

E. Clinical Research

The use of individual participants’ race as an eligibility criterion for medical research is widespread. A search of the NIH’s clinical trial website reveals numerous U.S.-based studies currently recruiting participants based on individual race, including trials specific to African-American, Asian-American, and Latino/Hispanic patients or that stratify trial participants within these trials by race. Some of these trials seem to use self-identified race as a proxy for exposure to social drivers of health outcomes.²⁷⁵ Others use self-identified race as a proxy for genetic risk factors.²⁷⁶ But many simply describe race as a potentially predictive factor for health outcomes without explaining why race is being used.²⁷⁷

Many trials reference beneficent justifications for considering race, but some also reference egalitarian ones. For instance, a trial of the intervention “Black Girls Move: A Daughter/Mother Intervention to Prevent Obesity by Increasing Physical Activity and Improving Dietary Intake Among Black Adolescent Daughters” referenced two purposes: to “enhance health, lengthen life, and reduce illness and disability” and “to reduce population health disparities impacted by structural racism.”²⁷⁸

Unlike in educational contexts or other health care contexts, there appears to be little political controversy over clinical trials using individual race as an inclusion criterion or in data analysis. Indeed, both NIH and FDA recommend the consideration of individual participants’ race in trial design.²⁷⁹ Additionally, a bipartisan group of senators endorsed the goal of “sufficient representation of racial and ethnic minorities in clinical trials.”²⁸⁰

Despite its widespread use, the use of individual race in clinical trials stands in tension with the Supreme Court’s increasingly colorblind commitments.²⁸¹ The Supreme Court has not recognized optimizing the medical benefit of clinical trials as a compelling interest for purposes of strict scrutiny, although lower courts have approved the use of racial classifications for educational research.²⁸² More importantly, individually race-neutral alternatives are likely to exist for many trials. Rather than using race as a proxy, trial designers could enroll for the characteristic for which race is proxying.²⁸³ If such alternatives were workable and comparably effective, the consideration of individual participants’ race would not satisfy strict scrutiny as the Supreme Court conceives it.

Some trials already aim to enroll participant populations with a particular racial makeup without excluding individual participants because of their self-identified race. For example, one study of “a program to promote awareness of and access to pre-exposure prophylaxis (PrEP) for HIV prevention” enrolled only residents of the Durham Housing Authority (DHA).²⁸⁴ The investigators explained that “[t]he vast majority of DHA residents are African American, so the investigators anticipate recruiting participants who are majority African American without explicitly excluding prospective participants on the basis of race.”²⁸⁵

III. The Future

Antiracist medical policies are likely to come into conflict with legal enactments as interpreted by an increasingly colorblind Supreme Court, as well as lower court judges who endorse colorblindness. Organizations are already seeking to sue medical decisionmakers who make race-aware decisions or employ racial classifications.²⁸⁶ In order to pursue race-aware beneficent, egalitarian, or reparative goals, health professionals and decisionmakers must understand how the Supreme Court approaches the use of race.

A comprehensive view of the Supreme Court’s present precedent on race reveals some important themes. First, individual racial classifications are treated very differently from policies that are motivated by, or consider, race without classifying anyone by race. Racial classifications must always satisfy strict scrutiny. Strict scrutiny requires, at a minimum, that the use of the classification be narrowly tailored to realize the compelling interest. The cases applying the strict scrutiny standard to racial classifications emphasize the need for consideration of race neutral alternatives, and the importance of avoiding over- and under-inclusiveness.

Second, a policy that considers race without classifying anyone by race does not have to satisfy strict scrutiny. Absent an individual racial classification, strict scrutiny is only triggered if some additional condition is met. Such conditions include an intended adverse impact on a racial group or the use of a classification that is a pretext for racial discrimination or a proxy for a racial classification. Strict scrutiny may also apply if a policy aims at racial balancing or if it is motivated exclusively or predominantly by a racial purpose.

This Part will consider two distinct ways that medicine might pursue race-aware beneficent, egalitarian, and reparative goals in light of the precedents detailed in Part I. The first would use individual racial classifications and argue that these classifications satisfy strict scrutiny because they are narrowly tailored to realize a compelling interest. The second would avoid individual racial classifications in favor of policies that advance beneficent, racially egalitarian, or racially reparative ends via facially race-neutral means.

