You are using an outdated browser.
Please upgrade your browser
and improve your visit to our site.

How Medical Examiners Shield Violent Cops From Scrutiny

They wield the important power of determining a victim’s manner of death—but they’re susceptible to both political pressure and racial bias.

In the summer of 2021, the body of 29-year-old Damien Cameron was delivered to the crime lab in Jackson, Mississippi. Cameron, a Black man, had died after a struggle with police. His face was swollen and bloodied, and there was bleeding in his neck. Witnesses described seeing two deputies kneeling on Cameron for more than 10 minutes. Yet Staci Turner, the state’s medical examiner, ruled Cameron’s manner of death to be “undetermined.” That decision effectively headed off any further investigation. The officers returned to work.

Two years later, Rankin County Deputy Hunter Elward, one of the officers who had knelt on Cameron, pleaded guilty to federal charges in another case—the horrific torture of two Black men. Elward and a gang of his fellow deputies, who called themselves the Goon Squad, admitted to breaking into the men’s home, torturing and sexually humiliating them, then sticking a gun into the mouth of one of the men, at which point the deputies claim it mistakenly fired, leaving the man with severe and disfiguring injuries.

The fallout from the torture cases brought new attention to Turner’s work in Cameron’s death. In a subsequent review, three separate medical examiners said that Turner was wrong to classify Cameron’s manner of death as “undetermined.” They say it was clearly a homicide. If it had been classified that way at the time, it’s possible that a criminal investigation might have taken Elward off the street, and might have revealed the rot in the sheriff’s department.

Cameron’s death came 14 months after the death of George Floyd, and there are clear similarities. Both men died after one or more police officers knelt on their backs. Both cases involve disputes between forensic pathologists about manner of death, and both raise important questions not only about how that determination is made but about whether medical examiners are adequately shielded from political pressure when making it.

Floyd’s case again came under scrutiny in right-wing media last October. In an unrelated gender discrimination lawsuit against the Hennepin County District Attorney’s Office, prosecutor Amy Sweasy testified that in the days after Floyd died under the knee of Police Officer Derek Chauvin, medical examiner Andrew Baker told her that “there were no medical findings that showed any injury to the vital structures of Mr. Floyd’s neck.” Sweasy says Baker then asked her, “What happens when the actual evidence doesn’t match up with the public narrative that everyone’s already decided on?” and warned, “This is the kind of case that ends careers.”

That testimony expands on a memo Sweasy wrote that was well publicized by Chauvin’s defense team in August 2020, though the comments she attributed to Baker in that memo were somewhat less inflammatory. Baker did eventually testify for the state and classified Floyd’s death as a homicide, prodding far-right pundits like Tucker Carlson to conclude that the doctor had changed his testimony to appease activists and political leaders. Under oath at Chauvin’s trial, Baker denied changing his diagnosis or bowing to pressure.

The overwhelming consensus in the medical community—among medical examiners, pulmonologists, cardiologists, and addiction specialists—is that the actions of Chauvin and other officers were the proximate cause of George Floyd’s death. That said, it’s certainly understandable why Baker’s comments to Sweasy would fuel conspiracists. The medical examiners I consulted found them to be puzzling.

And Carlson has something of a point, although it isn’t the point he intended. Medical examiners are often subjected to political and professional pressure that can bias their judgment. Moreover, Carlson and his fellow demagogues are correct that what happened in Chauvin’s case was an anomaly. It’s just that the anomaly, in this instance, is that the state eventually got it right.

When it comes to in-custody deaths, there are far more cases like Damien Cameron’s—in which a faulty manner of death determination heads off any further investigation of police misconduct—than like George Floyd’s, in which police use of force is correctly found to have caused a death.

About two months before the Chauvin trial, a team of researchers led by London-based cognitive neurologist Itiel Dror published a study in the Journal of Forensic Sciences that, along with the trial, would open an ongoing feud within the medical examiner community.

Dror and his team asked 133 medical examiners to review an autopsy report of a child’s death and determine if the death was accidental, a homicide, or if there was insufficient information to make a determination. The participants were given identical autopsy reports, with two exceptions. Half were told that the deceased child was Black and had died while in the care of the mother’s boyfriend. The other half were told the child was white and had been in the care of a grandmother.

