The smell of death was overpowering the moment a relief worker cracked open one of the hospital chapel’s wooden doors. Inside, more than a dozen bodies lay motionless on low cots and on the ground, shrouded in white sheets. Here, a wisp of gray hair peeked out. There, a knee was flung akimbo. A pallid hand reached across a blue gown.

Within days, the grisly tableau became the focus of an investigation into what happened when the floodwaters of Hurricane Katrina marooned Memorial Medical Center in Uptown New Orleans. The hurricane knocked out power and running water and sent the temperatures inside above 100 degrees. Still, investigators were surprised at the number of bodies in the makeshift morgue and were stunned when health care workers charged that a well-regarded doctor and two respected nurses had hastened the deaths of some patients by injecting them with lethal doses of drugs. Mortuary workers eventually carried 45 corpses from Memorial, more than from any comparable-size hospital in the drowned city.

Investigators pored over the evidence, and in July 2006, nearly a year after Katrina, Louisiana Department of Justice agents arrested the doctor and the nurses in connection with the deaths of four patients. The physician, Anna Pou, defended herself on national television, saying her role was to ‘‘help’’ patients ‘‘through their pain,’’ a position she maintains today. After a New Orleans grand jury declined to indict her on second-degree murder charges, the case faded from view.

In the four years since Katrina, Pou has helped write and pass three laws in Louisiana that offer immunity to health care professionals from most civil lawsuits — though not in cases of willful misconduct — for their work in future disasters, from hurricanes to terrorist attacks to pandemic influenza. The laws also encourage prosecutors to await the findings of a medical panel before deciding whether to prosecute medical professionals. Pou has also been advising state and national medical organizations on disaster preparedness and legal reform; she has lectured on medicine and ethics at national conferences and addressed military medical trainees. In her advocacy, she argues for changing the standards of medical care in emergencies. She has said that informed consent is impossible during disasters and that doctors need to be able to evacuate the sickest or most severely injured patients last — along with those who have Do Not Resuscitate orders — an approach that she and her colleagues used as conditions worsened after Katrina.

Pou and others cite what happened at Memorial and Pou’s subsequent arrest — which she has referred to as a "personal tragedy" — to justify changing the standards of care during crises. But the story of what happened in the frantic days when Memorial was cut off from the world has not been fully told. Over the past two and a half years, I have obtained previously unavailable records and interviewed dozens of people who were involved in the events at Memorial and the investigation that followed.

The interviews and documents cast the story of Pou and her colleagues in a new light. It is now evident that more medical professionals were involved in the decision to inject patients — and far more patients were injected — than was previously understood. When the names on toxicology reports and autopsies are matched with recollections and documentation from the days after Katrina, it appears that at least 17 patients were injected with morphine or the sedative midazolam, or both, after a long-awaited rescue effort was at last emptying the hospital. A number of these patients were extremely ill and might not have survived the evacuation. Several were almost certainly not near death when they were injected, according to medical professionals who treated them at Memorial and an internist’s review of their charts and autopsies that was commissioned by investigators but never made public.

In the course of my reporting, I went to several events involving Pou, including two fund-raisers on her behalf, a conference and several of her appearances before the Louisiana Legislature. Pou also sat down with me for a long interview last year, but she has repeatedly declined to discuss any details related to patient deaths, citing three ongoing wrongful-death suits and the need for sensitivity in the cases of those who have not sued. She has prevented journalists from attending her lectures about Katrina and filed a brief with the Louisiana Supreme Court opposing the release of a 50,000-page file assembled by investigators on deaths at Memorial.

The full details of what Pou did, and why, may never be known. But the arguments she is making about disaster preparedness — that medical workers should be virtually immune from prosecution for good-faith work during devastating events and that lifesaving interventions, including evacuation, shouldn’t necessarily go to the sickest first — deserve closer attention. This is particularly important as health officials are now weighing, with little public discussion and insufficient scientific evidence, protocols for making the kind of agonizing decisions that will, no doubt, arise again.

At a recent national conference for hospital disaster planners, Pou asked a question: ‘‘How long should health care workers have to be with patients who may not survive?’’ The story of Memorial Medical Center raises other questions: Which patients should get a share of limited resources, and who decides? What does it mean to do the greatest good for the greatest number, and does that end justify all means? Where is the line between appropriate comfort care and mercy killing? How, if at all, should doctors and nurses be held accountable for their actions in the most desperate of circumstances, especially when their government fails them?

A Shelter From the Storm

Memorial Medical Center was situated on one of the low points in the bowl that is New Orleans, three miles southwest of the city’s French Quarter and three feet below sea level. The esteemed community hospital sprawled across a neighborhood of double-shotgun houses. Several blocks from a housing project but a short walk to the genteel mansions of Uptown, it served a diverse clientele. Built in 1926 and known for decades as Southern Baptist, the hospital was renamed after being purchased in 1995 by Tenet Healthcare, a Dallas-based commercial chain. For generations, the hospital’s sturdy walls served as a shelter when hurricanes threatened: employees would bring their families and pets, as well as coolers packed with muffulettas.

