In November of 2022, a twenty-three-year-old woman named Shania Arms posted a video on TikTok. The footage shows her at a photo shoot, wearing jean cutoffs, black heels, and a crown of white flowers, lounging under palm trees. A building behind her with tall, rounded windows could be a hotel. But next to her, in every shot, is a metal trolley stacked with medical equipment. Two large plastic tubes, one cherry red, the other dark plum, loop from the trolley to Arms, entering her body beneath her clavicle. She has cystic fibrosis, a condition that damages the lungs. Hers are failing, and the machine on the trolley has replaced them.

The machine, called ECMO, which stands for extracorporeal membrane oxygenation, removes carbon dioxide from the blood and replaces it with oxygen. It can perform the work of the heart and lungs entirely outside the body. When Arms made the video, she had been on ECMO for forty-seven days, living in the I.C.U. of an Orlando hospital—the building behind her in the video—hoping for a lung transplant. Without ECMO, she would die. But owing to the complexity of the machine and its attendant risks—catastrophic bleeding, stroke, infection, malfunction—she couldn’t leave. She was waiting, stuck in a kind of limbo between life and death.

Although ECMO technology is more than half a century old, it was in some ways still considered an experimental treatment until recently. ECMO devices began to be used to replace lungs wrecked by COVID-19; stories began to spread of incredible recoveries after as long as a hundred and forty-nine days on ECMO. Family members of COVID patients called dozens of hospitals, hunting for the machines.

But some physicians worry that ECMO is creating entirely new ethical conundrums. “The unfortunate reality is that, sometimes, people get put on this machine and they don’t get better,” Jessica Zitter, a palliative- and critical-care physician, told me. A patient whose heart has stopped could potentially live on the machine for months, awake, able to walk and read the newspaper. But he might never leave the I.C.U. “It’s a trap,” Zitter said. ECMO is transforming medical care, saving lives. But it also complicates care when life inevitably begins to end, committing some patients to a liminal state with no hope for recovery. When should it be used, or withheld? And who should decide?

The beginnings of ECMO date back to 1931, when a surgical resident in Boston named John Gibbon cared for a woman with a pulmonary embolism—a blocked artery in the lungs, which made it harder for her blood to receive oxygen and ferry it to the rest of her body. Gibbon had no treatment, and could only watch her decline and die. “During that long night . . . the idea naturally occurred to me that if it were possible to remove continuously some of the blue blood . . . put oxygen into that blood . . . and then to inject continuously the now-red blood back into the patient’s arteries, we might have saved her life,” Gibbon later wrote. “We would have bypassed the obstructing embolus and performed part of the work of the patient’s heart and lungs outside the body.” The woman was gone by morning, but Gibbon set to work. By 1934, he had developed a machine that could support the circulation of a cat for thirty minutes. In 1952, it was finally ready for humans. Gibbon used the machine to replace the heart and lungs of a college student while he operated on her heart, repairing a cardiac defect. She survived; this event marked the advent of modern open-heart surgery.

Gibbon’s bypass circuit soon became an integral part of cardiothoracic surgery. But the technology relied on the direct mixing of oxygen and blood—a process that rendered the blood toxic after about an hour. Gibbon’s system left too much oxygen floating free in the bloodstream. In 1965, Robert Bartlett, another surgical resident in Boston, set out to solve this problem. Bartlett built a “membrane oxygenator” out of silicone, which mitigated oxygen’s noxious effects by allowing it to diffuse slowly into the blood across a semipermeable membrane, essentially in the same manner that it travels across the thin walls of the alveoli in our lungs. This gave the hemoglobin in the patient’s bloodstream time to absorb the oxygen. Within three years, his devices were keeping animals alive through “extracorporeal circulation” for up to four days. In 1971, his colleagues used a similar setup to support a man with respiratory failure. He lived on the machine for thirty-six hours, becoming the first human survivor of ECMO.

