KLEINFELD, Senior Circuit Judge

Gregg Zucker (argued) and Victoria Niewrzol, Affeld Grivakes Zucker LLP, Los Angeles, California; Tara Burd and Benjamin Pavone, Pavone & Fonner, San Diego, California; Matthew B. Pavone, Law Offices of Matthew B. Pavone, Novato, California; Frederik Spiess and Edward Burns, Burns & Schaldenbrand, Oceanside, California; David Elliot, Law Offices of David Elliot, San Diego, California; for Plaintiffs-Appellants Smith and Gregge, et al.

Greg W. Garrotto (argued), Law Offices of Garrotto & Garrotto, Los Angeles, California, for Plaintiff-Appellant Hines.

Milin Chun (argued), Brian M. Bush, and Raymond P. Boucher, Boucher LLP, Woodland Hills, California; Ian Wallach and Jason Feldman, Feldman & Wallach, Santa Monica, California; Mark Ozzello, Arias Ozzello & Gignac, Los Angeles, California; for Plaintiffs-Appellants/Cross-Appellees Jackson, et al.

Jay Russell (argued), Supervising Deputy Attorney General; Kevin A. Voth, Martine D. Agostino, and Maureen Onyeagbako, Deputy Attorneys General; Jon S. Allin and Thomas S. Patterson, Supervising Attorneys General; Jonathan L. Wolff, Senior Assistant Attorney General; Office of the Attorney General, San Francisco, California; for Defendants-Appellees/Cross-Appellants.

Kristina Doan Gruenberg and Susan E. Coleman, Burke Williams & Sorensen LLP, Los Angeles, California, for Defendants-Appellees Igbinosa and Winslow.

Before: Andrew J. Kleinfeld and Sandra S. Ikuta, Circuit Judges, and Rosanna Malouf Peterson, District Judge.

KLEINFELD, Senior Circuit JudgeInmates in several California state prisons were exposed to a heightened risk of getting Valley Fever, so they sued state officials for money damages under 42 U.S.C. § 1983. The inmates claim that exposing them to a heightened risk of getting Valley Fever was cruel and unusual punishment in violation of the Eighth Amendment. African-American inmates add a challenge under the Equal Protection Clause of the Fourteenth Amendment. They claim that because African-American inmates were particularly likely to get Valley Fever and suffer serious consequences, they should have been segregated from the prisons with the highest infection rates. In each of the four cases on appeal, we hold that the state officials are entitled to qualified immunity.

FACTS

A. The Federal Receiver

For years, inmates in California state prisons have claimed that the state violates the Eighth Amendment by failing to provide sufficient medical care. Many inmates have sued. In 2002, California signed a consent decree in one such case, Plata v. Davis . As part of that decree, California promised to implement specific procedures to ensure that inmates statewide received constitutionally adequate medical care. But the state did not satisfy the terms of the decree, so in 2006 the Plata district court appointed a federal Receiver. The court conferred on the Receiver "all powers vested by law in the Secretary of the [California Department of Corrections and Rehabilitation] as they relate to the administration, control, management, operation, and financing of the California prison medical health care system." The court concurrently "suspended" the Department of Corrections and Rehabilitation's exercise of those powers "for the duration of the Receivership." The Receiver has filed papers with the Plata district court, and the district court has entered orders to improve medical care.

Therefore, since 2006, state officials have made decisions about prison medical care while under the control of a federal Receiver, appointed by a federal district court to ensure compliance with the Eighth Amendment. This case challenges how those state officials responded to Valley Fever outbreaks in several prisons in the Central Valley of California, despite the Receiver's control.

B. Valley Fever

Valley Fever is a disease caused by inhaling certain fungal spores. The spores, which live in dry soil, are common in much of the southwestern United States. Millions of people live where the spores are common, and tens of thousands of people are infected each year. Two-thirds of infections are reported in Arizona. One-fourth are reported in California. The rest are typically reported in Nevada, Utah, New Mexico, and Texas.