By the time Katrina began lashing New Orleans in the early hours of Monday, Aug. 29, some 2,000 people were bunking in the hospital, including more than 200 patients and 600 workers. When the storm hit, patients screamed as windows shattered under a hail of rocks from nearby rooftops. The hospital groaned and shook violently.

At 4:55 a.m., the supply of city power to the hospital failed. Televisions in patient rooms flicked off. But Memorial’s auxiliary generators had already thumped to life and were humming reassuringly. The system was designed to power only emergency lights, certain critical equipment and a handful of outlets on each floor; the air-conditioning system shut down. By that night, the flooding receded from the surrounding streets. Memorial had sustained damage but remained functional. The hospital seemed to have weathered one more storm.

The Evacuation Begins

Anna Pou was a 49-year-old head- and neck-cancer surgeon whose strong work ethic earned respect from doctors and nurses alike. Tiny and passionate, with coiffed cinnamon hair and a penchant for pearls, Pou was funny and sociable, and she had put her patients at the center of her life.

The morning after Katrina hit, Tuesday, Aug. 30, a nurse called to Pou: ‘‘Look outside!’’ What Pou saw from the window was hard to believe: water gushing from the sewer grates. Other staff members gaped at the dark pool of water rimmed with garbage crawling up South Claiborne Avenue in the direction of the hospital.

Senior administrators quickly grasped the danger posed by the advancing waters and counseled L. René Goux, the chief executive of Memorial, to close the hospital. As at many American hospitals in flood zones, Memorial’s main emergency-power transfer switches were located only a few feet above ground level, leaving the electrical system vulnerable. ‘‘It won’t take much water in height to disable the majority of the medical center,’’ facilities personnel had warned after Hurricane Ivan in 2004. Fixing the problem would be costly; a few less-expensive improvements were made.

Susan Mulderick, a tall, no-nonsense 54-year-old nursing director, was the rotating ‘‘emergency-incident commander’’ designated for Katrina and was in charge — in consultation with the hospital’s top executives — of directing hospital operations during the crisis. The longtime chairwoman of the hospital’s emergency-preparedness committee, Mulderick had helped draft Memorial’s emergency plan. But the 246-page document offered no guidance for dealing with a complete power failure or for how to evacuate the hospital if the streets were flooded. Because Memorial’s chief of medical staff was away, Richard Deichmann, the hospital’s soft-spoken medical-department chairman, organized the physicians.

At 12:28 p.m., a Memorial administrator typed ‘‘HELP!!!!’’ and e-mailed colleagues at other Tenet hospitals outside New Orleans, warning that Memorial would have to evacuate more than 180 patients. Around the same time, Deichmann met with many of the roughly two dozen doctors at Memorial and several nurse managers in a stifling nurse-training room on the fourth floor, which became the hospital’s command center. The conversation turned to how the hospital should be emptied. The doctors quickly agreed that babies in the neonatal intensive-care unit, pregnant mothers and critically ill adult I.C.U. patients would be at great risk from the heat and should get first priority. Then Deichmann broached an idea that was nowhere in the hospital’s disaster plans. He suggested that all patients with Do Not Resuscitate orders should go last.

A D.N.R. order is signed by a doctor, almost always with the informed consent of a patient or health care proxy, and means one thing: A patient whose heartbeat or breathing has stopped should not be revived. A D.N.R. order is different from a living will, which under Louisiana law allows patients with a ‘‘terminal and irreversible condition’’ to request in advance that ‘‘life-sustaining procedures’’ be withheld or withdrawn.

But Deichmann had a different understanding, he told me not long ago. He said that patients with D.N.R. orders had terminal or irreversible conditions, and at Memorial he believed they should go last because they would have had the ‘‘least to lose’’ compared with other patients if calamity struck. Other doctors at the meeting agreed with Deichmann’s plan. Bill Armington, a neuroradiologist, told me he thought that patients who did not wish their lives to be prolonged by extraordinary measures wouldn’t want to be saved at the expense of others — though there was nothing in the orders that stated this. At the time, those attending the meeting didn’t see it as a momentous decision, since rescuers were expected to evacuate everyone in the hospital within a few hours.

There was an important party missing from the conversation. For years, a health care company known as LifeCare Hospitals of New Orleans had been leasing the seventh floor at Memorial. LifeCare operated a ‘‘hospital within a hospital’’ for critically ill or injured patients in need of 24-hour care and intensive therapy over a long period. LifeCare was known for helping to rehabilitate patients on ventilators until they could breathe on their own. LifeCare’s goal was to assist patients until they improved enough to return home or to nursing facilities; it was not a hospice.

The 82-bed unit credentialed its own doctors, most of whom also worked at Memorial. It had its own administrators, nurses, pharmacists and supply chain. It also had its own philosophy: LifeCare deployed the full array of modern technology to keep alive its often elderly and debilitated patients. Horace Baltz, one of the longest-serving doctors at Memorial, told me of spirited debates among doctors over coffee about what some of his colleagues considered to be excessive resources being poured into hopeless cases. ‘‘We spend too much on these turkeys,’’ he said some would say. ‘‘We ought to let them go.’’