This information should have been irrelevant to the manner of death. While it’s true that Black children are more likely to die from homicide, and it’s also true that stepfathers or love interests of a child’s mother are more likely to kill a child than a grandmother, neither of those things should be relevant when determining how a specific child died in a specific case. We don’t convict people based on generalizations about the demographic groups to which they or their victims belong.

Or at least we shouldn’t. The study suggested that perhaps sometimes we do. The medical examiners given the scenario with the Black child were five times more likely to rule the death a homicide than those with the white child. There was also broad disagreement among the study’s participants about whether there was sufficient information to even make a manner of death determination.

Predictability is a core tenet of established science. If you ask a group of scientists to apply the settled principles of their field to the same set of facts, they should reach the same conclusion. That didn’t happen here. Dror has since put out two more studies further confirming that extraneous information can bias the conclusions of medical examiners.

On some level this isn’t surprising. Medical examiners are human, and human beings are affected by cognitive bias. Yet the study sparked a furious backlash. Forensic pathologists from around the country went after Dror and his colleagues—some of whom are medical examiners themselves—with a barrage of personal and professional attacks.

“It was an extremely disturbing experience to be openly attacked by my colleagues, some of whom I thought were friends,” says Dr. Jonathan Arden, one of the study’s authors. Arden himself is a former president of the field’s largest professional organization, the National Association of Medical Examiners, or NAME, and the former head of the Washington, D.C., crime lab. “We were attacked baselessly, inappropriately, in a public forum. I had to retain counsel, and ended up spending a substantial amount of time, money, and effort responding to these accusations.”

“We expected people to disagree with the study,” says Dror. “But these are supposed to be scientists. You write a letter to the editor. These doctors make decisions that send people to prison. You’d think they’d welcome feedback that will help them minimize mistakes. Instead they sent a clear message: Don’t criticize us. We’ll go after you. We’ll go after your career.

“They turned this into the Spanish Inquisition.”

The study’s publication just before the Chauvin trial was coincidental. But the two events would become inextricably intertwined. Three of the study’s most vocal critics signed on as expert witnesses for Chauvin’s defense, while one of the study’s authors consulted for the prosecution.

The Dror studies, the response to them, and what we’ve learned from these high-profile deaths in police custody should prompt some weighty questions about how we investigate suspicious deaths in the United States. If medical examiners can’t even agree on manner of death in one of the most closely followed, consequential trials in recent memory—one involving what is likely the most-witnessed homicide in human history—how can we rely on their expertise to send people to prison, or to determine if we should continue to trust that a police officer should remain on the job after that officer has taken a life?

What if the life-and-death determinations that we think are governed by scientific principles and careful deliberation are susceptible to the prejudices, misplaced incentives, and inherent bias of a broken system?

When medical examiners perform an autopsy after a suspicious death, they’re asked to make two determinations: the cause of death and the manner.

The cause is usually well grounded in medicine, and while there are some exceptions, it typically isn’t controversial. Examples might include cardiac arrest, asphyxiation, exsanguination (blood loss), or blunt-force trauma.

Manner of death is the mechanism by which the death occurred. It’s much more subjective and far more consequential. It often determines whether or not police and prosecutors proceed with a criminal investigation. Typical manner of death determinations include suicide, homicide, accident, natural causes, and undetermined.

Manner of death is subjective enough that it can be hotly disputed among medical examiners. Yet the consequences can be profound. Parents, grandparents, and other caretakers have gone to prison because a medical examiner ruled a child’s death to be a homicide rather than natural or accidental. Medical examiners are also often asked to determine if a death was due to suicide or murder, or if there are signs that someone was killed in self-defense. Meanwhile, classifying a death as a suicide, accident, or undetermined instead of a homicide typically means no criminal investigation.

“Manner of death is not a medical determination,” says Peter Neufeld, co-founder of the Innocence Project. “It’s a legal determination that necessarily involves processing nonmedical information. Why is a doctor in a better position to evaluate the veracity of a suicide note, the truthfulness of a police report, or the reliability of a witness identification? They aren’t. Medical examiners simply don’t have the training to make those calls.”

In fact, the United States is the only country in the developed world in which medical examiners are asked to testify about the manner of death. It’s a vestige of the quirky, anachronistic coroner-based death investigation system we inherited from England.

Physicians belong to one of the most esteemed professions in American life. So they tend to be persuasive expert witnesses. “They bring this inherent trustworthiness and sense of authority,” says Neufeld. But that can become a problem when they start testifying outside their area of expertise, which manner of death determinations often require.