A. Surmounting Strict Scrutiny

As explained in Section I.A, an important takeaway for health policy design is that making decisions—including health-related or health-promoting decisions—on the basis of individuals’ race inevitably triggers strict legal scrutiny. Before adopting a policy that treats individuals differently based on their race, health policymakers must be certain that the use of racial classifications is necessary and narrowly tailored to realize a compelling interest.

Health professionals at times have assumed that they are equally free under the law to use any individual characteristic that is predictive of health outcomes.²⁸⁷ Legally, they are not: “Treating someone differently because of their skin color is not like treating them differently because they are from a city or from a suburb.”²⁸⁸ Even health professionals who have served in high office often dismiss legal complexities in cavalier fashion or conflate different types of race-conscious policies. For instance, the former head and present Chief Medical Officer of New York City’s health department asserted that “[l]egal counterarguments to addressing racism as a risk factor are . . . dubious” because “[t]he authors of the Civil Rights Act of 1964 recognized the need to incorporate race into remedies for racial inequity, ensuring the legality of race-conscious approaches.”²⁸⁹ But the Supreme Court’s justices are uninterested in what the authors of the Civil Rights Act recognized—they are interested in the text they produced, and at least two of these justices read that text to proscribe all use of racial classifications.²⁹⁰ Moreover, even if the Civil Rights Act were read to permit all race-conscious approaches, the Equal Protection Clause and state laws might still restrict their use. The Chief Medical Officer and colleagues also believe that because “legal authority supports programs like the Minority and Women-owned Business Enterprise Program”²⁹¹ in New York City, consideration of race in health would pass muster. This overlooks the legal difference between minority business enterprise programs, which remedy identified past governmental discrimination and mirror carefully tailored federal programs,²⁹² and novel policies seeking to mitigate racial health inequity broadly construed. Likewise, the Chief Medical Officer and colleagues fail to identify the crucial legal distinction between programs that counteract racism’s harmful effects without using racial classifications and those that employ such classifications.

Health programs will, however, fare better than educational ones with respect to SFFA’s requirement that the interest justifying racial classifications must be “sufficiently coherent for purposes of strict scrutiny” and capable of being “subjected to meaningful judicial review.”²⁹³ SFFA faults admissions policies for using racial classifications in pursuit of “engaged and productive citizens, sufficiently enhancing appreciation, respect, and empathy, or effectively train[ing] future leaders.”²⁹⁴ In contrast, health policies typically aim at outcomes like deaths averted, lives extended, or hospitalizations avoided (for beneficent aims); reductions in racial disparities with respect to these outcomes (for egalitarian aims);²⁹⁵ or improvements in a historically disadvantaged individual’s or group’s health outcomes (for reparative aims). These outcomes are measurable. If medical programs fail strict scrutiny, it will be because they do not realize a compelling interest or are not narrowly tailored to do so, not because they are nonassessable.

The proscription against racial stereotyping, meanwhile, lacks clear relevance in the medical context. SFFA’s invocations of a rule against stereotyping emphasize the wrongness of equating a person’s race with their beliefs, choices, or achievements.²⁹⁶ Medical treatment does not typically assess a person’s beliefs or viewpoint, but rather whether a given intervention will improve or worsen their bodily functioning. Classifying patients by race in order to improve their health does not send any stigmatic message regarding the connection between race and belief.

1. Beneficent Justifications

While SFFA does not explicitly include health as a compelling interest, health improvements should pass the compelling interest test. The Supreme Court in dicta has recognized health as a compelling interest justifying the use of racial classifications, as have appellate courts.²⁹⁷ Additionally, saving lives and improving health care have been recognized as compelling interests that satisfy strict scrutiny in other contexts.²⁹⁸

To be narrowly tailored to realize a compelling interest, as Part I explains, two criteria are most important. First, race-neutral alternatives must have been seriously considered in good faith and found unworkable. Second, the racial classification must be neither an overinclusive nor an underinclusive means of improving health.

