Dr. Joel Zivot stared at the autopsy reports. The language was dry and clinical, in stark contrast to the weight of what they contained — detailed, graphic accounts of the bodies of inmates executed by lethal injection in Georgia.

It was 2016, and the autopsy reports had been given to him by lawyers representing inmates on death row. He had received simple instructions: Interpret the levels of an anesthetic in the blood to determine whether the inmates were conscious during their execution. As an anesthesiologist at Emory University Hospital in Atlanta, Zivot specialized in reading these levels. But as he looked beyond the toxicology reports, something else caught his eye. The lungs were way too heavy.

He checked another autopsy. Again, heavy lungs. The average human lung weighs about 450 grams. Many of these lungs weighed twice that, sometimes more. His best guess was that they were filled with fluid — but he needed a second opinion.

His colleague Mark Edgar, an anatomical pathologist at Emory, agreed to help. Zivot didn't mention the lungs at all, to see if Edgar would catch the same aberrations. He did. And he confirmed that Zivot's hunch had been correct — the lungs were filled with a mixture of blood and plasma and other fluids.

It was a severe form of a condition called pulmonary edema, which can induce the feeling of suffocation or drowning.

Maybe it was a fluke? Zivot and Edgar needed more autopsies to be sure. Lawyers in other states shared autopsies of former clients who had been executed. The evidence explained why multiple inmates in recent years had gasped for air after their executions began.

Eventually, Zivot and Edgar found pulmonary edema occurring in about three-quarters of more than three dozen autopsy reports they gathered.

"The autopsy findings were quite striking and unambiguous," says Zivot. He had imagined that lethal injection induced a quick death and would leave an inmate's body pristine, or at least close to it. But the autopsies told another story.

"I began to see a picture that was more consistent with a slower death," he says. "A death of organ failure, of a dramatic nature that I recognized would be associated with suffering."

In some cases, there was even froth and foam in the airways: "Frothy fluid present in the lower airways," read one report.

The froth was a clue: It meant that the inmates were still alive and trying to breathe as their lungs filled with fluid, because froth could form only if air was still passing through the lungs. It also meant that the pulmonary edema was being caused by the first drug given during a lethal injection, since the second drug, a paralytic, stops the inmate's breathing altogether.

Most states use three drugs during a lethal injection: The first is supposed to anesthetize inmates; the second paralyzes them; the third stops the heart.

"How do we ask an inmate whether or not they experience their own death as cruel?" says Zivot. "Here was, to my mind, the beginning of a piece of evidence that has been critically absent."

Zivot and Edgar brought their findings of pulmonary edema to federal courts in Georgia, Arkansas, Missouri, Tennessee and Ohio. That evidence is now at the forefront of constitutional challenges to the death penalty in the United States. It has even made its way to the Supreme Court, where lawyers for inmates on federal death row have used autopsies to argue that lethal injection protocols constitute cruel and unusual punishment under the Eighth Amendment.

Now, an NPR investigation has expanded the scope of this evidence of pulmonary edema significantly. A review of more than 200 autopsies — obtained through public records requests — showed signs of pulmonary edema in 84% of the cases. The findings were similar across the states and, notably, across the different drug protocols used.

Doctors who spoke with NPR about the findings also raised serious concerns that many inmates are not being properly anesthetized and are therefore feeling the suffocating and drowning sensation brought on by pulmonary edema. The findings come at a time when death penalty states are already facing scrutiny over drug shortages, untrained execution personnel and a series of high-profile botched executions.

"These autopsy reports show definitively without question that these inmates are developing pulmonary edema," says Allen Bohnert, a federal public defender who represents Ohio inmates with upcoming executions. "That evidence continues to build and continues to get better every time another execution happens, unfortunately."

Bohnert first came across Zivot and Edgar's findings in the summer of 2018, when Edgar testified in Tennessee at a federal court hearing. Tennessee was about to use a drug called midazolam in an upcoming execution. Edgar brought autopsies from all 32 inmates who had so far been executed using midazolam and showed the court that 87% of them had developed pulmonary edema.