Compounding the problem, medical examiners often operate in systems that can bias their decisions, often by design. In many jurisdictions, the medical examiner’s office falls under the auspices of a law enforcement agency, which means they ultimately report to police or prosecutors who have a strong interest in these determinations. In other jurisdictions, prosecutors or elected coroners have contracted autopsies out to private pathologists, creating a strong incentive for those doctors to tell law enforcement what they want to hear in order to get more referrals. In some places, crime labs actually receive more funding when someone is convicted (it’s a fee that’s tacked on to a defendant’s punishment)—practically the platonic ideal of a perverse incentive.

Even in cities and counties where medical examiners aren’t directly plugged into the law enforcement apparatus, they’re still commonly considered to be part of the prosecution’s “team.” They’ll often consult with prosecutors or police investigators before starting an autopsy. Sometimes, law enforcement officials are even present during the procedure—access rarely if ever granted to a defense team.

In a system where medical examiners had true independence, they’d also be free to consult with defense attorneys when they believe the evidence doesn’t support a prosecution. But that just isn’t done. Medical examiners can communicate with defense attorneys who reach out to them, but proactively making contact is generally frowned upon, and medical examiners who have done so have faced retaliation.

Judy Melinek, a medical examiner and one of the besieged co-authors of the controversial Dror-led study, wrote in an article for MedPageToday, “There is a subset of prosecutors who believe, and will not be dissuaded, that the investigative work done by a medical examiner always ought to align with the goals of law enforcement. Some forensic pathologists feel the same way.”

About two months after the death of George Floyd, defense attorneys for the officers charged with his death filed the exhibit that ignited Derek Chauvin’s far-right defenders.

According to Sweasy, shortly after Floyd’s death, Baker told her that Floyd had a preexisting heart condition. He also had some meth and a potentially fatal amount of fentanyl in his system. “The autopsy revealed no physical evidence suggesting that Mr. Floyd died of asphyxiation,” Sweasy wrote, quoting Baker, and “if Mr. Floyd had been found dead in his home (or anywhere else) and there were no other contributing factors he would conclude that it was an overdose death.” Baker also said Floyd had “sickle-cell trait,” a condition sometimes—and dubiously—cited when Black people die in police custody.

The following day, in handwritten notes, Baker documented his comments to Sweasy although, notably, he added, “I’m not saying this killed him.”

The medical examiners with whom I consulted had a hard time with Baker’s initial observations to Sweasy. They were confused both by the observations themselves and how they conflicted with Baker’s testimony at trial.

Overdoses are notoriously difficult to diagnose, particularly with respect to opioids. Frequent users build up tolerance, so a fatal concentration of an opioid in one person could be nontoxic in another. Moreover, addiction and drug specialists have argued that if Floyd had been in the midst of a fatal overdose, he wouldn’t have been able to communicate with the police in the manner he did, and he certainly wouldn’t have been able to walk.

Toxicologists and addiction specialists have since asserted—including at Chauvin’s trial—that the amount of fentanyl in Floyd’s system was about a fifth of the average concentration in postmortem cases, and the amount of meth in his system was negligible.

There’s also no truth to the claim made by Carlson and others that the lack of physical injury to Floyd’s neck proves that Chauvin couldn’t have caused his death. Pressure on the neck and constriction of the diaphragm can quickly cause breathing to gradually grow more shallow. That can cause oxygen levels to drop, eventually causing brain damage and heart failure. There’s even a term for killing in this manner. It’s called “burking,” after William Burke, who with an accomplice killed 16 people that way in the 1820s in order to sell the cadavers to a medical college.

Baker ultimately concluded that Floyd died this way—heart failure brought on in part by asphyxiation due to the pressure Chauvin put on Floyd’s chest and neck. He classified the death as a homicide, and he testified for the prosecution at Chauvin’s trial. Under cross-examination, Baker said his initial statements to Sweasy were based on incomplete, preliminary observations, and he denied he had been influenced by activists or state officials.

But that only raises more questions: Why would Baker consult with the prosecution before he’d seen all the results, and why would he make such sweeping claims about Floyd’s death before he’d obtained information that ultimately was important enough to change his conclusions?

It’s actually common for medical examiners to consult with police prosecutors early on in an investigation. There’s also a lot of debate about whether it’s proper.