Using racial classifications in service of generally beneficent aims—as opposed to racially egalitarian or reparative aims—exemplifies the use of “a racially conscious means to further a nonracial goal.”²⁹⁹ For this and other reasons, narrow tailoring will be difficult, but not impossible, for beneficent health programs to satisfy. The programs most likely to succeed are those that make evidence-based recommendations, such as recommending when to screen for cancer, for individual patients. Even such a program, however, would only pass muster under a pragmatic rather than formalist understanding of the prohibition on over- and under-inclusiveness, and a willingness to rule out some race-neutral alternatives as too costly. An individual’s race itself is never a direct health risk factor.³⁰⁰ Rather, race may be correlated with adverse health outcomes because race is correlated with subjection to racism, which in turn worsens health outcomes.³⁰¹ Or race may be correlated with health outcomes because it correlates with some biological factor, such as a genetic or epigenetic variant that affects health outcomes.³⁰² Rather than using an individual’s race as a proxy for their subjection to racism,³⁰³ health decisionmakers could use alternative proxies that eschew racial classifications, such as disadvantage indices. Disadvantage indices have the important advantage of recognizing that not all members of a societally or governmentally defined racial group have been equally harmed by racism. Likewise, rather than using race as a proxy for biological differences, health decision-making could use some other proxy, such as self-reported ancestry, or could directly test for the biological risk factor.

Medical experts recognize that race is an imperfect proxy for underlying risk factors.³⁰⁴ As such, the use of racial classifications to realize beneficent medical goals is inevitably overinclusive or underinclusive. The question is how much imperfection in fit is acceptable. Some courts have insisted that none is.³⁰⁵ But others have accepted some degree of over-inclusiveness.³⁰⁶

Programs that do not compare patients might also hope that standing doctrine will protect them against legal challenges, since they do not injure third parties. But if a patient is misdiagnosed by health professionals using a policy that classifies them by race, they would have standing to sue for discrimination even if the policy was designed to further a compelling interest.

Programs that compare patients to one another based on their race are likely to face greater difficulty. SFFA’s rule against uses of race as a negative appears particularly applicable to strictly zero-sum policies such as the allocation of scarce medical resources. It also applies to policies like the racially classificatory nudges used in the Brigham and Women’s policy, about which the court could also ask whether “in its absence, members of some racial groups would be admitted in greater numbers than they otherwise would have been”³⁰⁷—here, admitted to the cardiology service rather than to Harvard or UNC.

Notably, the zero-sum status of higher education admissions is more debatable than the zero-sum status of scarce medical resource allocation. Whether a policy is zero-sum requires considering its effects both on access to an opportunity and on the value of that opportunity. In some circumstances, a policy that reduces access to an opportunity may increase the value of that opportunity, rendering it not zero-sum. For instance, a policy that increases racial diversity in medical schools may lower the odds of admission for members of some racial groups, yet the increase in diversity may increase the value of education for admittees. Even members of racial groups whose odds of admission would decrease might rationally prefer reduced odds of receiving a more valuable education.³⁰⁸ A similar argument could be made for the use of race as an eligibility criterion for clinical trials if its use leads to more socially valuable data. By contrast, policies that classify patients competing for a scarce resource by race are zero-sum, because the patients who lose out do not stand to benefit from the classification—the classification merely benefits others at their expense.

Medical uses of race for beneficent purposes might attempt to invoke technical expertise in their support. SFFA and other cases counsel caution in such efforts. SFFA rejects the idea that “what experts and evidence tell us is required (as a matter of social science) to solve for pernicious race-based inequities”³⁰⁹ should guide or even affect constitutional analysis.³¹⁰ Relatedly, even if the distinctive nature of physicians’ professional role debars them from legally challenging concordance requests,³¹¹ this role-based restriction would not extend to other professionals such as nurses or health aides. Ultimately, beneficent justifications can often satisfy the compelling interest prong of strict scrutiny, but whether they can satisfy narrow tailoring depends on how punctiliously narrow tailoring is interpreted.