Many of the 52 patients at LifeCare were bedbound or required electric ventilators to breathe, and clearly, they would be at significant risk if the hospital lost power in its elevators. The doctors I spoke to who attended the meeting with Deichmann did not recall discussing evacuating LifeCare patients specifically, despite the fact that some of the doctors at the meeting worked with both Memorial and LifeCare patients.

In the afternoon, helicopters from the Coast Guard and private ambulance companies began landing on a long-unused helipad atop an eight-story parking garage adjacent to the hospital. The pilots were impatient — thousands of people needed help across the city. The intensive-care unit on the eighth floor rang out with shouts for patients: ‘‘We need some more! Helicopters are waiting!’’

A crew of doctors, nurses and family members carried Memorial patients down flights of stairs and wheeled them to the hospital wing where the last working elevator brought them to the second floor. Each patient was then maneuvered onto a stretcher and passed through a roughly three-by-three-foot opening in the machine-room wall that offered a shortcut to the parking garage. Many patients were placed in the back of a pickup truck, which drove to the top of the garage. Two flights of metal steps led to the helipad.

At LifeCare that afternoon, confusion reigned. The company had its own ‘‘incident commander,’’ Diane Robichaux, an assistant administrator who was seven months pregnant. At first everything seemed fine; Robichaux established computer communications with LifeCare’s corporate offices in Texas and was assured that LifeCare patients would be included in any FEMA evacuation of Memorial. But as the day wore on, the texts between LifeCare staff members and headquarters grew frantic as it became clear that the government’s rescue efforts and communications were in chaos.

According to the messages, Robichaux asked Memorial administrators to add her 52 patients to transport plans being organized with the Coast Guard. An executive at the hospital told Robichaux that permission would be requested from Memorial’s corporate owner, Tenet Healthcare. ‘‘I hope and pray this is not a long process for getting their approval,’’ Robichaux said in an e-mail message to her colleagues at headquarters. (A Tenet spokesman, David Matthews, wrote me in an e-mail message that LifeCare staff members turned down several offers of evacuation assistance from Memorial staff members on Tuesday afternoon.)

The doctors had now spent days on duty, under stress and sleeping little. Ewing Cook, one of the hospital’s most senior physicians, told me that he decided that in order to lessen the burden on nurses, all but the most critical treatments and care should be discontinued. When Bryant King, a 35-year-old internist who was new to Memorial, came to check on one of his patients on the fourth floor, he canceled the senior doctor’s order to turn off his patient’s heart monitor. When Cook found out, he was furious and thought that the junior doctor did not understand the circumstances. He directed the nurse to reinstate his instructions.

It was dark when the last of the Memorial patients who had been chosen for immediate evacuation were finally gone. Later that night, the Coast Guard offered to evacuate more patients, but those in charge at Memorial declined. The helipad had minimal lighting and no guard rail, and the staff needed rest.

Memorial had shaved its patient census from 187 to about 130. On the seventh floor, all 52 LifeCare patients remained, including seven on ventilators. ‘‘Been on the phone with Tenet,’’ a LifeCare representative outside the hospital wrote to Robichaux. ‘‘Will eventually be to our patients. Maybe in the morning.’’

Fateful Triage Decisions

At about 2 a.m. on Wednesday, Aug. 31 — nearly 48 hours after Katrina made landfall near New Orleans — Memorial’s backup generators sputtered and stopped. Ewing Cook later described the sudden silence as the ‘‘sickest sound’’ of his life. In LifeCare on the seventh floor, critically ill patients began suffering the consequences. Alarm bells clanged as life-support monitors and ventilators switched to brief battery reserves while continuing to force air into the lungs of seven patients. In about a half-hour, the batteries failed and the regular hiss of mechanical breaths ceased. A Memorial nurse appeared and announced that the Coast Guard could evacuate some critical patients if they were brought to the helipad immediately. Volunteers began carrying the LifeCare patients who relied on ventilators down five flights of stairs in the dark.

A LifeCare nurse navigated the staircase alongside an 80-year-old man on a stretcher, manually squeezing air into his lungs with an Ambu bag. As he waited for evacuation on the second floor, she bagged him for nearly an hour. Finally a physician stopped by the stretcher and told her that there was no oxygen for the patient and that he was already too far gone. She hugged the man and stroked his hair as he died.

Anna Pou began bagging another patient on the second floor to relieve a nurse whose hands were growing tired. That patient, along with two other LifeCare patients who relied on ventilators, also died early that morning, but the others were evacuated by helicopter. The hospital chaplain opened a double door with stained-glass windows down the hallway, and the staff began wheeling bodies into the chapel. Distraught nurses cried, and the chaplain held them and prayed with them.

The sun rose and with it the sultry New Orleans temperature, which was on its way to the mid-90s. The hospital was stifling, its walls sweating. Water had stopped flowing from taps, toilets were backed up and the stench of sewage mixed with the odor of hundreds of unwashed bodies.