In 1975, Bartlett and his associates used their device to sustain a newborn who developed respiratory failure after birth. Within fifteen years, hundreds of babies with similar problems had been saved with ECMO, with a survival rate of eighty per cent. The technology became standard in the largest pediatric centers. But a trial with adult patients, in 1979, found a mortality rate of ninety per cent: it seemed that using ECMO in adults turned out to be far more difficult, because they often have multiple medical problems, and their illnesses are more complex. The adoption of ECMO for adult patients largely stalled, until H1N1 influenza emerged in the Southern Hemisphere, in 2008. “The Australians were reporting that they were having an epidemic, and the only thing that was improving outcomes was ECMO,” Bartlett told me. The technology had advanced, and doctors were better at using it; the machines now kept patients alive while their bodies fought the infection. A 2009 trial found that ECMO helped with respiratory failure from other causes. The number of hospitals offering ECMO increased threefold within ten years. And then the coronavirus pandemic arrived.

On November 20, 2020, Henry Garza came down with a headache. He was fifty-four, living in Illinois with his wife, Michele, and four children. His COVID test was positive, and he felt short of breath. “People who go to the hospital don’t come home,” Garza told his wife, but he went anyway. A month later, he was still in the hospital, and his lungs were getting worse. He was wary of intubation—he’d heard that many patients didn’t survive for long after they’d been put on ventilators—and a nurse told him about ECMO, which would let his lungs heal by sparing them the work of breathing while inflamed by a viral infection. Michele recalled that ECMO was presented as “either a bridge to healing or a bridge to transplant.” She remembers thinking, “If it’s a bridge, let’s get on it.”

A ventilator, like many other medical technologies, can sometimes harm as well as help. It assumes the work of breathing and controls it, but forcing air into damaged lungs can damage them further. Early in the pandemic, the mortality rate for ventilated COVID patients was as high as sixty per cent. In the spring of 2020, physicians at N.Y.U. treated thirty ventilated COVID patients with ECMO. Ninety per cent survived. Later that year, a study, published by The Lancet, of more than a thousand COVID patients treated with ECMO found a sixty-per-cent survival rate. In 2022, at Vanderbilt University, physicians studied a group of ventilated COVID patients whose doctors suggested that they be put on ECMO. Thirty-five were approved, and twenty survived. Among the rest, the survival rate was ten per cent.

Jeffrey DellaVolpe, an intensivist in Texas, began to wonder if skipping the ventilator entirely might be better. From fifty-two COVID patients that he and his colleagues treated with ECMO, twelve were placed on it without going on a ventilator first. His team found that, although the average survival rate for ventilated patients with ECMO was around fifty per cent, it was seventy-five per cent among the ECMO-first patients. DellaVolpe speculated that one reason the ECMO-first patients had more positive outcomes was that they could be awake. Since ECMO doesn’t require sedation, as ventilators do, he could get them out of bed to work with physical therapists. “As I read the tea leaves, I think this is going to be a big part of the I.C.U.,” DellaVolpe told me. In theory, ECMO could largely replace the use of ventilators for respiratory failure. Today, I.C.U.s are often hushed and sepulchral; perhaps they’ll one day be full of walking, talking patients.

Garza was on ECMO for a hundred and nineteen days. He came to know the machine so well that he could tell when the oxygenator needed to be changed. He hallucinated being caught in a spiderweb, and sometimes dreamed that he was being held against his will. “It was this feeling of being trapped all the time,” he said. His lungs didn’t recover, but ECMO allowed him to be wait-listed for a transplant. He asked his nurses, “Can I just wait at home?” Their answer was blunt. “If we take you off these machines, you will die. And you can’t take them home with you.” Finally, in April, 2021, he got new lungs.

Jon Marinaro, an emergency physician and intensivist at the University of New Mexico, is pushing the use of ECMO further. Every year, nearly four hundred thousand Americans suffer cardiac arrest outside of a hospital. Despite the use of CPR, defibrillators, and powerful drugs, fewer than one in ten survive. “That’s where ECMO comes in,” Marinaro told me. In a small room crowded with ECMO pumps, next to his hospital’s pediatric I.C.U., Marinaro showed me how to place an ECMO tube, or cannula, on a model he’d built out of white PVC pipe. Hand-drawn thank-you cards from a class of first graders were tacked to a cabinet behind him: Marinaro and his team had saved their teacher with ECMO. “If you can do better than CPR, you’re going to save more lives,” he said.