2. Racially Egalitarian Justifications

Many governmental programs explicitly aim to reduce racial health disparities.³¹² It is unclear, however, whether this governmental aim constitutes a compelling interest that could justify the use of racial classifications in health decisionmaking, as opposed to merely a permissible aim that can be pursued by individually race-neutral means. Some argue that Supreme Court precedent denies that reducing racial health disparities is a compelling interest.³¹³ Others argue the reverse: that racial disparity reduction is a compelling interest.³¹⁴

The ambiguous status of mitigating racial health disparities under the compelling interest test parallels the more general, longstanding ambiguity regarding whether race-conscious aims other than remediating the effects of the government’s own past discrimination can constitute a compelling governmental interest. The Supreme Court has been clear that remediating the effects of societal discrimination is not a compelling interest.³¹⁵ In contrast, it has also held—though less recently—that racial diversity and the avoidance of racial isolation are compelling interests.³¹⁶ Racial diversity and reduced racial isolation, however, both directly aim at creating spaces where individuals of different races can interact. They remediate participatory racial injustice.³¹⁷ In contrast, mitigating racial health disparities, understood as differences in health outcomes that track disadvantage or group membership, seems closer to remediating disparities in income, education, or employment—distributive racial injustice rather than exclusion from social participation.³¹⁸ Courts deny that remediating disparate societal outcomes, unless they stem from governmental discrimination, is a compelling government interest at least for the purposes of strict scrutiny.³¹⁹

Even if the use of racial classifications to reduce racial health disparities can satisfy the compelling interest test, such use would also have to be narrowly tailored. Encouragingly, using racial classifications in service of the racially egalitarian goal of mitigating disparities has better prospects of avoiding over- and under-inclusiveness than using racial classification in service of beneficent goals. Whereas patient’s race is only a proxy for their medical risk, it is directly relevant to whether a racial disparity exists.³²⁰ Using racial classifications to mitigate racial disparities uses racial means to pursue racial rather than nonracial ends.

The other narrow tailoring factors, however, would still need to be satisfied. Even though the goal of mitigating racial disparities explicitly references race, individually race-neutral alternatives, such as disadvantage indices, would nevertheless need to be seriously considered. Programs would need to be durationally limited. Any burdens on third parties would need to be minimal.

3. Racially Reparative Justifications

The governmental use of racial classifications to rectify the effects of past illegal discrimination remains acceptable even after SFFA.³²¹ The past discrimination in question, however, must be intentional discrimination either initiated or participated in by the governmental actor seeking to remedy the discrimination.³²² The most demanding aspect of this test is that the rectifying actor must have themselves been the past discriminator: In contrast, the beneficiaries need not have personally suffered the discrimination in question, nor do third parties burdened by the remedial use of classifications need to have profited from the discrimination.³²³

An example fitting this rationale is the United Network for Organ Sharing’s recent directive that patients unfairly disadvantaged by past racially classificatory eligibility calculations should be restored in the transplant queue to where they would have been but for the racially classificatory policy.³²⁴ This guidance fits perfectly with SFFA’s approval of reparative policies that restore the status quo that would have existed without discrimination.³²⁵ Additionally, even though this policy may set back the interests of third parties also waiting for transplants, it does not classify third parties by race.³²⁶

It is less clear whether a reparative policy initiated by a past discriminator may exclude third parties from a health program, or deprioritize them for health services, on the basis of their race. No Supreme Court or appellate cases discuss the remedial allocation of health services, even in dicta. Rather, the remedial cases all concern the allocation of employment, contracting, or educational opportunities.³²⁷ Because of the high stakes of access to health care, however, delaying or denying a patient’s access to scarce treatment on the basis of their race would likely be regarded as unacceptably imposing “the entire burden of achieving racial equality on particular individuals, often resulting in serious disruption of their lives.”³²⁸ Put another way, the high stakes of health care decisionmaking means that any direct health burden on “innocent” third parties inflicted via a racial classification will typically be viewed as too severe to pass muster. This conclusion presents normative problems, since the stakes are equally high for the third parties who would be excluded and the racial minority beneficiaries who would be included.³²⁹ But it is an implication of current precedent’s decision to regard racial classification as a distinctively objectionable means of achieving a policy outcome.³³⁰

B. Seeking Rational Basis Review

If health decisionmakers avoid the use of racial classifications, they can implement policies pursuing beneficent, racially egalitarian, or racially reparative goals without facing strict judicial scrutiny.