Once someone has been infected with the fungal spores, they are immune from future infections. But infections affect different people in different ways. About 60% of infected people do not develop any symptoms. Another 30% develop only mild flu-like symptoms (such as fever, cough, rash, headaches, and muscle aches) that usually go away after a few weeks. But around 10% of people develop a severe case of Valley Fever. About 8% of infections lead to a severe respiratory disease. And 1–5% of infections spread from the lungs to other parts of the body, a serious condition known as "disseminated cocci." Patients with disseminated cocci can be effectively treated, but they cannot be cured. Many disseminated cocci patients need expensive treatment for the rest of their lives to prevent their symptoms from recurring. In rare cases, such as when disseminated cocci spread to the brain and are not effectively treated, Valley Fever is fatal.

Some groups of people have an above-average risk of experiencing severe symptoms or developing disseminated cocci. One risk factor is having an underlying medical condition, such as HIV, diabetes, or heart disease. Another risk factor is being on a medication that suppresses the immune system, such as chemotherapy. Adults over 55 and pregnant women are at a greater risk. Men are more likely than women to develop disseminated cocci. And for unknown reasons, people of African and Filipino descent are several times more likely to develop disseminated cocci than are people of other racial or ethnic backgrounds.

C. Valley Fever in California Prisons

In 2005, California prison officials noticed a "significant increase" in the number of Valley Fever cases among prisoners. The federal Receiver asked the California Department of Health Services to investigate the outbreak at Pleasant Valley State Prison, the prison with the highest infection rate. After its investigation, the Department of Health Services issued a report in January 2007. It stated that Pleasant Valley State Prison had 166 Valley Fever infections in 2005, including 29 hospitalizations and four deaths. The infection rate inside the prison was 38 times higher than in the nearby town and 600 times higher than in the surrounding county. According to the report, "the risk for extrapulmonary complications [was] increased for persons of African or Filipino descent, but the risk [was] even higher for heavily immunosuppressed patients." The report then explained that physically removing heavily immunosuppressed patients from the affected area "would be the most effective method to decrease risk." The report also recommended ways to reduce the amount of dust at the prisons. After receiving the health department's recommendations, the Receiver convened its own committee. In June 2007, the Receiver's committee made recommendations that were similar to those from the health department.

In response, a statewide exclusion policy went into effect in November 2007. The inmates who were "most susceptible to developing severe or disseminated cocci" would be moved from prisons in the Central Valley or not housed there in the first place. The prisons used six clinical criteria to identify which inmates were most likely to die from Valley Fever: "(a) All identified HIV infected inmate patients; (b) History of lymphoma ; (c) Status post solid organ transplant ; (d) Chronic inmmunosuppressive therapy (e.g. severe rheumatoid arthritis ); (e) Moderate to severe Chronic Obstructive Pulmonary Disease (COPD) requiring ongoing intermittent or continuous oxygen therapy; [and] (f) Inmate-patients with cancer on chemotherapy." Inmates were not excluded from the Central Valley prisons based on race. The Receiver refined the exclusion policy in 2010 and created a list of "inmates who [were] at institutions within the Valley Fever hyperendemic area that [needed] to be transferred out." The record does not indicate that the 2010 policy excluded inmates from the outbreak prisons based on race.

In April 2012, the prison system's own healthcare services released a report examining Valley Fever in prisons. The report concluded that despite the "education of staff and inmates" and the "exclusion of immunocompromised inmates," there had been "no decrease in cocci rates." The authors found that Pleasant Valley State Prison inmates were still much more likely to contract Valley Fever than citizens of the surrounding county. From 2006 to 2010, 7.01% of inmates at Pleasant Valley State Prison and 1.33% of inmates at Avenal State Prison were infected. By comparison, the highest countywide infection rate was 0.135%, and the statewide rate was just 0.007%. From 2006 to 2011, 36 inmates in the Central Valley prisons died from Valley Fever. Prison healthcare services also found that male African-American inmates were twice as likely to die as other inmates. Each year, about 29% of the male inmates in California are African-American, but 50% of the inmates who developed disseminated cocci between 2010 and 2012 were African-American, and 71% of the inmates who died from Valley Fever between 2006 and 2011 were African-American.