“It’s a pattern with in-custody deaths,” says Joye Carter, a co-author of the Dror study who consulted for Floyd’s family and the prosecution. In 1992, Carter was named director of the Washington, D.C. medical examiner’s office, becoming the first Black woman in the country to head such an office. “There’s a knee-jerk desire to quickly clear these police officers. Some medical examiners bend over backwards to blame these deaths on race-based conditions like ‘sickle-cell trait’ or excited delirium. That conditions law enforcement officers to be more aggressive with Black people.”

Carter speculates that Baker first acted as a cog in a machine that’s programmed to rapidly exonerate police, but may have slowed down and taken another look once he saw that he’d be contradicted by an overwhelming consensus among cardiology, pulmonary, and addiction specialists.

Other forensic pathologists took a more sympathetic view of how Baker handled the case. “Sweasy is a litigant in a lawsuit, and the comments she attributed to Baker are hearsay,” one medical examiner told me, who did not wish to be quoted by name. “We haven’t heard Baker’s version of that conversation.”

It would be brazen and risky for a prosecutor like Sweasy to lie under oath. It also isn’t at all unusual for two people to have earnest but strikingly disparate recollections of the same conversation. It’s also worth noting that for all of the right-wing embrace of Sweasy’s deposition, the prosecutor herself said she thought Chauvin should have been charged—although with third-degree murder instead of second.

“I’ve been on both sides of this,” this medical examiner added. “I know what it’s like to have police pressuring you to clear officers who killed someone, and what it’s like when you discover evidence that exonerates the officer while protests are raging. I wouldn’t want any comments I made in those moments to appear out of context in some court document.”

Another medical examiner, a former officer with the National Association of Medical Examiners who asked not to be identified by name, also offered a mild defense of Baker, suggesting he may have felt compelled to leave a meticulous record given the high-profile nature of the investigation. “It’s unfortunate, because that then opened him to allegations that he was pressured to change his conclusions,” the medical examiner said.

But both doctors still said Baker’s initial comments to Sweasy, if accurately portrayed, are puzzling and that at some point he should publicly explain them. Baker’s office did not reply to my requests for an interview, but the office has told other outlets that he stands by his testimony at Chauvin’s trial and won’t respond to allegations in an unrelated lawsuit.

In the early 1990s, DNA testing began to show that many forensic specialties once touted as nearly infallible had put innocent people in prison. As exonerations mounted, the scientific community took notice and began to test the veracity of some of these fields.

In the 30 or so years since, studies and reviews by scientific bodies like the National Academies of Science and the President’s Committee on Science and Technology have been sharply critical of controversial forensic fields like bite-mark and hair and fiber analysis, but also of long-trusted specialties like ballistics matching and fingerprinting. The forensics community and its supporters in law enforcement have responded to these with defensiveness and resentment, often lashing out at scientists who scrutinize their fields.

Until the Dror studies, medical examiners had largely escaped such scrutiny. That’s in part because forensic pathology straddles the line between medicine and forensics. As doctors, medical examiners get much of the prestige and respect we afford physicians. But the field has also liberally co-mingled with forensics and law enforcement, and even before the Dror studies, there were signs that these attachments could sometimes corrupt the medical decisions of its practitioners.

Perhaps the best example is “shaken baby syndrome,” a theory that put hundreds of people in prison before scientists outside the criminal justice system proved that symptoms medical examiners contended could only be caused by violent shaking could also be caused by other factors like falls or illness.

Investigations by ProPublica and the Medill Justice Project have also found that child deaths are far more likely to be classified as crimes in some states and counties than others, suggesting that such determinations aren’t grounded in hard science so much as the priorities and personal predilections of local police and prosecutors.

Forensic pathology has separated from medicine in other ways too. Most medical examiners work in the public sector, earning significantly less money than people in other medical specialties. These status disparities, combined with the perception that proper physicians work with the living rather than the dead, has brought disparagement to the field as a sort of professional dumping ground for those who didn’t excel in medical school.

There are of course plenty of smart and conscientious medical examiners driven by an earnest commitment to public service. But it’s also indisputable that medical students aren’t rushing into forensic pathology. Most medical schools no longer even teach it. Moreover, state legislatures chronically underfund medical examiners’ offices. As a result, the United States has faced a critical shortage of qualified forensic pathologists for decades. All of which means these offices can’t always afford to be picky about who they hire.