1. Design Options

Building on the work of decisionmakers in other areas such as primary and secondary education, health decisionmakers could use geographic indices of disadvantage to pursue public health aims or to allocate scarce medical resources. After Parents Involved established that policies treating individuals differently on the basis of their race were subject to strict scrutiny, many school districts adopted geographically-based assignment policies that considered factors like local disadvantage.³³¹ Both before and during the COVID-19 pandemic, public health decisionmakers likewise used geographic indices of disadvantage to prioritize patients for access to interventions, such as antiviral therapies and vaccines.³³² These proposals were defended on both beneficent and egalitarian grounds, since disadvantaged patients were at greater risk of severe COVID-19 outcomes.³³³

Other health policies have prioritized different factors that correlated with disadvantage. For instance, some have prioritized workers in risky jobs.³³⁴ Others have prioritized Medicaid recipients.³³⁵ Yet others have prioritized patients who become ill earlier in life.³³⁶ All these factors correlate with disadvantage. Considering them will narrow racial disparities (an egalitarian goal) and often advance beneficent and reparative goals as well.

Additionally, policies that consider Native American status, even as an individual classification, are not subject to strict scrutiny under current precedent because Native Americans are viewed as a political group rather than a race.³³⁷ Policies that classify individuals by Native American status need only satisfy rational basis scrutiny.

Institutions training professionals could also design policies in light of the SFFA majority’s concession that “nothing in this opinion should be construed as prohibiting universities from considering an applicant’s discussion of how race affected his or her life, be it through discrimination, inspiration, or otherwise.”³³⁸ For instance, medical school or residency admissions could consider applicants’ personal experience overcoming racial health disparities or interest in working to mitigate such disparities.

2. Best Practices

The Supreme Court has been clear that programs seeking to address racial disparities without classifying anyone by race do not trigger strict scrutiny merely because their programmatic goals are race-aware. Careful program design, however, can further reduce the risk of strict scrutiny, whereas incautious program design could prompt strict scrutiny even of a program that does not classify individuals by race.

First, health policy designers and implementers should avoid conceiving of race-conscious policies that classify no individual by race as “proxies” or “workarounds” for policies that impermissibly classify individuals by race.³³⁹ Race-aware policies should instead be justified by their potential to realize legitimate goals that cannot be cast as individual discrimination. For instance, arguing that “[c]ommunity colleges include a high proportion of premedical students from diverse backgrounds whose lived experience will make them excellent physicians” is legally preferable to contending that “[i]n this post-affirmative action era, community college attendance could be a proxy for race and ethnicity in medical schools admissions.”³⁴⁰

Second, race-aware policies should avoid designs that could be interpreted as having been adopted because of their adverse impact on an identifiable group. Designers should also be aware that in zero-sum contexts, some judges appear willing to view a race-aware policy with inevitable adverse impact as equivalent to a race-aware policy with discriminatory intent.³⁴¹ Some have gone further to suggest that race-conscious intent always triggers strict scrutiny.³⁴² But this is not the law at present.

Third, when race-aware policies aim at racially egalitarian—and thus race-conscious—rather than purely beneficent goals, they are in a stronger position if racial disparity reduction is not the policy’s sole or predominant purpose. Even though no case law squarely holds that strict scrutiny applies whenever race-aware motives predominate, some of the redistricting cases draw this distinction.³⁴³ Racial disparity reduction also must avoid shading into the proscribed practice of “racial balancing.” The Supreme Court’s clearest statement of what racial balancing involves is Fisher I, which concludes that a university may pursue “the educational benefits that flow from student body diversity,” but may not “define diversity as some specified percentage of a particular group merely because of its race or ethnic origin.”³⁴⁴ While Fisher I concerns education rather than health, it presents the question of whether mitigating racial health disparities is akin to pursuing racial diversity. The wide adoption of racial disparity reduction in federal and state policies suggests that racial disparity reduction is not inherently objectionable racial balancing. But it leaves open whether racial disparity reduction must be viewed as a means to realizing some other benefit, just as racial diversity is an acceptable goal not for its own sake but because of the educational benefit it produces. The goal of reducing racial health disparities, however, need not be improving population health—it could instead be some other end such as equal opportunity, civic vitality, or public trust.³⁴⁵ Health systems could also take a more cautious tack by seeking merely to eliminate serious health shortfalls rather than eliminating all racial health disparities.³⁴⁶ This approach would mirror the shift away from the broader pursuit of racial diversity toward the achievement of a “critical mass” or the avoidance of racial isolation.³⁴⁷