Following this report, the Receiver issued another exclusion policy—one that would effectively suspend the transfer of African-American and diabetic inmates to the Central Valley prisons. The state objected, but the district court ordered the prisons to comply with the new exclusion policy.

There are several theories for why Valley Fever was more common inside the Central Valley prisons than in the surrounding areas. One theory is that new construction and excavation stirred up the soil, allowing the breeze to circulate the fungal spores. Many of the prisons were newly constructed or were being expanded during the outbreaks. Pleasant Valley State Prison, which had the highest rate of Valley Fever, was next door to a large construction project. Some prisons did not stop the airflow into their buildings on windy days. The prisons also might be built where there are more fungal spores or where the spores are more virulent.

Prison demographics were certainly relevant, as inmates were more likely to have certain risk factors. For example, adult males are at greater risk than women and children, and the prisons at issue in this case housed only adult males. African-Americans were also over-represented in the prison population, and they are more likely to develop disseminated cocci. Also, it could be that many prisoners were brought into the Central Valley from places that did not have the fungal spores, meaning that the inmates were not immune to the disease when they arrived at the prisons. By contrast, many civilians in the Central Valley could have been infected when they were young and healthy, and as a result, many civilians might have developed immunity without experiencing severe symptoms.

Finally, there may be differences in identifying people with Valley Fever. Inmates may be more likely than civilians to seek and obtain medical attention when they are sick. They may know about Valley Fever and request medical attention, while civilians with flu-like symptoms that go away in a few weeks may not. Prison doctors may be more aware of the Valley Fever problem than many doctors or other medical care providers outside the prisons. And it may be that Valley Fever is more widespread among the civilian population than the statistics indicate, because of lower diagnosis rates rather than lower incidence rates among civilians.

Even though Valley Fever is more common in prisons, it is important to remember that it is not unique to prisons. More than a million people freely live in the Central Valley, and many of them contract Valley Fever each year. Nor is the disease confined to the Central Valley. It occurs throughout the southwestern United States and is especially common in Arizona. Since the prisoners are confined together, it is especially important that Valley Fever is not contagious.

D. The Cases on Appeal

There are four cases consolidated on appeal. Each is a suit for money damages brought under 42 U.S.C. § 1983.

In Smith v. Schwarzenegger , current and former inmates of prisons in the Central Valley who were diagnosed with Valley Fever sued various state officials for Eighth Amendment violations. They alleged that the officials were deliberately indifferent to the inmates' exposure to an unreasonable risk of getting Valley Fever and developing disseminated cocci. The defendants moved to dismiss the complaint under Rule 12(b)(6), claiming that they were entitled to qualified immunity. The district court granted the motion to dismiss. It did not grant leave to amend the complaint. The inmates appeal.

In Gregge v. Kate , prison doctors diagnosed Gregge with cocci-meningitis while he was incarcerated at Pleasant Valley State Prison. He sued the prison warden and others for violating the Eighth Amendment. As in Smith , the district court dismissed the complaint under Rule 12(b)(6) based on qualified immunity. It did not grant leave to amend. Gregge appeals.

In Hines v. Youseff , Hines was incarcerated at Corcoran State Prison when he contracted Valley Fever. He brought an Eighth Amendment claim. The officials moved for summary judgment based on qualified immunity. The district court granted that motion and denied leave to amend. Hines appeals.And in Jackson v. Brown , inmates at Pleasant Valley State Prison and Avenal State Prison who got Valley Fever sued various officials. The defendants moved for judgment on the pleadings under Rule 12(c). The district court held that the officials were entitled to qualified immunity against the inmates' Eighth Amendment claim. But a subgroup of African-American inmates in Jackson had also alleged that the officials violated the Equal Protection Clause of the Fourteenth Amendment by intentionally failing to protect African-American inmates, whom the officials knew had a heightened risk of developing disseminated cocci. The court held that the officials were not entitled to qualified immunity against the Fourteenth Amendment claim. The officials appeal that decision. The inmates do not appeal the ruling on their Eighth Amendment claim.