Bohnert watched the hearing and contacted Edgar for help with his own case. Ohio was about to execute an inmate named Robert Van Hook with midazolam. Bohnert asked if Edgar could come — this time, to do the autopsy himself. Edgar agreed.

Van Hook wheezed and gasped for air as he was executed on July 18, 2018, with a three-drug cocktail beginning with midazolam. The next day, Edgar performed an autopsy on the body at the county coroner's office in Dayton.

"Bloody froth fills both main stem bronchi," he wrote in his report, referring to the two large airways that enter each lung. He also found "frothy fluid" deeper inside the lungs, which were heavier than usual.

Edgar added the Van Hook autopsy to his existing findings and presented them at a federal court hearing in the Southern District of Ohio in December of 2018.

He told the court that, based on his examination of Van Hook's body, inmates executed using midazolam "would experience severe respiratory distress with associated sensations of drowning, asphyxiation, panic and terror."

Magistrate Judge Michael Merz deliberated for a month, then wrote his decision.

"All medical witnesses to describe pulmonary edema agreed it was painful, both physically and emotionally, inducing a sense of drowning and the attendant panic and terror, much as would occur with the torture tactic known as waterboarding," he wrote.

For the first time, a federal judge ruled that pulmonary edema, as shown in autopsies, reached the Supreme Court's standard for cruel and unusual punishment and that it "certainly or very likely causes severe pain and needless suffering."

Citing the decision, Ohio Gov. Mike DeWine delayed upcoming executions and instructed the Department of Rehabilitation and Corrections to reassess the state's lethal injection protocol. Ohio state officials declined to be interviewed for this article.

For Bohnert and his team of federal public defenders, the autopsies were becoming a powerful tool to get around a legal problem inherent in death penalty cases:

"We can't ask the clients what is happening to them during the course of their executions," he says. "This was the first time that we had a collection of autopsy data that allowed us to say what actually is happening from a scientific standpoint."

"Water on the banks"

The legal battle currently unfolding in federal courts hinges on two interconnected questions. First, why are lethal injection drugs causing pulmonary edema? And second, how much pain can inmates feel as their lungs fill with fluid?

Outside the execution chamber, pulmonary edema can develop for a number of reasons, including sepsis, congestive heart failure, or even intense exercise at high altitudes.

But according to doctors who spoke to NPR — and others who have testified in federal court — inmates develop pulmonary edema during lethal injection for a different reason: Extremely high doses of drugs, given quickly, are directly damaging the delicate architecture of the lungs. It's a phenomenon often seen in fatal heroin overdoses.

"In the '70s it was a very common way for a drug addict to just die after self-injecting heroin," says Philippe Camus, a pulmonologist in Dijon, France.

Camus has spent decades studying and compiling the various ways that drugs can negatively affect the lungs. He says that when a high dose of drugs is rapidly injected into the body, it pushes a concentrated "front" through the bloodstream. Doses vary slightly by states, but many inmates receive 500 milligrams of midazolam; for comparison, in a hospital setting patients may receive 1 or 2 milligrams.

"The quicker the injection, the denser the front, and the higher the risk of causing damage," Camus says.

Specifically, that concentrated front of drugs damages the thin barrier between blood vessels and air sacs in the lungs. Jeffrey Sippel, a pulmonologist who reviewed autopsies obtained by NPR, likens this phenomenon to a river flooding its banks.

"Water is supposed to be in the river, and the banks are supposed to be dry," he says. In this case, the dry banks are the lungs' air sacs, and the river is a network of capillaries; in healthy lungs, they are separated by a thin membrane. "When there is pulmonary edema, that normal relationship is awry. There's water on the banks where it doesn't belong."

When that membrane breaks, fluid from the capillaries enters the air sacs, impeding one's ability to breathe.

