By 1:56 p.m., the intensive care unit had tried everything: aggressive CPR, four shocks to the chest, seven doses of adrenaline and two bags of fluids. But the 11-month-old girl lay still, her body in cardiac arrest. At 1:58 p.m., after two minutes flatlining without a pulse, she was pronounced dead.

“The family wanted a little time to just be with the patient,” saysLouis Daugherty, an associate professor of pediatrics at the University of Rochester Medical Center and a member of the team handling the case. After about 15 minutes, the mother asked for the breathing tube to be removed so that she could hold her daughter. And then, the team witnessed the unimaginable.

“Soon after the breathing tube was removed, she started to have spontaneous breathing. Her heart rate came back, her color improved and she had a gag reflex,” says Daugherty. “I had never seen anything like this.” Although the young girl’s condition stabilized, she succumbed to progressive heart failure in a chronic care facility four months later.

The girl had experienced a rare resurrection called the “Lazarus Phenomenon,” in which patients who appear to be clinically dead sometimes spontaneously return to life. While the majority of these patients eventually succumb to death’s grip, as many as a third make a full recovery. But according to several surveys, this marvel may be more common than most people suspect due to under-reporting tied to legal concerns.

For centuries, people have had anxieties about incorrect death pronouncement and premature burials. In the 1800s, the fear of being buried alive, known as taphophobia, was so widespread that many people included provisions in their wills calling for tests to confirm death, such as pouring hot liquids on their skin or making surgical incisions. Others were buried with crowbars and shovels. This paranoia eventually led to a new class of “safety coffins” with breathing tubes and a variety of flags, bells or pyrotechnics that would allow anyone buried prematurely to signal passersby.

Auto-resuscitation in hospitals wasn’t reported in medical literature until1982. Anesthesiologist Jack Bray, Jr.gave the phenomenon its monikerin 1993, based on the Biblical story of Lazarus of Bethany, who died and was resurrected by Jesus Christ four days later. Since then, though, the phenomenon has remained scarce in the scientific literature.

Vedamurthy Adhiyaman, a consultant geriatrician at Glan Clwyd Hospital in North Wales, became interested in the Lazarus Phenomenon after encountering it firsthand in the early 2000s. His team had conducted CPR on an elderly man in is his late 70s for about 15 minutes with no response.

“There isn’t any definite time frame for how long you should attempt CPR before you stop,” says Adhiyaman. “It really varies on a case by case basis.” Although Adhiyaman did not officially declare death immediately after stopping CPR, a member from his team told the family that the man had died. As it turns out, the situation was not that straightforward.

“After about 15 to 20 minutes, he started breathing,” recalls Adhiyaman. “But he remained unconscious in a coma for the next two days until he died on day three.”

The family believed that the CPR should not have been stopped and that the team had provided substandard care, so they took Adhiyaman to court. “It was around that time that I began researching this phenomenon, because I had to show evidence that these things do happen,” he says.

After scouring the medical literature, Adhiyaman unearthed 38 cases of Lazarus Phenomenon, which proved sufficient to demonstrate its legitimacy and exonerate him of negligence. In his2007 reviewof the subject, published in theJournal of the Royal Society of Medicine, Adhiyaman found that on average, these patients returned from death’s door seven minutes after stopping CPR, though close monitoring in many cases was inconsistent. Three patients were left unattended for several minutes, with one making it all the way to the hospital mortuary before being discovered alive.

While the vast majority of patients died soon after auto-resuscitation, 35 percent of them were eventually sent home with no significant neurological consequences. Adhiyaman’s analysis also showed that these positive outcomes were not really affected by the duration of CPR or the amount of time it took for patients to auto-resuscitate.

Coming back from the brink this way is undoubtedly rare. In 2010, a team at McGill University conducted anextensive reviewof medical literature and found just 32 cases of the Lazarus Phenomenon since 1982. That same year, a German team was able toround up45 articles on the subject. Many of the same cases appear in both reports.

A spattering of new cases has emerged since then. In 2012, a 65-year-old patient in Malaysia was found with a pulse 40 minutes after he was pronounced dead.In 2013, an 89-year-old woman in New Haven regained a pulse five minutes after resuscitation efforts were abandoned. And in 2015, two cases popped up—one in a 67-year-old man in Denmark and another in the 11-month-old girl in Rochester.