STANDARDS OF REVIEW

We have jurisdiction over all four appeals. We do not have jurisdiction over the Plata decree, and it is not on appeal. We review whether the officials are entitled to qualified immunity de novo and the denial of leave to amend for abuse of discretion.

In Smith and Gregge , the district court granted the officials' Rule 12(b)(6) motions to dismiss the complaint. And in Jackson , the district court denied the officials' Rule 12(c) motion for judgment on the pleadings. So for those three appeals, we must accept as true all of the inmates' factual allegations, and we must draw all reasonable inferences in their favor. We must affirm the dismissal of the Smith and Gregge complaints if those complaints do not "contain sufficient factual matter, accepted as true, to state a claim to relief that is plausible on its face." We must reverse the denial of judgment on the pleadings in Jackson if "there is no issue of material fact in dispute" and the officials are "entitled to judgment as a matter of law."

Hines was decided at the summary judgment stage, not at the pleading stage. We therefore evaluate the grant of summary judgment based on the cognizable evidence. We must affirm the grant of summary judgment if there are no genuine issues of material fact and if, as the district court concluded, the officials are entitled to judgment as a matter of law.

Despite these different procedural stages and legal tests, the facts alleged in the Smith , Gregge , and Jackson complaints are largely identical to the evidence produced in Hines . Each of the appeals also presents the same basic question: whether the constitutional rights that the officials allegedly violated were "clearly established" when the officials acted. We therefore consider all four appeals together.

ANALYSIS

The officials in these cases are entitled to qualified immunity against claims that they were deliberately indifferent to a substantial risk of serious harm in violation of the Eighth Amendment. They are also entitled to qualified immunity against claims that they racially discriminated against African-American inmates. But first, we hold that several of the defendants cannot be sued at all because they were not personally involved in any alleged violations.

I. PERSONAL INVOLVEMENT

The inmates sued the officials under 42 U.S.C. § 1983. That means the inmates must show that each defendant personally played a role in violating the Constitution. An official is liable under § 1983 only if "culpable action, or inaction, is directly attributed to them."

The plaintiff in Hines argues that prison officials were deliberately indifferent to a substantial risk of serious harm when they housed him in the Central Valley. But the plaintiff has failed to demonstrate that defendants Joshua Garza, Dr. Godwin Ugeze, and Dr. Ashrafe Youseff were personally involved in any Eighth Amendment violations. Garza, a nurse practitioner, did not have any discretion to determine whether Hines should have been excluded from prisons in the Central Valley. There is also no evidence that Garza actually determined whether Hines should have been excluded from the Central Valley. There is no evidence that Dr. Ugeze was personally involved in determining what categories of inmates to exclude from the Central Valley. Instead, he was instructed to simply follow the exclusion criteria developed by others. And there is no evidence that Dr. Youseff had any contact with Hines. So the district court was right to dismiss those defendants from the case.

II. CRUEL AND UNUSUAL PUNISHMENT

The inmates allege that the defendant state officials violated the Eighth Amendment's prohibition on "cruel and unusual punishments" by being deliberately indifferent to the inmates' heightened risk of getting Valley Fever. The district courts in Smith , Gregge , and Hines held that the officials are entitled qualified immunity against those claims. Reviewing de novo, we affirm. Any Eighth Amendment right to be free from heightened risk of Valley Fever was not clearly established when the officials acted.

None of the cases before us seek an injunction that would regulate how the state assigns inmates to the Central Valley or how it addresses the risk of Valley Fever. That is the subject of the Plata case, which is not before us. The cases before us are only about whether individual defendants can be held liable for money damages because of allegedly unconstitutional acts and omissions.