In France, emergency responders have been using ECMO to treat cardiac arrest since 2011, placing patients with heart attacks onto the machines, whether in the Louvre or on the subway. In 2014, Demetris Yannopoulos, a cardiologist at the University of Minnesota, started a similar program in Minneapolis. What he found surprised him. “We could save almost half” of the patients, he told me. He paused. “I wasn’t expecting that kind of survival.” In a randomized study of cardiac-arrest patients, he found that forty-three per cent of those treated with ECMO and CPR survived; with CPR only, only six per cent did.

Marinaro initially became interested in ECMO as a way of saving patients with pulmonary embolisms—the same disease that inspired John Gibbon, in 1931. In 2016, four patients at his hospital died from it. The next year, Marinaro and his team started a program to treat such patients with ECMO. Since then, they have not had a single death in patients treated promptly. He soon began using ECMO to treat cardiac arrest, too—the machine can take over for the heart until that organ is re-started—and he became a convert. Most of his patients have had about fifty minutes of CPR before they make it to ECMO. “As an E.R. doc yourself, you know what that means,” he said to me. I work in an emergency department in New Mexico, and, in my experience, after an hour of CPR with no improvement, the possibility of survival is almost nil. I would likely pronounce those patients dead.

And yet, after this point, about thirty per cent of Marinaro’s patients survive. “If you were to triple survival with a cancer drug, people wouldn’t believe it,” he said, while a humming ECMO pump whisked water through his PVC model. “But we have tripled survival for out-of-hospital cardiac arrest. In medicine, tripling survival is unheard of!” He told me about one patient who received CPR for three hours, and was taken to two different hospitals, before Marinaro’s team put him on ECMO. “That guy walked out of the hospital,” he said. Recently, he treated a sixteen-year-old girl named Sophia who collapsed in cardiac arrest in a Starbucks bathroom. She received CPR for an hour and nine minutes. “They were about to quit,” he told me. Her mother, a physician named Angelina Villas-Adams, waited in the E.R. with her husband as the team worked on their daughter; a chaplain came by. Marinaro’s own protocols recommend against deploying ECMO after more than an hour of CPR, because neurologic recovery becomes very unlikely. But then a paramedic said he had seen her move. Sophia had exhibited something called CPR-induced consciousness; the chest compressions were circulating enough blood for her to move an arm. Marinaro made an exception and put Sophia on ECMO. Later, in the I.C.U., someone whispered “Squeeze my hand” into her ear, and she did. “We’re thinking, Is she going to wake up?” her father, Buddy, told me.

The ECMO pump, he said, reminded him of a red Slurpee machine. After two days, Sophia’s heart recovered, and she came off ECMO. Then she spoke: “My tootsies are squished.” Pneumatic boots, meant to prevent blood clots, were squeezing her feet. “ECMO saved my life,” Sophia told me when I met her in Albuquerque, several months later. “I lived when I should have died,” she said.

“We should be doing this way more,” Marinaro said. He thinks that ECMO should be everywhere. “How do you make ECMO available to everybody so that you can die anywhere, anytime, and someone will be able to save you?” In the back of an ambulance, he showed me a hand-crank ECMO circuit that he and another physician, Darren Braude, had pioneered for cardiac arrests in the field. “We do Crank-MO!” he said, grinning. Hunched over in the back of the ambulance, I turned the crank. It whined like a Hot Wheels car. The setup was astonishingly simple. As we stepped out of the ambulance, Marinaro suggested sending a few such circuits to the rural hospital where I work. I thought of all the patients we could save.

Around a decade ago, a teen-ager who couldn’t be saved was admitted to a New England hospital. Like Shania Arms, he had cystic fibrosis. A previous lung transplant was failing, and his only hope was another transplant. He was put on ECMO while he waited. Two months later, doctors discovered that he had developed an incurable cancer. Now there was no way for him to leave the I.C.U. His lungs were beyond recovery, and the cancer made him ineligible for transplant. He was caught on a bridge to nowhere.