Visitors who had come to the hospital for safety felt so desperate that they cheered when two airboats driven by volunteers from the Louisiana swamplands roared up to the flooded emergency-room ramp. The flotilla’s organizers, Mark and Sandra LeBlanc, had a special reason to come to Memorial: Vera LeBlanc, Mark’s 82-year-old mother, was at LifeCare, recovering from colon-cancer surgery. Sandra, an E.M.T., knew that her mother-in-law couldn’t swallow, so she was surprised when she saw that Vera and other patients who needed IVs to keep hydrated were no longer getting them. When her husband asked a Memorial administrator why, the administrator told him that the hospital was in survival mode, not treating mode. Furious, Mark LeBlanc asked, ‘‘Do you just flip a switch and you’re not a hospital anymore?’’

That morning, doctors and nurses decided that the more than 100 remaining Memorial and LifeCare patients should be brought downstairs and divided into three groups to help speed the evacuation. Those who were in fairly good health and could sit up or walk would be categorized ‘‘1’s’’ and prioritized first for evacuation. Those who were sicker and would need more assistance were ‘‘2’s.’’ A final group of patients were assigned ‘‘3’s’’ and were slated to be evacuated last. That group included those whom doctors judged to be very ill and also, as doctors agreed the day before, those with D.N.R. orders.

Though there was no single doctor officially in charge of categorizing the patients, Pou was energetic and jumped into the center of the action, according to two nurses who worked with her. Throughout the morning, makeshift teams of medical staff and family members carried many of the remaining patients to the second-floor lobby where Pou, the sleeves of her scrubs rolled up, stood ready to receive them.

In the dim light, nurses opened each chart and read the diagnoses; Pou and the nurses assigned a category to each patient. A nurse wrote ‘‘1,’’ ‘‘2’’ or ‘‘3’’ on a sheet of paper with a Marks-A-Lot pen and taped it to the clothing over a patient’s chest. (Other patients had numbers written on their hospital gowns.) Many of the 1’s were taken to the emergency-room ramp, where boats were arriving. The 2’s were generally placed along the corridor leading to the hole in the machine-room wall that was a shortcut to the helipad. The 3’s were moved to a corner of the second-floor lobby near an A.T.M. and a planter filled with greenery. Patients awaiting evacuation would continue to be cared for — their diapers would be changed, they would be fanned and given sips of water if they could drink — but most medical interventions like IVs or oxygen were limited.

Pou and her co-workers were performing triage, a word once used by the French in reference to the sorting of coffee beans and applied to the battlefield by Napoleon’s chief surgeon, Baron Dominique-Jean Larrey. Today triage is used in accidents and disasters when the number of injured exceeds available resources. Surprisingly, perhaps, there is no consensus on how best to do this. Typically, medical workers try to divvy up care to achieve the greatest good for the greatest number of people. There is an ongoing debate about how to do this and what the ‘‘greatest good’’ means. Is it the number of lives saved? Years of life saved? Best ‘‘quality’’ years of life saved? Or something else?

At least nine well-recognized triage systems exist. Most call for people with relatively minor injuries to wait while patients in the worst shape are evacuated or treated. Several call for medical workers to sort the injured into another category: patients who are seen as having little chance of survival given the resources on hand. That category is most commonly created during a devastating event like a war-zone truck bombing in which there are far more severely injured victims than ambulances or medics.

Pou and her colleagues had little if any training in triage systems and were not guided by any particular triage protocol. Pou would later say she was trying to do the most good with a limited pool of resources. The decision that certain sicker patients should go last has its risks. Predicting how a patient will fare is inexact and subject to biases. In one study of triage, experienced rescuers were asked to categorize the same patients and came up with widely different answers. And patients’ conditions change; more resources can become available to help those whose situations at first appear hopeless. The importance of reassessing each person is easy to forget once a ranking is assigned.

After several helicopters arrived and rescued some of the LifeCare patients, Air Force One flew over New Orleans while President Bush surveyed the devastation. Few helicopters arrived after that. Pou told me she heard that the Coast Guard was focusing on saving people stranded on rooftops around the city. Meanwhile dozens of patients sweltered on the lower two floors of Memorial and in the parking garage as they waited to leave.

Many of the doctors and nurses had shifted from caring for patients to carrying them and were loading people onto helicopters and watercraft. Vera LeBlanc, the LifeCare patient whose son arranged the airboat flotilla that had arrived hours earlier, was among the patients massed on the second floor. Her chart read ‘‘Do Not Resuscitate,’’ as it had during several hospital admissions for more than a decade, so that her heart would not be restarted if it were to stop. Mark LeBlanc decided he was going to put his mother on one of the airboats he and his wife had directed to the hospital. When the LeBlancs tried to enter the patient area on the second floor, a staff member blocked them, and several doctors told them they couldn’t leave with Vera. ‘‘The hell we can’t,’’ Sandra said. The couple ignored the doctors, and Vera smiled and chatted as Mark and several others picked her up and carried her onto an airboat.

On a seventh-floor hallway at LifeCare, Angela McManus, a daughter of a patient, panicked when she overheard workers discussing the decision to defer evacuation for D.N.R. patients. She had expected her frail 70-year-old mother, Wilda, would soon be rescued, but her mother had a D.N.R. order. ‘‘I’ve got to rescind that order,’’ Angela begged the LifeCare staff. She says they told her that there were no doctors available to do it.