To determine whether an official is entitled to qualified immunity, we ask two questions: (1) whether the official's conduct violated a constitutional right; and (2) whether that right was "clearly established" at the time of the violation. Helling v. McKinney sets out the constitutional framework for Eighth Amendment claims about involuntary exposure to environmental hazards. It held that an Eighth Amendment claim against an official for unconstitutional prison conditions requires an inmate to prove both an objective and a subjective factor.

For the objective factor, inmates must establish "that it is contrary to current standards of decency for anyone to be ... exposed against his will" to the hazard. This "requires more than a scientific and statistical inquiry into the seriousness of the potential harm and the likelihood that such injury to health will actually be caused." Instead, courts must "assess whether society considers the risk that the prisoner complains of to be so grave that it violates contemporary standards of decency to expose anyone unwillingly to such a risk," meaning that the risk "is not one that today's society chooses to tolerate."

For the subjective factor, inmates must show that the official is "deliberately indifferent" to the inmate's suffering. In Farmer v. Brennan , the Supreme Court explained that this standard means that an official is liable "only if he knows that inmates face a substantial risk of serious harm and disregards that risk by failing to take reasonable measures to abate it."

The courts below did not decide whether exposing inmates to a heightened risk of Valley Fever violates the Eighth Amendment. Neither do we. Instead, we go straight to the second prong of the qualified immunity analysis: whether a right to not face a heightened risk was "clearly established" at the time. A right is clearly established if it was "sufficiently clear that every reasonable official would [have understood] that what he is doing violates that right." That is, the issue must have been "beyond debate." In determining what is clearly established, we must look at the law "in light of the specific context of the case, not as a broad general proposition."

Applying those principles to the cases at hand, we conclude that the specific right that the inmates claim in these cases—the right to be free from heightened exposure to Valley Fever spores—was not clearly established at the time. A reasonable official could have concluded that the risk was not so grave that it violates contemporary standards of decency to expose anyone unwillingly to such risk, or that exposure to the risk was lawful.

A. Other Valley Fever Cases

The inmates' alleged constitutional right would be "clearly established" if "controlling authority or a robust consensus of cases of persuasive authority" had previously held that it is cruel and unusual punishment to expose prisoners to a heightened risk of Valley Fever. But no such precedent exists. The inmates argue that several of our memorandum dispositions clearly establish their right to not face an unreasonable risk of Valley Fever. But memorandum dispositions do not establish law. They are, at best, persuasive authority. And more importantly, none of the cited memorandum dispositions held that inmates have an Eighth Amendment right to not be exposed to a heightened risk of Valley Fever. The inmates also point us to unpublished district court decisions about Valley Fever exposure. We have previously said that unpublished district court decisions "may inform our qualified immunity analysis." But we have also noted that "it will be a rare instance in which, absent any published opinions on point or overwhelming obviousness of illegality, we can conclude that the law was clearly established on the basis of unpublished decisions only." And at most, the cited district court opinions show that the law was developing—not that it was already clearly established.

We therefore conclude that when the officials acted, existing Valley Fever cases did not clearly establish that they were violating the Eighth Amendment.

B. Eighth Amendment Principles

Of course, we do not require that heightened exposure to Valley Fever must have been previously held unlawful. The qualified immunity analysis does not require a case on all fours. What matters is whether "existing precedent ... placed the statutory or constitutional question beyond debate," not whether the debate has already taken place. An officer loses qualified immunity, even in novel factual circumstances, if he or she commits a "clear" constitutional violation. This rule prevents absurd results. As then-Judge Gorsuch once explained, "some things are so obviously unlawful that they don't require detailed explanation and sometimes the most obviously unlawful things happen so rarely that a case on point is itself an unusual thing. Indeed, it would be remarkable if the most obviously unconstitutional conduct should be the most immune from liability only because it is so flagrantly unlawful that few dare its attempt."