In addition, recent investigations suggest that the phenomenon may be underreported. A2013 studyindicated that nearly half of all French emergency room physicians claim to have seen a case of auto-resuscitation during their career, while according to a2012 survey, more than one-third of Canadian critical care doctors reported encountering at least one case.

It may be that doctors are not reporting it officially due to the embarrassing professional and legal consequences associated with a premature declaration of death. Adhiyaman also believes that many cases go unreported due to privacy laws.

“In order to publish a case report in the scientific literature, you need the consent of the family. And it’s going to be really hard to get them to agree when all trust between the medical profession and the family has been broken,” he says.

This all makes auto-resuscitation extremely difficult to study, and the exact mechanisms that produce the phenomenon remain speculative. Notably, though, all official reports of auto-resuscitation have one thing in common—the use of CPR.

One popular theory is dynamic hyperinflation, which can occur during CPR if the lungs are rapidly filled with air without adequate time to exhale. The increased pressure in the lungs could limit blood flow back to the heart and even inhibit the heart’s ability to pump altogether, producing cardiac arrest.

“When we breathe we suck in air, which creates negative pressure, whereas a ventilator [or CPR] blows in air, which creates positive pressure,” says Daugherty. “If someone has an abnormal heart that is not functioning normally, and then you add this pressure to the chest, it decreases the amount of blood that is being returned to the heart, which further impairs its function.”

In theory, when emergency doctors stop CPR, the lung pressure caused by dynamic hyperinflation returns to normal and the blood begins to circulate with greater ease, producing an auto-resuscitation effect.

Other researchers have proposed that dynamic hyperinflation instead plays a role in delaying drugs administered during CPR from reaching the heart. Once CPR is curtailed and blood flow returns to normal, the drugs reach their destination and may produce further improvements in circulation.

Hyperkalemia, or an elevated level of potassium in the blood, has also been proposed as a contributing cause in some cases of auto-resuscitation. These heightened levels interfere with heart function. After physicians prescribe calcium, glucose and insulin, sodium bicarbonate or other drugs that reduce potassium levels, the heart is able to resume beating.

While the nuts and bolts of the “Lazarus Phenomenon” remain an enigma, doctors can still take precautions to ensure that they don’t quit on a patient too early. Adhiyaman recommends that physicians notify family members that CPR has been stopped and then monitor the patient for at least 10 to 15 minutes before declaring death.

“Death is not an event, it is a process. It happens gradually as your organs start shutting down. And so unless you are absolutely certain, you should not certify death,” he says.

But in some situations, physicians are under time pressure and must draw a discrete line between life and death as quickly as possible—especially when it comes to organ donation and transplantation.

Thedead donor rule, which serves as the ethical standard for organ transplantation, states that “vital organs should only be taken from dead patients and, correlatively, living patients must not be killed by organ retrieval.” For organs to be transplanted successfully, they must be quickly removed to minimize any damage from lack of blood supply.

For brain-dead patients, the answer is simple: Keep them hooked up to a ventilator, which ensures circulation. But for patients who are donating after a cardiac death, doctors are put in the difficult situation of waiting long enough to ensure that a patient can be declared dead, but short enough to be left with viable organs that could save another life.

“There is an inherent tension, because the longer you wait, the more time the organs are not getting enough blood, which increases the likelihood that they go bad. So it cannot be too long,” saysJames Kirkpatrick, an associate professor of medicine and a member of the ethics consultation committee at the University of Washington School of Medicine. “But you also want to make sure the patient is not going to auto-resuscitate, because theoretically their heart and lungs are not irreversibly damaged and could come back.”

Right now, recommendations for wait times in cases of organ donation after a cardiac death vary significantly. The Institute of Medicine suggests at least five minutes, while the American Society of Transplant Surgeons and the Society for Critical Care Medicine each propose two minutes. A2012 study, for instance, closely tracked 73 potential organ donors after cardiac death. That research found no occurrence of auto-resuscitation after two minutes—but none of those patients had received CPR.

Also, adopting national guidelines may be challenging, because some people remain skeptical about auto-resuscitation. “Frankly, some people don’t really believe in it,” says Daugherty. “And so a couple of examples like this are not going to change everything in how physicians declare someone dead.”

In the meantime, advancements in life-sustaining medical technologies and resuscitation techniques have only added nuance and complexity—prompting further questions, such as at what point death, clinically speaking, becomes irreversible?

“Although this is such a rare phenomenon and it is poorly understood, a lot of caution still needs to be taken on when we should declare someone dead,”