Some members of the medical team thought that ECMO should be stopped. Transplant was no longer possible, and ECMO machines were scarce. As long as the patient was on the machine, it couldn’t be used to save someone else. It’s also expensive; according to a 2023 study, the median hospitalization charge for COVID patients on ECMO was around eight hundred and seventy thousand dollars, and prolonged cases can exceed several million. These resources might be needed to help other patients, and the boy couldn’t live in the I.C.U. indefinitely.

But others on the team disagreed. “He was texting with his friends,” Robert Truog, a pediatrician and bioethicist who was involved with and wrote about the case in The Lancet, said. He was spending time with family, and doing homework online. Because he could be awake on ECMO, he could still engage in activities that were meaningful. Situations like this represent a “profound ethical dilemma,” Raghu Seethala, an intensivist and ECMO specialist at Brigham and Women’s Hospital, told me. “The technology is ahead of the ethics,” another expert said.

According to Truog, those in favor of continuing the treatment pointed out that patients are routinely kept alive on devices such as ventilators, or through treatments such as dialysis, without expectation of recovery. Why was ECMO any different? “It is different,” Kenneth Prager, the director of clinical ethics at Columbia University Irving Medical Center, told me. Patients can sometimes utilize those machines from home, whereas ECMO requires an I.C.U. Meanwhile, I.C.U. patients with severe organ failure are rarely fully conscious; with ECMO, “you can have a patient who is awake and alert, walking around, riding a stationary bicycle, and yet their heart and lungs are incapable of supporting life,” Prager said. “The contrast is totally unmatched by any other technology.”

It’s this striking likeness to life, when a patient is upon the very threshold of death, that makes the situation created by ECMO so startling and difficult. It is a “precarious form of existence, which sometimes raises questions about our traditional definitions of life and death,” Prager and his colleagues have written, in a paper published by the journal Circulation. Using ECMO, a patient with severe organ failure may look well, and feel well, but without recovery or a transplant, he cannot leave the I.C.U. and has no hope of long-term survival. The conversation about when to stop involves the patients themselves, who must contemplate choosing the moment of their own demise. Whoever decides, it’s like delivering a death sentence, Prager said.

In the end, the team in charge of the teen-age boy’s care presented a compromise to the family. Rather than actively withdrawing ECMO—literally flipping a switch to “off”—they would no longer maintain the machine. When the oxygenator began to wear out, they chose not to swap in a replacement. One day, as it failed, the boy gently lost consciousness and died. In other instances, however, patients have refused to consider any de-escalation in their care. Those patients lived in the I.C.U. for as long as they could, until a complication killed them. Here, too, ECMO introduces new complexities. A ventilator, or dialysis, staves off a cascade of deterioration that ultimately causes cardiac arrest—the final common pathway of death. And, once cardiac arrest occurs, it is often difficult to reverse. But ECMO can take over the work of the heart completely, pausing the usual progression toward death. Cardiac arrest becomes meaningless.

It seems almost certain that, the more we use ECMO to avert death, the more people will end up living in I.C.U.s, kept alive solely by the machines. “This is a big problem already,” Prager told me. “It’s just going to get worse.” In part, this is because doctors cannot perfectly predict who will benefit from ECMO, or from the treatments it makes possible. “We put people on ECMO believing it will be a bridge to something, and it turns out not to happen,” Prager said, perhaps because they aren’t able to recover on their own, or are ineligible for a transplant. Patient expectations can also play a role. Seethala told me about a lawsuit brought against a team of doctors for not offering access to ECMO: In 2019, a jury found the doctors liable for not transferring a Bronx woman to an ECMO center. (After a 2021 appeal reduced her settlement, she was awarded ten million dollars for past and future pain and suffering.) Other observers have expressed concern that patients may come to expect ECMO as a default treatment for any cardiac arrest, as CPR is now.

Blair Bigham, another emergency physician and intensivist, told me that during the pandemic, he saw many hospitals acquire ECMO machines, then deploy them indiscriminately for patients who were unlikely to recover. “As soon as you have the ECMO button, you’re going to use it,” he told me. Jessica Zitter, the palliative-care physician, calls this tendency to always add another machine, or another drug, the end-of-life conveyor belt. “We’re always trying to do more and more and more to people, when maybe we shouldn’t,” Bigham said. “The overarching problem here is that we have this fear of letting people die,” he explained.