But this case does not involve a "clear" or "obvious" violation. The inmates must show that "every reasonable official would [have understood]" that exposing them to a heightened risk of Valley Fever violated the Eighth Amendment. More specifically, they must show that no reasonable officer could have thought that free society tolerated that risk. They have not met that burden for two reasons: a federal court supervised the officials' actions, and there is no evidence that "society's attitude had evolved to the point that involuntary exposure" to such a risk "violated current standards of decency," especially given that millions of free individuals tolerate a heightened risk of Valley Fever by voluntarily living in California's Central Valley and elsewhere. Those two facts mean that a reasonable official could have thought that he or she was complying with the Constitution.

It is especially significant that state officials could have reasonably believed that they were not violating the inmates' Eighth Amendment rights because the officials reported to the federal Receiver. The Plata district court appointed a federal Receiver in 2006—just a year after the Valley Fever outbreak began. The receiver entered orders about Valley Fever. Studies were conducted, and in 2010, the Receiver amended the policy excluding certain inmates from the Central Valley. Thus the federal Receiver appointed by the federal court to assure Eighth Amendment compliance actively managed the state prison system's response to Valley Fever.

Because the Receiver oversaw prison medical care and protective measures regarding Valley Fever, state officials could have reasonably believed that their actions were constitutional so long as they complied with the orders from the Receiver and the Plata court. The inmates do not claim that state officials defied the Plata Receiver. The Receiver promulgated orders directed specifically to the Valley Fever problem, and the inmates do not claim that the defendants defied those orders and that the defiance harmed them. The inmates fault the officials for not following various recommendations made before 2013. For example, in 2007 the California Department of Health Services recommended covering the prison grounds, but the soil was not stabilized until 2011 after the prisons got funding from the Receiver. Other recommendations were never adopted. But the inmates do not argue that the officials disobeyed the Receiver's binding orders, only that the officials did not promptly follow recommendations that were not orders. In determining what constituted the constitutionally sufficient level of protection, an official could reasonably have thought that it sufficed to comply with the Receiver's orders. As we once stated in a different context, "no reasonable prison official would understand that executing a court order without investigating its potential illegality would violate [a] prisoner's right to be free from cruel and unusual punishment."

Second, millions of people live in the Central Valley. This includes many African-Americans and others with a heightened risk of getting Valley Fever. Many people also work in the same prisons where the inmates live, exposed to the same fungal spores as the inmates. These people voluntarily live and work in the Central Valley despite a heightened risk of getting Valley Fever. Likewise, people live in Arizona despite the risk of getting Valley Fever. Each year, two-thirds of all Valley Fever cases are reported in Arizona. And from 1998 to 2016, the infection rate in Arizona nearly tripled. The infection rate is particularly high around Phoenix, Arizona. Yet Arizona's population grew an estimated 35.1% between 2000 and 2016. Where large numbers of people are exposed to a known risk, and yet no societal consensus has emerged that the risk is intolerably grave, a reasonable official can infer that the risk is one society is prepared to tolerate, like the risk of being injured or killed in a traffic accident.

Because so many people freely chose to live in the Central Valley despite the Valley Fever risk, and there is no evidence in the record that "society's attitude had evolved to the point that involuntary exposure" to either the heightened risk inside prison or the lower risk outside prison "violated current standards of decency," it would not have been "clear" to every reasonable officer that the inmates had a valid claim under Helling . The inmates have failed to show that every reasonable officer would have thought that "it violate[d] contemporary standards of decency to expose anyone involuntarily to such a risk," that is, that the risk of Valley Fever in the prisons was "not one that today's society chooses to tolerate."

We therefore affirm the district court rulings in Hines , Smith , and Gregge holding that the officials are entitled to qualified immunity against the Eighth Amendment claims. We also hold that the district courts did not abuse their discretion in denying the inmates' motions for leave to amend. Any attempt to amend the pleadings would be futile because we see no way to hold that the officials violated a clearly established Eighth Amendment right.