Taking these precautionary measures should be amply sufficient to ensure that race-aware health policies comply with present Supreme Court precedent. Recently, the Supreme Court faced one case inviting them to revisit that precedent. In Coalition for TJ, the petitioners suggested—albeit in “coy” fashion³⁴⁸—that efforts to mitigate racial disparities in the demographics of a magnet high school should trigger strict scrutiny even if pursued without classifying any student by race.³⁴⁹ After petitioners prevailed in the district court, the Fourth Circuit reversed in a 2–1 decision and remanded with instructions to grant the governmental defendants summary judgment.³⁵⁰ The Supreme Court initially declined to vacate the Fourth Circuit’s stay of a preliminary injunction authorized by the district court, although Justices Alito, Gorsuch, and Thomas would have granted the petitioners’ motion.³⁵¹ Subsequently, the Court denied certiorari over the disagreement of Justices Alito and Thomas.³⁵² This denial of certiorari leaves policies that seek to mitigate racial disparities without classifying anyone by race safe—for now.

C. Anticipating Statutory and Regulatory Limits

Congress or administrative agencies may also issue additional guidance constraining the use of racial classifications in the delivery of health care. In response to states’ use of racial classifications in allocating COVID-19 therapies, Members of Congress introduced the “Medical Equality and Discrimination Stoppage Act” (MEDS Act) which would prohibit the Department of Health and Human Services from granting “a preference to . . . any person or group based in whole or in part on race, color, ethnicity or national origin, in connection with . . . any . . . program or activity carried out by the Secretary.”³⁵³ Similar provisions would apply to any recipient of HHS funds. This legislation did not progress beyond referral to subcommittees.³⁵⁴

Arizona’s Secretary of State also submitted a petition for rulemaking to HHS asking that the agency “issue a rule prohibiting the allocation of medical therapies based on race and ethnicity, or issuance of guidance recommending the same, unless the action satisfies the strict scrutiny test.”³⁵⁵ Unlike the MEDS Act, Arizona’s petition did not categorically reject the medical use of racial classifications. Rather, Arizona asserted that beneficent justifications are valid if supported by evidence while egalitarian or reparative ones are not. The Arizona Secretary of State claimed that “the Constitution demands that HHS save as many lives as possible, all else being equal, rather than allowing some to die to achieve a ‘better’ racial mix in the death count,” and that “the Equal Protection Clause does not permit HHS to sacrifice lives of individuals upon the altar of equity.”³⁵⁶ In contrast, he conceded that HHS “has authority to address biological distinctions that put individuals at greater risk.”³⁵⁷

Though HHS has not taken formal action on Arizona’s petition, it has issued other guidance constraining the use of race. Its Office for Civil Rights has issued nonbinding guidance reiterating that health programs receiving federal financial assistance “generally cannot distinguish among individuals on the basis of race . . . either directly or indirectly, in the types, quantity, quality or timeliness of program services, aids or benefits that they provide or the manner in which they provide them.”³⁵⁸ HHS provides the example that “[a] state or local public healthcare authority or federally assisted health care provider that establishes a policy or procedure that provides priority to vaccines based on race . . . without a legitimate non-discriminatory reason for doing so would violate Title VI and Section 1557.”³⁵⁹ HHS’s guidance does not explain what would constitute a legitimate non-discriminatory reason for a race-based vaccination priority. It does, however, state that even if a non-discriminatory reason exists, the race-based policy cannot continue if there are “alternatives that would achieve the same objectives but without discriminating on the basis of race.”³⁶⁰