The profession has also taken a beating for its reluctance to self-police, despite some jaw-dropping scandals. High-ranking state medical examiners have been revealed to conduct astonishing numbers of autopsies, to conduct sloppy and unprofessional investigations, and to have given preposterous testimony that contributed to wrongful convictions. Some have been accused of criminal misconduct.

Forensic pathologists also operate in an inherently adversarial system that’s unique in the medical profession. As the Chauvin trial showed, it isn’t at all unusual for two doctors to give contradictory testimony in the same case. It makes for a contentious, often combative work life. Given the high stakes, it can be professionally damaging to admit to mistakes.

“When confronted with the possibility that they may have gotten a determination wrong, too many medical examiners double down rather than consider the merit of the criticism,” says Arden. “I’ve had medical examiners who, years after stubbornly defending their findings in a case, came back and said, ‘You know what, you were right about that.’ I’ve found that especially disturbing in cases where their findings sent someone to prison.”

The way the medical examiner community received the Dror study might have been a more important test of the profession than what the study actually found. “When I’ve found cognitive bias in other fields of medicine that was affecting patient health, those findings were welcomed,” Dror says. “They wanted to improve.” Dror says he had hoped the medical examiner community would embrace the feedback, as other fields of medicine have done.

That isn’t what happened. Dror and and his six co-authors, four of whom are medical examiners themselves, were subjected to acrid, unusually personal vitriol from colleagues across the country. NAME went particularly hard after Dror, who is not a medical examiner. The group’s president and executive director sent an official complaint to Dror’s then employer, University College of London. After an investigation, the university determined that Dror had done nothing wrong.

The reaction from medical examiners was so unhinged that the Journal of Forensic Sciences’ editor published an unusual note chastising the critics for their lack of decorum and collegiality.

Critics of the study seemed particularly irked by the racial component. “They keep saying we called them racist,” Dror says. “But the study was never about detecting racism. It was to measure how bias can affect a manner of death determination. We used race as one variable because it’s well established that the deaths of Black children are far more likely to be classified as homicides.”

“I was disappointed by the reaction, but I wasn’t surprised,” says Carter. “I’ve seen medical examiners say things like, because of the color of the decedent’s skin, they couldn’t tell if there were bruises. Which is just incredible. When I’d tell that story to explain why representation of nonwhite people is important, they would just shut down. They didn’t want to hear it. They just hear that as you calling them racist. So now you have a study finding that racial stereotypes can influence manner of death determinations. I can’t say I’m surprised that they don’t want to hear that, either.”

Dror and Neufeld speculate that the unusually unruly response stems from the fact that medical examiners aren’t accustomed to being proven wrong. “Doctors who work with the living get constant feedback for how they’re doing. If a patient doesn’t improve, you know to change the course of treatment,” Neufeld says. “But when you work with the dead, you don’t get that kind of feedback.”

“Unless there’s DNA testing involved, there’s rarely a ground truth in forensic science to tell a medical examiner when they’re wrong,” adds Dror. “So I think what we’ve seen with the reaction to the study is a profession being told for the first time that they can be wrong. And they don’t want to hear it.”

Critics filed eight complaints against Dror and his co-authors. All have since been dismissed. The Journal of Forensic Sciences initially sent the study to three peer reviewers, all of whom approved it for publication. After the wave of criticism, the journal sent the study to a fourth statistician. That statistician too found nothing wrong with the study.

Dror has since collaborated with other researchers and medical examiners to publish two additional studies on cognitive bias in forensic pathology. In the first, medical examiners were given identical autopsy reports, but half were accompanied by a police report suggestive of suicide, while the other half were given one suggestive of homicide. The medical examiners were far more likely to give a manner of death determination that matched the police reports. The final study, published in September, showed that when police give medical examiners just one theory of the crime, they’re likely to agree with that theory. Providing more than one theory changed their analysis.

The reaction to these subsequent studies has been less intense, but Dror says the criticism had an impact. “The response has made my medical examiner colleagues extremely skittish,” he says. “They want to water down the language in these subsequent studies, not because it’s wrong or overstated but because they don’t want their colleagues to come after them. One has asked that his name be redacted. It isn’t that he doesn’t agree with the results. He is afraid of retaliation. It’s definitely had a chilling effect.”