By Wednesday afternoon, Dr. Ewing Cook was physically and mentally exhausted, filthy and forlorn. A 61-year-old pulmonary specialist, he’d had his semi-automatic Beretta strapped to him since he heard on Monday that a nurse was raped while walking her dog near the hospital (a hospital official denies that this happened). Cook had had two heart attacks and could not help transport patients in the heat.

That afternoon, Cook stood on the emergency-room ramp and caught sight of a mattress floating up Napoleon Avenue. On it lay an emaciated black woman, with several young men propelling her through the fetid water. ‘‘The hospital is closed,’’ someone shouted. ‘‘We’re not accepting anybody.’’

René Goux, the hospital’s chief executive, told me he had decided, for reasons of safety, that people floating up to Memorial should generally be directed to dry ground about nine blocks south. Medical workers finally insisted that the woman and her husband be allowed to enter, but the men who swam in the toxic soup to rescue her were told to leave. When a couple with small children rowed up and were told to ‘‘go away,’’ Bryant King, who was one of Memorial’s few African-American physicians, lost his temper.

‘‘You can’t do this!’’ King shouted at Goux. ‘‘You gotta help people!’’ But the family was turned away.

King was out of touch with reality, Cook told me he thought at the time. Memorial wasn’t so much a hospital anymore but a shelter that was running out of supplies and needed to be emptied. Cook also worried that intruders from the neighborhood might ransack the hospital for drugs and people’s valuables.

Recently retired from clinical practice, Cook became a Memorial administrator a week before Katrina hit, but he had spent many years working on the eighth floor in the I.C.U. That afternoon, he climbed slowly upstairs to check what was happening there. Most of the patients had been evacuated on Tuesday, but a few with D.N.R. orders had not.

‘‘What’s going on here?’’ he asked the four nurses in the unit. ‘‘Whaddya have left?’’ The nurses said they were down to one patient: Jannie Burgess, a 79-year-old woman with advanced uterine cancer and kidney failure. She was being treated for comfort only and had been sedated to the point of unconsciousness with morphine. She was so weighted down by fluid from her diseases that Cook sized her up at 350 pounds.

Cook later told me he believed several things: 1. Given how difficult it had been for him to climb the steps in the heat, there was no way he could make it back to the I.C.U. again. 2. Given how exhausted everyone was and how much this woman weighed, it would be ‘‘impossible to drag her down six flights of stairs.’’ 3. Even in the best of circumstances, the patient probably had a day or so to live. And frankly, the four nurses taking care of her were needed elsewhere.

To Cook, a drug that had been dripping into Burgess’s IV for days provided an answer. Morphine, a powerful narcotic, is frequently used to control severe pain or discomfort. But the drug can also slow breathing, and suddenly introducing much higher doses can lead to death.

Doctors, nurses and clinical researchers who specialize in treating patients near the ends of their lives say that this ‘‘double effect’’ poses little danger when drugs are administered properly. Cook says it’s not so simple. ‘‘If you don’t think that by giving a person a lot of morphine you’re not prematurely sending them to their grave, then you’re a very naïve doctor,’’ Cook told me when we spoke for the first time, in December 2007. ‘‘We kill ’em.’’

In fact, the distinction between murder and medical care often comes down to the intent of the person administering the drug. Cook walked this line often as a pulmonologist, he told me, and he prided himself as the go-to man for difficult end-of-life situations. When a very sick patient or the patient’s family made the decision to disconnect a ventilator, for example, Cook would prescribe morphine to make sure the patient wasn’t gasping for breath as the machine was withdrawn.

Often Cook found that achieving this level of comfort required enough morphine that the drug markedly suppressed the patient’s breathing. The intent was to provide comfort, but the result was to hasten death, and Cook knew it. To Cook, the difference between something ethical and something illegal ‘‘is so fine as to be imperceivable.’’

Burgess’s situation was ‘‘a little different,’’ as Cook described it. Being comatose and on painkillers, she wasn’t uncomfortable. But the worst thing Cook could imagine would be for the drugs to wear off and for Burgess to wake up and find herself in her ravaged condition as she was being moved. ‘‘Do you mind just increasing the morphine and giving her enough until she goes?’’ Cook told me he asked Burgess’s nurse.

Cook scribbled ‘‘pronounced dead at’’ in Burgess’s chart, left the time blank and signed the note with a large squiggle. Then he walked back downstairs, believing that he had done the right thing for Burgess. ‘‘To me, it was a no-brainer, and to this day I don’t feel bad about what I did,’’ he told me. ‘‘I gave her medicine so I could get rid of her faster, get the nurses off the floor.’’ He added, ‘‘There’s no question I hastened her demise.’’

The question of what to do with the hospital’s sickest patients was also being raised by others. By the afternoon, with few helicopters landing, these patients were languishing. Susan Mulderick, the ‘‘incident commander’’ who had worked with Cook for decades, shared her own concerns with him. According to Cook, Mulderick told him, ‘‘We gotta do something about this.’’ Mulderick, who declined to be formally interviewed about the days after Katrina, did tell me: ‘‘We were well prepared. We managed that situation well.’’