Courts have also recently questioned the legitimacy of pursuing health equity and mitigating health disparities, albeit only in specific regulatory contexts. In Colville v. Becerra, a physician and several states sued the Secretary of Health and Human Services challenging the Center for Medicare and Medicaid Services’s (CMS) promulgation of a “Create and Implement an Anti-Racism Plan” clinical practice improvement activity.³⁶¹ Medical practices participating in the activity will “implement an anti-racism plan using the CMS Disparities Impact Statement or other anti-racism planning tools,” which will include a “clinic-wide review of existing tools and policies . . . to ensure that they include and are aligned with a commitment to anti-racism and an understanding of race as a political and social construct, not a physiological one.”³⁶² CMS believed this activity “has the potential to improve clinical practice or care delivery and is likely to result in improved outcomes” because it “supports clinicians in identifying health disparities and implementing processes to reduce racism and provide equitable quality health care.”³⁶³ Represented by the same attorney who argued for the plaintiffs in SFFA, Plaintiffs complained that CMS’s action both diverges from the statutory mandate that improvement activities concern “clinical practice or care delivery” and promotes unlawful “race-based decisionmaking in medical care.”³⁶⁴ Even though CMS’s action contains no facial racial classification, Plaintiffs’ complaint survived a motion to dismiss because the CMS language does not define what “anti-racism” constitutes, and there is no evidence that it “rejects the race-based decisionmaking that is alleged to be promoted.”³⁶⁵

Both the state plaintiffs and federal defendants subsequently moved for summary judgment, with plaintiffs asserting that CMS’s rule forces “clinicians to prioritize certain racial and ethnic minorities.”³⁶⁶ They claim that because CMS’s Disparities Impact Statement mentions “health equity” and “health disparities,” guides clinicians to identify which populations to prioritize, and suggests that clinicians “stratify[] measures and health outcomes by race.”³⁶⁷ Similarly to Arizona’s petition, the Colville plaintiffs view physiological differences associated with race as medically relevant, but not social ones.³⁶⁸ Most recently, the states’ motion was denied for lack of standing, with the court observing that “[a]lthough their laws preclude various forms of racial discrimination, State Plaintiffs have not pointed to any competent summary judgment evidence showing how the Anti-Racism Rule or the Disparities Impact Statement threatens actual or imminent discrimination in violation of those laws.”³⁶⁹ However, the court permitted the states to proceed with discovery to attempt to establish standing.³⁷⁰

Plaintiffs’ paranoid cherry-picking of text from the Disparities Impact Statement belies the unexceptionable nature of CMS’s proposal. Stratifying measures and health outcomes by race³⁷¹ merely means tracking outcomes by race as well as overall outcomes; CMS has reported race-stratified outcomes since 2015, across multiple presidential administrations.³⁷² The Colville plaintiff states encourage consideration of these very disparities in their own laws.³⁷³ The Impact Statement’s only other reference to race is in its first paragraph, which observes that “[t]his tool can be used by all health care stakeholders to achieve health equity for racial and ethnic minorities, people with disabilities, sexual and gender minorities, individuals with limited English proficiency, and rural populations.”³⁷⁴ There is no obvious connection between the pursuit of health equity and the Colville plaintiffs’ imaginings that patients will be denied services or discriminated against based on their race.³⁷⁵ Rather, the statement gives as its example that stakeholders can “[i]mplement a Language Access Plan for patients with limited English proficiency.”³⁷⁶ Such a plan would involve no denial of access or “discrimination against populations not prioritized,”³⁷⁷ because patients and populations are not being comparatively prioritized for access to a scarce resource. Rather, health care stakeholders are simply identifying areas of opportunity based on data on disparities and pursuing improvement in those areas.³⁷⁸ Such improvements can be readily achieved without classifying any patient by race: For instance, a clinical practice could revise policies that inappropriately classify patients by race,³⁷⁹ or could address root causes of disparities.