III. RACIAL DISCRIMINATION

For unknown reasons, Valley Fever disproportionally affects African-Americans. State officials did not exclude African-American inmates from the outbreak prisons until a federal court ordered them to do so in 2013. Some of the inmates in Jackson allege that this failure violated the Equal Protection Clause of the Fourteenth Amendment. According to the complaint, the officials "intentionally failed" to exclude African-American inmates from Pleasant Valley and Avenal State Prisons (or otherwise reduce the risk of harm) because the officials wanted to harm African-American inmates. Thus, the inmates allege, it was discriminatory to adopt a race-neutral exclusion policy that excluded inmates from those prisons based solely on medical conditions. That is, they allege it was discriminatory not to discriminate. On a motion for judgment on the pleadings, the district court held that the officials lacked qualified immunity. The officials appealed, and we reverse. We address an unusual Equal Protection claim that it was a denial of equal protection not to segregate prisoners by race.

The district court analyzed this case as being about "the right to non-discriminatory administration of prison services." The district court and the inmates both rely on Elliot-Park v. Manglona , but that case is inapposite. In Elliot-Park , a Micronesian drunk driver crashed into a Korean driver. The investigating police officers were all Micronesian. The Micronesian driver told an officer that "he had ‘blacked out’ while driving," but the officers did not test him for intoxication or arrest him for drunk driving. The Korean driver sued the officers, arguing that their failure to investigate or arrest the drunk driver was motivated by racial animus against Koreans. We held that the officers lacked qualified immunity because "[t]he right to non-discriminatory administration of protective services is clearly established." Because the officers considered race when deciding whom to help, strict scrutiny applied.

But Elliot-Park did not establish that state actors could violate the Equal Protection Clause by adopting a race-neutral policy. Implicit in our holding in that case was the fact that police officers typically arrest drunk drivers. The officers diverged from the norm, allegedly because of racial animus. That is, they allegedly treated Korean drivers differently than they treated Micronesian drivers.

Here, by contrast, the officials did not have one policy for African-American inmates and another for white inmates. All inmates were treated the same, regardless of race. The officials are said to have violated the Constitution precisely because they treated the inmates the same regardless of race—not, as in Elliot-Park , because they treated people differently because of their race. So for the officials here to lose qualified immunity, it would have to have been clearly established that treating people of all races the same violated the Equal Protection Clause. For three reasons, it would not have been clear to a reasonable person, acting on the officials' information and motivated by their purposes, that the Equal Protection Clause required excluding African-American inmates from these prisons based on race.

First, from 2006 onward, a federal Receiver supervised the prisons. During that time, multiple experts gave recommendations. An exclusion policy went into effect in 2007. The Receiver modified that policy in 2010. It was not until April 2012 that experts proposed excluding African-Americans from the Central Valley. The Receiver did not formally recommend a policy that would exclude African-Americans until November 2012. The inmates note that the prisons objected to excluding African-Americans from the affected prisons. But the inmates do not argue that the prisons failed to obey the district court's order once that order was made. And again, since 2006, the prisons were under the Receiver's supervision. The officials adopted exclusion policies in accord with the Receiver's directions and under the Receiver's watchful eye. Therefore, an official could have reasonably believed that the policies about excluding (or not excluding) African-Americans from Central Valley prisons did not violate the Equal Protection Clause. There is a second reason why the officials have qualified immunity: the Constitution generally demands race neutrality. Over and over again, the Supreme Court has unambiguously held that "all racial classifications" are invalid unless they pass strict scrutiny. That is, an express racial classification (like the ones the inmates want) is presumptively unconstitutional. It can survive only if the state proves that the classification is "narrowly tailored" to achieving a "compelling" state interest. Even so-called "benign" racial classifications must satisfy strict scrutiny. In Johnson v. California , prison inmates challenged a policy of temporarily segregating inmates based on race. Even though the prison adopted the policy to avoid racial gang violence, the Supreme Court plainly held that strict scrutiny applied.