"It would be a feeling of drowning, a feeling of suffocation — a feeling of panic, imminent doom," says Sippel.

The increase in fluid also causes the lungs to become heavy. The average human lung weighs between 400 and 450 grams; the inmate autopsies obtained by NPR showed average lung weights of 813 grams for the right lung and 709 grams for the left lung. Some surpassed 1,000 grams each.

So far, the legal fight over pulmonary edema has focused on the way that the drugs midazolam and pentobarbital cause this damage to the lungs during a lethal injection. However, the autopsy reports obtained by NPR show high rates of pulmonary edema in inmates executed with other drugs, too — including sodium thiopental, which was used in hundreds of executions before its manufacturer stopped producing it in 2011.

"There are 214 drugs which can produce pulmonary edema even when used at a normal therapeutic dosage," says Camus, referring to drugs he has tracked through an online database. He says that the method of injecting drugs rapidly at high doses is more problematic than the pharmacology of any individual drug — and that states may have a hard time finding any drugs that won't cause pulmonary edema in a lethal injection setting.

"If we increase a dose of almost any medication and give it intravenously, you may have a thousand drugs capable of causing pulmonary edema when given at a higher than therapeutic dosage."

Lawyers representing the state of Ohio have admitted that other drugs used in lethal injection — besides midazolam — cause pulmonary edema.

"This is part of what happens," said Anne Strait, an Ohio assistant attorney general, at a court hearing in September of last year. "And the evidence for that is the fact that it also happens in pentobarbital executions, and it also happens in thiopental executions."

Courts are also debating how much inmates can feel the suffocating effects of their lungs filling with fluid during a lethal injection.

The first drug given in the lethal injection cocktail is supposed to anesthetize inmates — and yet midazolam, which has been used in dozens of executions in eight states, does not block pain.

"The ability of the drug midazolam to produce an anesthetic state is reliably zero," says David Lubarsky, CEO of UC Davis Health and the former chair of the anesthesiology department at the University of Miami. He has also testified in court cases warning about the use of midazolam as an anesthetic. "Its use as a sole anesthetic agent in the lethal injection protocol would amount to malpractice if it were a medical application."

In Ohio, expert witnesses testifying on behalf of the state have been insistent that midazolam can knock inmates out to the point where they wouldn't feel the pain associated with pulmonary edema. Midazolam is a short-acting benzodiazepine that does induce some sedation — drowsiness or sleep — but doesn't stop a patient from feeling pain or being awoken by intense stimulation.

"Believe me, the drug companies that invented and made it would have loved for it to have been granted by the FDA the moniker as a general anesthetic," says Lubarsky. "But it didn't get that, because it doesn't do that."

David Greenblatt, one of the doctors who helped develop midazolam in the late 1970s, even testified in Ohio that midazolam was insufficient to render inmates unable to feel pain.

For years now, problems with using midazolam in lethal injections have been clear. In 2014, Arizona inmate Joseph Wood gasped and snorted for nearly two hours before he died; that same year, in Oklahoma, Clayton Lockett writhed on the execution table for 33 minutes until he died of a heart attack. Additionally, inmates in multiple executions in recent years have shown signs of pain — gasping, heaving against restraints, choking — after being administered midazolam.

Lubarsky warns that if the first drug isn't anesthetizing the inmate, then they're likely to feel not only the suffocating sensation of pulmonary edema, but also the pain of the third drug: potassium chloride.

"It's like a burning cocktail coursing through your veins," says Lubarsky, referring to potassium chloride. "Once it reaches the heart, it stops the heart, and you do die. But in the process there is a period of just intense and searing pain."

Lubarsky also questions whether the other drugs historically used in lethal injections — barbiturates — are able to properly anesthetize inmates.

In 2005, he co-authored a study in The Lancet that looked at the postmortem toxicology of inmates executed using sodium thiopental — a barbiturate that is used as a general anesthetic.