Another district court has regarded “health equity” as an impermissible factor for CMS to consider in a different context: when deciding whether a Medicaid demonstration project could continue. The court in Georgia v. LaSure concluded that “health equity, by which the Agency apparently means equal health outcomes across various demographic populations, is not a statutory factor.”³⁸⁰ Despite health equity being referenced only once in CMS’s decision, the court faulted CMS for considering “the physical health of racial and ethnic minority groups, who already experience disparities in health outcomes” and recognizing that “studies have also found that premium policies can exacerbate health disparities.”³⁸¹ The court rejected consideration of health equity even though the Medicaid statute describes medical assistance to people with low “income and resources” as a statutory objective, and racial disparities in income and wealth are extensively documented.³⁸²

In both cases, CMS’s lack of explicit definitions created an opening for plaintiffs to advance negative interpretations of terms like antiracism and health equity. The Colville plaintiffs equated antiracism with discriminatory treatment of patients based on their race. They referenced antiracism scholar Ibram Kendi’s endorsement of “present discrimination” and claimed that the CMS “[r]ule itself suggests that ‘doctors should engage in Antiracist discrimination to prioritize group disparities over individuals’ needs while providing care’ ”—even though their quoted language comes not from the “rule itself” but from a 2022 law review article by a Heritage Foundation fellow.³⁸³ But neither is the arbiter of what antiracism means,³⁸⁴ nor is either cited in HHS’s rule. It would nevertheless be preferable to offer a clear definition of antiracism and provide examples of appropriate antiracist policy changes. Likewise, rather than allowing the speculative inference in LaSure that health equity “apparently means equal health outcomes across various demographic populations,”³⁸⁵ HHS could have explicitly defined health equity in terms of fair processes and opportunities for health improvement.³⁸⁶ Notably, in their briefing, HHS makes clear that “[h]ealth equity . . . means ‘the attainment of the highest level of health for all people, where everyone has a fair and just opportunity to attain their optimal health,’ ” and that “[h]elping underserved populations achieve optimal health does not require a reduction in the services to other populations.”³⁸⁷

Importantly, opportunity-based definitions of health equity also align with public endorsement of equal opportunity and rejection of racial discrimination.³⁸⁸ In contrast, the public questions equality of outcome.³⁸⁹ Accordingly, health equity advocacy may be strategically ill-served to lament those health interventions that improve overall outcomes, including for racial minorities, but widen health gaps,³⁹⁰ and better-served to emphasize that health disparity data illuminates opportunities for overall health improvement.³⁹¹

Conclusion

The Supreme Court has adopted an increasingly restrictive regime for race-aware actions. A successful racially classificatory policy not only has to further a compelling interest but also satisfy the exacting requirements of narrow tailoring. Satisfying these requirements requires seriously considering alternative pathways to realizing the compelling interest that do not classify individuals by race. It also requires painstakingly crafting the racial classifications contained in the policy to ensure that they contain, as far as possible, all and only the individuals whose differential race-based treatment is necessary to realize the compelling interest. For policies that do not classify by race, decisionmakers have a freer hand. But even for these policies, courts seem willing to invalidate certain race-conscious motives, such as motives that predominantly or adversely affect certain races.

In the current legal environment, racial health inequities cannot be effectively addressed without recognizing present legal limits and how those limits might evolve. A policy cannot prevent future Black children from dying of asthma if it is enjoined from taking effect or withdrawn once challenged. And it can cause needless future deaths if it prompts new limits. Yes, professional ethics sometimes requires lawbreaking.³⁹² But naïve policy design should not be excused as principled civil disobedience.

Awareness of legal constraints on the pursuit of racial health equity should not be equated with endorsement of these constraints,³⁹³ nor with a belief that present constraints are permanent. Health professionals and systems can and should support the enactment of laws and regulations that advance racial justice. But that work must begin with an understanding of legal precedent as it stands today. This Article offers such an understanding.

* Associate Professor, University of Denver Sturm College of Law. JD, PhD, Stanford University. Thanks to Rebecca Aviel, Bernard Chao, Alan Chen, Julie Cramer, Ian Farrell, Tim Holbrook, Elizabeth Jordan, Sam Kamin, Kevin Lynch, Viva Moffat, Marty Katz, Nancy Leong, Wyatt Sassman, Laurent Sacharoff, Sarah Schindler, Michael Siebecker, Michael Sousa, and Zahra Takhshid for helpful comments, and to the University of Wisconsin for inviting me to present an early version of these ideas at their 14th Annual Bioethics Symposium, Toward an Anti-Racist Bioethics. I am grateful to Dani Fishman and Nikki Musco for research assistance.