For all the protests, legislation, racial reckoning, and backlash inspired by George Floyd’s death, the most lasting—and perhaps improbable—legacy may be to have exposed and even helped correct the problems with how in-custody deaths are investigated in the U.S.

After the Sweasy memo first appeared in defense filings in August 2020, George Floyd’s family hired two outside medical examiners to conduct their own autopsy. Those doctors concluded that Floyd died of asphyxiation caused by pressure applied by Chauvin, and that his death was indeed a homicide.

That review provoked a bizarrely defensive—but perhaps revealing—response from the National Association of Medical Examiners. In a public statement, NAME discouraged obtaining second opinions for forensic autopsies. It was a strange stand to take—second opinions and peer review are a core tenet of science and medicine. Given the historic use of manner of death determinations to cover up racial violence, it was also inexplicably tone-deaf. After a flurry of public criticism, NAME retracted the statement.

The group has since hired Carter to head up a diversity and inclusion program to educate medical examiners on sickle cell trait, excited delirium, and other matters white medical examiners may encounter when conducting autopsies on Black people. Carter says she had long advocated such a program, but until the 2020 protests, she’d received little support.

One of the most vocal critics of the first Dror study was Brian Peterson, who at the time was chief medical examiner for Milwaukee. He was the lead signatory of the most prominent letter to the editor of the Journal of Forensic Sciences—the one that inspired a note from the editor. Peterson also attacked the studies’ authors in particularly personal terms on the NAME listserv. In one especially odd—and perhaps revealing—email to the group, Peterson wrote, “Is there anyone in our profession that has not, at one point or another, quipped about ‘spinning the wheel of death’ and picking one?”

“I don’t even know how you respond to that,” says Carter. “It’s such an embarrassingly cavalier thing to write about what we do—about the decisions we make that can send people to prison.”

Peterson was also a consultant for Chauvin’s defense, though he didn’t ultimately testify at the trial. Peterson himself has a controversial history with in-custody deaths. A decade ago, he issued a contentious diagnosis about another Black man who died with a police officer’s knee on his back, even as that man complained that he couldn’t breathe. Though Peterson and his office did ultimately classify that death as a homicide, he attributed the death to sickle-cell trait, a condition he speculated might have been brought on by the man’s decision to flee and resist police officers. Last fall, Peterson either retired or was fired from his position in Milwaukee, depending on whom you believe. He then stopped showing up for court hearings in pending cases, including the highly publicized trial of a police officer accused of killing a 25-year-old Black man with a choke hold.

Another prominent critic of the study was David Fowler, the former state medical examiner in Maryland. Fowler actually testified for Chauvin’s defense, and his testimony was so controversial that the Maryland attorney general appointed a panel to conduct an unprecedented review of Fowler’s work on in-custody deaths.

That review is ongoing. But last year, attorneys for the Maryland ACLU said they had found dozens of cases in which someone died while under some sort of restrictive police restraint, after which Fowler’s office ruled the death “undetermined.” In some of those cases, “the decedent had been Tased, pepper sprayed, subject to police baton strikes, prone restraint, or other uses of force.”

Critics like Dror say that while those sorts of determinations could be the product of a personal bias in favor of exonerating police, they’re more likely due to systematic bias.

“If you bring in a homeless person who was beaten to death by a street gang and covered in bruises, most medical examiners will quickly rule that a homicide. But if the same homeless person was beaten by police, they’ll run the gamut of tests. If you look hard enough, you can always find some other condition that might have contributed to death,” Dror says. In some jurisdictions, Dror says, in-custody deaths are required to undergo more tests as a matter of policy.

This is where diagnoses like “excited delirium” come into play. Despite dubious origins and little support in clinical and academic research, the condition caught on after the company formerly known as Taser spent millions to promote it among coroners and medical examiners. It posits that people apprehended by police who are intoxicated, high, or panicked, or who have ailments like heart disease or genetic conditions like sickle cell trait, can go into cardiac arrest and die when restrained or apprehended by police, through no fault of the police.

Excited delirium isn’t recognized by either the American Medical Association or the American Psychiatric Association—and after 2020, both groups explicitly recommended against its use. But it has been used to exonerate hundreds of police officers for in-custody deaths, and it’s disproportionately invoked in Black and Latino deaths. Chauvin’s attorneys managed to work it in at his trial, but the judge in the 2022 trial of two other officers barred them from claiming it was a factor in Floyd’s death. (Tellingly, excited delirium diagnoses are all but nonexistent outside a police context, and nearly all involve some sort of restraint.)