"What we found is a very large percentage of executed inmates did not have a sufficient level of anesthesia in their blood at death in order to assure that they were asleep and not feeling the pain of the rest of the process," says Lubarsky.

The study found that 43 of the 49 inmates whose autopsies they reviewed had postmortem sodium thiopental concentrations in their blood below the level required for surgery. Lubarsky and his co-authors concluded that the stark differences between a surgery and an execution were to blame: In an operating room, an anesthesiologist will continuously deliver a "maintenance dose" of anesthesia to keep a patient unconscious, which inmates don't receive.

A review of the autopsy findings obtained by NPR shows that nearly all of the inmates executed using sodium thiopental had blood concentrations below that level required for surgery, too.

However, the 2005 findings proved controversial.

"You can't take these postmortem drug levels at face value," says Derrick Pounder, a forensic pathologist at the University of Dundee in Scotland. In 2005, he co-wrote a letter to The Lancet disputing the conclusions that Lubarsky's team had reached.

The problem, they said, was that too much time had elapsed between death and when the autopsy was performed — sometimes as much as 24 hours. During that time, the level of a drug in the blood can change drastically as it moves in and out of the bloodstream.

"You're not taking into account this massive redistribution that would explain why these levels are so low," says Pounder, who concluded that the levels must have been higher during the execution.

The disagreement shows the limitations of autopsies. Unless blood is drawn immediately after death, it's difficult to obtain definitive data about whether the inmate was conscious.

But there's a bigger problem with determining whether an inmate is conscious or not. In many states, the inmate is given a paralytic — right after the anesthetic — that masks any signs of pain.

"You can be paralyzed and totally awake," says Susi Vassallo, a toxicologist and professor of emergency medicine at New York University. She says that in a hospital setting, paralytics aren't used because they eliminate the doctors' ability to monitor their patients.

"We never paralyze the patient, because we need to look at them. We need to see if their face shows any pain. We need to make sure they're unconscious," Vassallo says. "But if they're paralyzed, we don't know anything. They could be having a seizure. They could be screaming. Whatever they're doing, we don't see or hear anything."

Eighteen states — including Ohio — use a paralytic in their lethal injection protocol.

"Severe pain and needless suffering"

At 90 years old, Norman Stout has no patience for courtroom deliberations about pain and suffering: "I just consider that stupidity."

Still, Stout follows Ohio's legal battles closely for one reason: He has been waiting 36 years to see the execution of the man convicted of murdering his wife.

On the night of May 14, 1984, Stout was at home in New Concord, Ohio, with his wife, Mary Jane. It was a Sunday, the one day of the week when he wasn't away from home, working on a construction crew building stretches of Interstate 70.

Just after dinner, two men came to the back door. They said their car had broken down, and they asked to use the phone.

"I thought nothing of it," says Stout.

Once inside, the men drew guns and ordered Norman and Mary Jane into a bedroom. Norman lunged at one of the intruders and was shot twice in the head. He managed to keep fighting and stayed conscious just long enough to hear the shots that killed Mary Jane.

"I heard four shots," he says. "I could shut my eyes, but not my ears."

Stout woke up in the hospital three weeks later. He could barely move the left side of his body, and the bullets had left five pieces of lead in his head — which remain there to this day.

The intruders were eventually caught, in Texas, and brought back to Ohio for a trial. In early 1985, one of the men, John Stumpf, was sentenced to death for the murder of Mary Jane Stout. But the execution itself would be delayed numerous times over the next three decades by a series of appeals and court decisions. Stumpf was scheduled to be executed this April, but in February, the governor of Ohio delayed it again, until next year.

"What is the system that takes 35 years? I don't know," says Stout. "How can you get away with 35 years of doing nothing?"

For now, the federal courts aren't standing in the way of John Stumpf's execution. After the district court in Dayton ruled that the pain caused by pulmonary edema would violate the Eighth Amendment, the 6th Circuit overturned the decision. The judges wrote that, because hanging is still technically permitted under the Constitution, suffocation of any kind is permitted, too: "It follows that Ohio's use of midazolam — which could cause pulmonary edema, i.e., suffocation — is not constitutionally inappropriate."