Cook sat on the emergency-room ramp smoking cigars with another doctor. Help was coming too slowly. There were too many people who needed to leave and weren’t going to make it, Cook said, describing for me his thinking at the time. It was a desperate situation and he saw only two choices: quicken their deaths or abandon them. ‘‘It was actually to the point where you were considering that you couldn’t just leave them; the humane thing would be to put ’em out.’’

Cook went to the staging area on the second floor where Anna Pou and two other doctors were directing care. Cots and stretchers seemed to cover every inch of floor space. Rodney Scott, an obese I.C.U. patient who was recovering from heart problems and several operations, lay motionless on a stretcher, covered in sweat and almost nothing else. A doctor had decided that he should be the last patient to leave the hospital because he weighed more than 300 pounds and might get stuck in the machine-room hole, backing up the evacuation line. Cook thought Scott was dead, and he touched him to make sure. But Scott turned over and looked at him.

‘‘I’m O.K., Doc,’’ Scott said. ‘‘Go take care of somebody else.’’

Despite how miserable the patients looked, Cook said, he felt there was no way, in this crowded room, to do what he had been thinking about. ‘‘We didn’t do it because we had too many witnesses,’’ he told me. ‘‘That’s the honest-to-God truth.’’

Richard Deichmann, Memorial’s medical-department chairman, also remembers being stopped by Mulderick for a quick conversation that afternoon, an episode he wrote about in ‘‘Code Blue,’’ a memoir he published in 2006 about the days after Katrina. He was startled, he wrote, when Mulderick asked him his thoughts about whether it would be ‘‘humane’’ to euthanize the hospital’s D.N.R. patients. ‘‘Euthanasia’s illegal,’’ he said he told her. ‘‘There’s not any need to euthanize anyone. I don’t think we should be doing anything like that.’’ He had figured the D.N.R. patients should go last, but the plan, he told Mulderick, was still to evacuate them eventually. Through her lawyer, Mulderick denied that she discussed euthanasia of patients with Deichmann or anyone else at Memorial.

As darkness fell, rumor spread that evacuations would halt for the night because people were shooting at rescuers. In the adjacent parking garage, Goux distributed guns to security and maintenance staff, who cordoned off the hospital’s entrances. That night, dozens of LifeCare and Memorial patients lay on soiled and sweaty cots in the second-floor lobby. Pou, several doctors and crews of nurses worked in the dim light of a few lamps powered by a portable generator. For the third night in a row, Pou was working with scarcely an hour’s sleep, changing patients’ diapers, giving out water, comforting and praying with nurses.

Kamel Boughrara, a LifeCare nursing director, walked past the A.T.M. area on the second floor where some of the sickest patients — most of whom had been given 3’s — lay. Carrie Hall, a 78-year-old LifeCare patient with long, braided hair whose vast family called her Ma-Dear, managed to grab him and indicate that she needed her tracheostomy cleared. The nurse was surprised at how fiercely Hall was battling to stay alive. He suctioned her with a portable machine and told her to fight hard.

Comfort Care Or Mercy Killing?

Soon after sunrise on Thursday, Sept. 1 — more than 72 hours into the crisis — Memorial’s chief financial officer, Curtis Dosch, delivered good news to hospital staff gathered on the emergency-room ramp. He had reached a Tenet representative in Dallas and was told that Tenet was dispatching a fleet of privately hired helicopters that day. Dosch later said that the dejected staff was skeptical. But soon the hospital’s voice chain began echoing with shouts for women and children to evacuate. Boats were arriving, including fishing vessels that had been parked on trailers in the neighborhood and were now commandeered by hospital workers. Helicopters at last converged on the hospital within a couple of hours of daylight, according to a Memorial nurse from the Air Force Reserve who oversaw helipad operations. The Tenet spokesman and testimony by Mulderick in a 2008 deposition also confirm this. The hospital filled with the cacophony of military and private crafts hovering and landing. Down on the emergency-room ramp that morning, stone-faced State Police officers wielding shotguns barked that everyone had to be out of the hospital by 5 p.m. because of civil unrest in New Orleans; they would not stay later to protect the hospital.

Meanwhile, Cook strapped on his gun again and prepared to leave the hospital by boat to rescue his son, who had been trapped at his house since Tuesday’s flooding. He told me that Mulderick asked him before he left to talk to Pou.

On the second floor, Cook says, he and Pou, both weary, discussed the Category 3 patients, including nine who had never been brought down from the seventh floor. According to Cook, Pou was worried that they wouldn’t be able to get them out. Cook hadn’t been on the seventh floor since Katrina struck, but he told me that he thought LifeCare patients were ‘‘chronically deathbound’’ at the best of times and would have been horribly affected by the heat. Cook couldn’t imagine how the exhausted Memorial staff would carry nine patients down five flights of stairs before the end of the day. Nobody from outside had arrived to help with that task. If there were other ways to evacuate these patients, he didn’t see them.

Cook said he told Pou how to administer a combination of morphine and a benzodiazepine sedative. The effect, he told me, was that patients would ‘‘go to sleep and die.’’ He explained that it ‘‘cuts down your respiration so you gradually stop breathing and go out.’’ He said he believed that Pou understood that he was telling her how to achieve this. He said that he viewed it as a way to ease the patients out of a terrible situation.