In March, 2023, I visited Robert Barlett’s lab, which moved in 1980 from U.C. Irvine, where he originally developed ECMO in the nineteen-seventies, to the University of Michigan. Ann Arbor was gray, but the lab was bright and full of activity. Bartlett, who is now eighty-four, was at home, recovering from a medical procedure. He called in during my visit to tell me what they were working on. One focus is on finding ways to get patients who are on ECMO out of the I.C.U., or even out of the hospital. “They need to be able to eventually go home with these devices,” Bartlett said.

To that end, the lab is trying to create a mini-ECMO—a kind of wearable artificial lung. Down the hall from the conference room where I spoke to Bartlett, a research intern named Gabriele Seilo sat at a lab bench, winding grooved pucks with a gauzy plastic fabric. The plastic fibres contain micropores, allowing them to act like lung capillaries, which introduce oxygen to the bloodstream in precise quantities. “It facilitates gas exchange,” Seilo explained, turning from his bench. Beside him, a bin held prototypes of different sizes. The idea is for the puck to be worn outside the body, tethered to a portable oxygen tank. Blood can flow through the device passively, so it doesn’t need a pump.

In another room, the lab was developing an ECMO membrane that secretes nitric oxide. ECMO oxygenators tend to create blood clots, and patients require medications to prevent this. But those medications can in turn cause dangerous bleeding. “Bleeding and clotting are the biggest problems with ECMO,” Bartlett said on the phone. Nitric oxide discourages clots from forming, but doesn’t cause bleeding. It could make ECMO far safer, and potentially suitable for use outside the hospital.

Down the hall, I found what may be the lab’s most astonishing innovation. Alvaro Rojas-Peña, the co-director of the lab, led me into an anteroom that adjoined an operating suite.

“Look at this,” Rojas-Peña said. He gestured to a pedestal in the middle of the room. I leaned forward. On the pedestal was an isolated, beating heart. Cannulas connected it to a special ECMO circuit. The heart, which had been harvested from a research animal, was swaddled in a plastic bag to prevent desiccation; it writhed and twisted with each beat, alive.

Rojas-Peña showed me an iPad, on which software tracked the heart using nearly a dozen physiological parameters. Incredibly, the heart was generating a normal blood pressure. With the help of ECMO, the heart had been kept alive and pumping, outside of any body, for almost twenty-four hours. In a few minutes, it would be transplanted into an animal in the operating room next door.

Receiving an organ transplant is like winning several lotteries at once. A donor must be healthy enough that the organ’s function is preserved, and close enough geographically that transport time is minimized; the organ must match the recipient’s blood type, and the recipient can’t be too ill to survive the operation. All of these conditions must align at the moment an organ becomes available; every day, seventeen people die in the U.S. because some part of this equation fails, or simply because organs can’t be found. “Seventy per cent of donor organs are rejected,” Rojas-Peña told me, because something isn’t optimal.

The implications of applying ECMO to organ transplantation could be profound. If organs could be reliably kept alive outside of the body, they could be sent to a centralized organ bank. Transport time would no longer be a factor. Organs could be perfectly matched to recipients, and marginal organs could be tuned up outside the body with medications. Wait lists could disappear. “Transplant would not be an emergency surgery anymore,” Rojas-Peña said. It could be planned, like any other operation. In Toronto, a group has already begun doing this with human lungs, and that has allowed them to utilize about seventy per cent of donor lungs for transplantation, compared with an average in the United States of about twenty per cent.

One of the risks of ECMO, as it’s employed today, is that it will strand patients on a bridge to nowhere. But the technology itself may be a bridge to somewhere new. Ultimately, Bartlett suggested, it may have uses beyond transplantation. “We could have organs that do things,” he has said, such as manufacturing clotting factors or red blood cells. “We could take out a liver that’s full of cancer, treat it, and then return it to the patient,” he told me. “We think all those things could be possible.” ♦