Even if excited delirium were a real condition, it shouldn’t affect a manner of death determination. “A homicide is a homicide,” says Carter. “A prosecutor can decide later that it was a justifiable homicide like self-defense. But if the person would still be alive if not for the police officer’s actions, it’s a homicide.”

Floyd’s death and the resulting protests prompted more skepticism of these sorts of deaths across the country.

In early November, as part of a settlement to the lawsuit related to the ACLU investigation, the state of Maryland agreed to sweeping changes to shield medical examiners from pressure by police and prosecutors.

But there have been changes elsewhere too. After Floyd’s death, authorities in Colorado reopened an investigation into the death of Elijah McClain, a healthy, 23-year-old Black man who died after police stopped him, threw him to the ground, and restrained him. Paramedics forcibly injected McClain with ketamine after officers at the scene claimed he may have been suffering from excited delirium.

After consulting with police, the county coroner classified McClain’s manner of death as “undetermined” and speculated that he may have died of excited delirium brought on by an undiagnosed mental health condition. Citing the coroner’s report, the local district attorney then cleared the officers of wrongdoing.

In the summer of 2020, Colorado Governor Jared Polis appointed a special prosecutor to reinvestigate McClain’s death. A grand jury subsequently found that police officers had no probable cause to stop McClain. The autopsy report was then amended to say McClain died of “complications of ketamine administration following forcible restraint.” One officer has since been convicted of manslaughter. Another was acquitted.

The 2020 protests also prompted scrutiny of police tactics like hog-tying and restraint chairs. They spurred reexamination of other in-custody deaths in Colorado, Utah, Oakland, and Illinois and inspired a critical academic review of the research police and prosecutors cite when they claim there’s no risk to restraining suspects while they’re lying face down.

Floyd’s death was also likely a factor in a nationwide reassessment of excited delirium. After years of criticism from medical professionals, NAME finally disavowed the diagnosis in March. In October, a panel for the last holdout medical organization—the American College of Emergency Physicians—voted to withdraw its endorsement of the seminal medical paper upon which the diagnosis is based. A week later, the California legislature voted to ban coroners from citing excited delirium as a cause of death.

But critics like Neufeld urge a more substantive change. They say the Dror studies demonstrate why the U.S. should join the rest of the world and bar medical examiners from testifying about manner of death in court—an opinion shared by Dror’s medical examiner co-authors, Melinek and Arden. Dror himself is on the fence on the issue, but believes that medical examiners should be required to notify prosecutors and juries when a manner of death determination was influenced by nonmedical information.

Some have suggested the U.S. adopt one of the less adversarial models of expert testimony used in Europe, but such a radical departure from the current system seems unlikely. Instead, Dror and other critics who study cognitive bias have pushed for a process called linear sequential unmasking, in which a medical examiner first conducts an autopsy and makes only medical observations, then carefully factors in any external information as necessary. Advocates of this process say it separates objective, medicine-based observations from the more subjective conclusions.

Dror says for this reason he was especially baffled by one critic of his study, in particular: Andrew Baker. “It’s interesting that Baker would be one of our biggest critics,” Dror says. “The most charitable interpretation of his actions in the Floyd case is that he did exactly what I recommend.”

At Chauvin’s trial, Baker testified that he avoided watching the videos of Floyd’s death because he didn’t want to bias his conclusions. “It wasn’t until after he saw the video that he concluded Chauvin’s actions contributed to Floyd’s death. He made preliminary medical observations first, then factored in the nonmedical information as necessary,” Dror says. “This is exactly how it should be done.”

The ease with which the deaths of young Black men like Cameron and McClain were initially dismissed—and with which others continue to be dismissed today—evokes the ugly racial history of death investigation in the U.S., one in which clear lynchings were classified as “undetermined,” “accidental,” “suicide,” or “natural causes,” rulings that foreclosed any further criminal investigation.

When I first saw the coroner’s report speculating that Elijah McClain’s death might have been caused by an “undiagnosed mental illness,” I recalled a particularly chilling quote from one of those cases, which a young Black woman gave to the journalist John Howard Griffith back in 1954. “We couldn’t even count the bullet holes in my brother’s head,” she said. “But they called it heart failure.”