Mitchell v. Washington demonstrates how strict scrutiny applies to race-based medical decisions. There, an African-American inmate with Hepatitis C asked a prison doctor to treat him with certain drugs. The doctor did not prescribe the drugs because they "had been largely unsuccessful on African-American males" with Hepatitis C. The inmate sued the doctor on the theory that basing treatment decisions on race violated the Equal Protection Clause. We held that strict scrutiny applied because "even medical and scientific decisions are not immune from invidious and illegitimate race-based motivations and purposes." Even though the doctor might have had good intentions and good data—the inmate later got the demanded treatment, and it was unsuccessful—"there is simply no way of determining what classifications are benign or remedial and what classifications are in fact motivated by illegitimate notions of racial inferiority or simple racial politics." The doctor did not give any compelling reason for why he considered the inmate's race, so he violated the Equal Protection Clause.

Third, a reasonable official could have believed that not excluding African-Americans from the prisons was consistent with the scientific data and pre-2012 expert recommendations. The California Department of Health Services began investigating Valley Fever at Pleasant Valley State Prison in 2005. It summarized its findings in a January 2007 report that assessed the relative risk of contracting Valley Fever for various populations. Overall, inmates with a chronic medical condition had a 2.7 "relative risk," meaning that they were 2.7 times more likely to contract Valley Fever than inmates without a chronic condition. Inmates with pulmonary conditions had a 3.8 relative risk. Diabetics had a 2.7 relative risk. Those with chronic heart conditions had a 1.6 relative risk. Inmates over age 40 had a 1.6 relative risk. And African-American inmates had a 1.9 relative risk compared to white inmates.

According to the report, 47% of African-American inmates' risk was due to race alone. But the report also estimated that removing African-American inmates from the prison would only reduce the number of Valley Fever cases by, at most, 16%. And even though being African-American was a risk factor for getting Valley Fever, the report said that being African-American "was not associated with more severe disease." The two biggest risks were having a chronic medical condition and being housed in a facility with more outdoor exposure. So the report concluded that targeting chronic conditions and outdoor exposure could do more to decrease Valley Fever than targeting race or age. Based on these relative risks, the state health department recommended the following:

Consider relocating the highest risk groups to areas that are not hyper-endemic for [the fungal spores]. Previous studies have suggested that the risk for extrapulmonary complications is increased for persons of African or Filipino descent, but the risk is even higher for heavily immunosuppressed patients. In this investigation, we found an increased risk among persons with chronic medical conditions, especially pulmonary conditions. Prevention efforts are critical for these higher risk populations and may mitigate the risk, but physical removal of these highest risk groups from highly endemic regions, if possible, would be the most effective method to decrease risk.

A reasonable official could have read this report and its recommendations and concluded that African-Americans did not need to be excluded from the Central Valley based on race. Even though African-American inmates had a higher risk of getting Valley Fever than did white inmates, those with chronic diseases typically had even higher risks. And because nearly one-third of inmates were African-American, a reasonable official could have decided that it was better to try less burdensome measures first.

In short, it was reasonable to exclude inmates based on medical conditions rather than based on race. Even if state officials should have been more aggressive in excluding inmates whose higher risk appeared to be on account of (or at least connected to) their race, that does not mean their conduct violated clearly established law. The inmates did not have a clearly established right to be segregated from certain Central Valley prisons based on their race. We therefore reverse the Jackson court's ruling on the equal protection claim.

CONCLUSION

We are sympathetic to the inmates' plight. Valley Fever is a serious and potentially fatal disease. When state officials know that inmates face a substantial risk of serious harm, the officials are constitutionally required to take reasonable steps to abate that risk. State officials cannot shut their eyes to inmate suffering; they are responsible for the safety of the people in their custody. But it would not have been "obvious" to any reasonable official that they had to segregate prisoners by race or do more than the federal Receiver told them to do. So we conclude that the defendants are entitled to qualified immunity. The rights that the inmates claim were not clearly established when the officials acted. Granting leave to amend would be futile. We therefore AFFIRM the judgments in Hines , Gregge , and Smith , and we REVERSE the judgment on appeal in Jackson .