"Judges may vary on how one views pain, period," says Deborah Denno, a law professor at Fordham School of Law who has studied lethal injection for more than two decades.

Judges must decide whether the execution method in question will result in "severe pain and needless suffering" — the standard for cruel and unusual punishment set by the Supreme Court in 2015.

The Supreme Court has not yet ruled on whether the pain associated with pulmonary edema violates this standard. However, in July, lawyers for inmates on federal death row presented evidence of pulmonary edema before the court — and justices allowed the executions to proceed.

Denno says it's a standard that is extremely subjective and may be interpreted differently judge by judge.

"Somebody might think that suffocation certainly constitutes pain," Denno says says. "And another judge may think it's not — that suffocation alone doesn't qualify as severe pain and needless suffering."

But despite federal courts giving Ohio the green light for upcoming executions, it's unclear whether the state will carry them out. Over the past year and a half, Gov. Mike DeWine has delayed every scheduled execution, citing concerns about both pulmonary edema and a shortage of drugs.

Norman Stout sees the delays — and the courtroom arguments about reducing suffering — as a misplaced process never afforded to his wife before she was murdered.

"Cruel and unusual punishment is lying out there in the cemetery," he says. He believes that Ohio should use the firing squad instead. "There is absolutely no suffering whatsoever. He doesn't know what hit him. He's dead."

It's the exact same argument also being made by Allen Bohnert — one of the lawyers for John Stumpf, the man convicted of killing Norman Stout's wife.

"The inmate would be dead before his brain would ever have the chance to register the sound of the report of the rifle," says Bohnert. In court, he and his team have presented the firing squad as a readily available alternative to lethal injection. Supreme Court precedent requires them to offer an alternative. It's an extreme measure, but it's one that eliminates the risk of pulmonary edema.

"They significantly reduce the risk of severe pain," he says, "when we compare to the risks posed by the current protocol."

The last firing squad execution in America took place in Utah in 2010. Yet inmates in Georgia, Tennessee, Alabama and Virginia have also requested the firing squad over lethal injection in the past few years — and lawmakers in multiple states have introduced legislation to bring the firing squad back.

This push to return to older methods of executions comes on top of a litany of problems for death penalty states — drug shortages, highly publicized botched executions, and now concerns over pulmonary edema. As a result, states are performing fewer and fewer executions every year. In 2019, there were only 22 executions in seven states.

Yet despite the accumulation of legal challenges to lethal injection, public support is still behind capital punishment — a nationwide poll in October showed 56% of Americans in favor of the death penalty.

Methodology

For this reporting, NPR obtained 305 autopsy reports of inmates executed in nine states between 1990 and 2019. Nearly all autopsies were gathered through public records requests; several were taken from public evidence submitted in federal court cases. The states included Alabama, Arizona, Arkansas, Florida, Georgia, Ohio, Oklahoma, Tennessee and Virginia.

NPR consulted with pathologists, pulmonologists and anesthesiologists to interpret the autopsies. Lung analysis was based on the internal examination notes written by medical examiners, and, when available, microscopy provided in the reports. Prevalence of pulmonary edema was determined, in consultation with doctors, by the presence of terms such as "pulmonary edema," "edema/edematous," "froth," "foam" and "congestion."

Since the autopsies spanned decades in multiple states, not all contained the same level of detail. Only 216 autopsies contained an internal examination of the lungs, so the prevalence of pulmonary edema was based on that sample. Other autopsies contained toxicology reports, but not internal examinations.

Not all states conduct autopsies after lethal injections. Texas, which has performed by far the most lethal injections of any state, has a policy of not conducting autopsies. When asked by NPR about this, a spokesperson for the Texas Department of Criminal Justice said, "We know how they died."

Copyright 2020 NPR. To see more, visit https://www.npr.org.