In an interview with Newsweek in 2007, Pou acknowledged that after discussions with other doctors, she did inject some Category 3 patients. But she said her intention was only to ‘‘help the patients that were having pain and sedate the patients who were anxious’’ because ‘‘we knew they were going to be there another day, that they would go through at least another day of hell.’’ Beyond that, Pou has not talked about the details of what happened on that Thursday, citing the pending legal cases and sensitivity to patients and their families. What follows is based on the recollections of others, some of which were recounted in interviews with Louisiana Justice Department investigators, as well as in interviews with me.

Therese Mendez, a LifeCare nurse executive, had worked overnight on the first floor, she later told investigators. (She declined to speak with me.) After daybreak, she heard the sound of helicopters and watched the evacuation line begin to move. According to Mendez, she returned at around 8 or 9 a.m. to the seventh floor and walked along a corridor. The patients she saw looked bad. Several were unconscious, frothing at the mouth and breathing in an irregular way that often heralds death. Still, while two patients died on the LifeCare’s seventh floor on Wednesday, the others had lived through the night, with only a few given small doses of morphine or the sedative lorazepam for comfort.

Mendez heard that Pou was looking for her. They sat down in an office with an open window. Pou looked distraught and told her that the LifeCare patients probably were not going to survive. Mendez told investigators that she responded, ‘‘I think you’re right.’’

Mendez said she watched Pou struggle with what she was saying, telling investigators that Pou told her that ‘‘the decision had been made to administer lethal doses’’ of morphine and other drugs. (Pou, through her lawyer, Richard Simmons Jr., denied mentioning ‘‘lethal doses.’’) Were the LifeCare patients being singled out? Mendez asked. She knew there were other sick patients at Memorial. Mendez recalled that Pou said ‘‘no’’ and that there was ‘‘no telling how far’’ it would go.

According to Mendez, Pou told her that she and other Memorial staff members were assuming responsibility for the patients on the seventh floor; the LifeCare nursing staff wasn’t involved and should leave. (Pou, through her lawyer, disputes Mendez’s account.) Mendez later said she had assumed that the hospital was under martial law, which was not the case, and that Pou was acting under military orders. Mendez left to dismiss her employees, she said, because she feared they would be forced downstairs by authorities.

Diane Robichaux, the senior leader on the LifeCare floor, later walked into the office, she recalled in interviews with investigators. (She declined to talk to me.) She and other LifeCare workers had gone downstairs at around 9:30 a.m. to ask Susan Mulderick when the LifeCare patients on the seventh floor would be evacuated. According to Robichaux, Mulderick said, ‘‘The plan is not to leave any living patients behind,’’ and told her to see Pou.

In Robichaux’s interview with investigators, she could not recall exactly what Pou told her, but she said that she understood that patients ‘‘were not going to be making it out of there.’’ She said that Pou did not use the word ‘‘euthanize.’’ Prompted by investigators, she said she thought Pou might have used the word ‘‘comfortable’’ in describing what she was trying to do for the patients.

Robichaux remembered Pou saying that the LifeCare patients were ‘‘not aware or not alert or something along those lines.’’ Robichaux recounted to investigators that she told Pou that that wasn’t true and said that one of LifeCare’s patients — Emmett Everett, a 380-pound man — was ‘‘very aware’’ of his surroundings. He had fed himself breakfast that morning and asked Robichaux, ‘‘So are we ready to rock and roll?’’

The 61-year-old Honduran-born manual laborer was at LifeCare awaiting colostomy surgery to ease chronic bowel obstruction, according to his medical records. Despite a freakish spinal-cord stroke that left him a paraplegic at age 50, his wife and nurses who worked with him say he maintained a good sense of humor and a rich family life, and he rarely complained. He, along with three of the other LifeCare patients on the floor, had no D.N.R. order.

Everett’s roommates had already been taken downstairs on their way to the helicopters, whose loud propellers sent a breeze through the windows on his side of the LifeCare floor. Several times he appealed to his nurse, ‘‘Don’t let them leave me behind.’’ His only complaint that morning was dizziness, a LifeCare worker told Pou.

‘‘Oh, my goodness,’’ a LifeCare employee recalled Pou replying.

Two Memorial nurses — identified as Cheri Landry and Lori Budo from the I.C.U. to investigators by a LifeCare pharmacist, Steven Harris — joined the discussion along with other LifeCare workers. (Through their lawyers, Landry and Budo declined to be interviewed. Harris never returned my calls.) They talked about how Everett was paralyzed and had complex medical problems and had been designated a ‘‘3’’ on the triage scale. According to Robichaux, the group concluded that Everett was too heavy to be maneuvered down the stairs, through the machine-room wall and onto a helicopter. Several medical staff members who helped lead boat and helicopter transport that day say they would certainly have found a way to evacuate Everett. They say they were never made aware of his presence.

In his interviews with investigators, Andre Gremillion, a LifeCare nurse, said that the female physician in the office (he didn’t know Pou’s name) asked if someone who knew Everett could explain to him that because he was so big they did not think they would be able to evacuate him. They asked Gremillion whether he could ‘‘give him something to help him relax and explain the situation.’’ Gremillion told investigators that he didn’t want to be the one who told Everett that ‘‘we would probably be leaving and he would be staying.’’ At that point, Gremillion said, he lost his composure.

Gremillion’s supervisor and friend, a LifeCare nursing director, Gina Isbell, told me she walked into the room around 11 a.m. and saw Gremillion crying and shaking his head. He brushed past her into the hallway, and Isbell followed, grabbing his arm and guiding him to an empty room. ‘‘I can’t do this,’’ he kept saying.

‘‘Do what?’’ Isbell asked. When Gremillion wouldn’t answer, Isbell tried to comfort him. ‘‘It’s going to be O.K.,’’ she said. ‘‘Everything’s going to be all right.’’

Isbell searched for Robichaux, her boss. ‘‘What is going on?’’ she asked, frantic. ‘‘Are they going to do something to our patients?’’

‘‘Yes, they are,’’ Isbell remembers Robichaux, in tears, saying. ‘‘Our patients aren’t going to be evacuated. They aren’t going to leave.’’ As the LifeCare administrators cleared the floor of all but a few senior staff members, Robichaux sent Isbell to the back staircase to make sure nobody re-entered. It was quiet there, and Isbell sat alone, drained and upset. Isbell said she thought about her patients, remembering with guilt a promise she made to the daughter of one of her favorites, Alice Hutzler, a 90-year-old woman who came to LifeCare for treatment of bedsores and pneumonia. Isbell fondly called her Miss Alice and had told Hutzler’s daughter that she would take good care of her mother. Now Isbell prayed that help would come before Hutzler and her other patients died.

According to statements made to investigators by Steven Harris, the LifeCare pharmacist, Pou brought numerous vials of morphine to the seventh floor. According to investigators, a proffer from Harris’s lawyer said that Harris gave her additional morphine and midazolam — a fast-acting drug used to induce anesthesia before surgery or to sedate patients for medical procedures. Like morphine, midazolam depresses breathing; doctors are warned to be extremely careful when combining the two drugs.

Kristy Johnson, LifeCare’s director of physical medicine, said she saw what happened next. She told Justice Department investigators that she watched Pou and two nurses draw fluid from vials into syringes. Then Johnson guided them to Emmett Everett in Room 7307. Johnson said she had never seen a physician look as nervous as Pou did. As they walked, she told investigators, she heard Pou say that she was going to give him something ‘‘to help him with his dizziness.’’ Pou disappeared into Everett’s room and shut the door.

As they worked their way down the seventh-floor hallway, Johnson held some of the patients’ hands and said a prayer as Pou or a Memorial nurse gave injections. Wilda McManus, whose daughter Angela had tried in vain to rescind her mother’s D.N.R. order, had a serious blood infection. (Earlier, Angela was ordered to leave her mother and go downstairs to evacuate.) ‘‘I am going to give you something to make you feel better,’’ Pou told Wilda, according to Johnson.

Johnson took one of the Memorial nurses into Room 7305. ‘‘This is Ms. Hutzler,’’ Johnson said, touching the woman’s hand and saying a ‘‘little prayer.’’ Johnson tried not to look down at what the nurse was doing, but she saw the nurse inject Hutzler’s roommate, Rose Savoie, a 90-year-old woman with acute bronchitis and a history of kidney problems. A LifeCare nurse later told investigators that both women were alert and stable as of late that morning. ‘‘That burns,’’ Savoie murmured.

According to Memorial workers on the second floor, about a dozen patients who were designated as ‘‘3’s’’ remained in the lobby by the A.T.M. Other Memorial patients were being evacuated with help from volunteers and medical staff, including Bryant King. Around noon, King told me, he saw Anna Pou holding a handful of syringes and telling a patient near the A.T.M., ‘‘I’m going to give you something to make you feel better.’’ King remembered an earlier conversation with a colleague who, after speaking with Mulderick and Pou, asked him what he thought of hastening patients’ deaths. That was not a doctor’s job, he replied. Patients were hot and uncomfortable, and a few might be terminally ill, but he didn’t think they were in the kind of pain that calls for sedation, let alone mercy killing. When he saw Pou with the syringes, he assumed she was doing just that and said to anyone within earshot: ‘‘I’m getting out of here. This is crazy!’’ King grabbed his bag and stormed downstairs to get on a boat.

Bill Armington, the neuroradiologist, watched King go and was upset at him for leaving. Armington suspected that euthanasia might occur, in part, he told me, because Cook told him earlier that there had been a discussion of  ‘‘things that only doctors talk about.’’ Armington headed for the helipad, “stirred up,’’ as he recalls, ‘‘to intensify my efforts to get people off the roof.’’ Neither Armington nor King intervened directly, though King had earlier sent out text messages to friends and family asking them to tell the media that doctors were discussing giving medication to dying patients to help accelerate their deaths. King told me that he didn’t think his opinion, which hadn’t mattered when he argued against turning away the hospital’s neighbors, would have mattered.

Only a few nurses and three doctors remained on the second floor: Pou; a young internist named Kathleen Fournier; and John Thiele, a 53-year-old pulmonologist, who had never before spoken publicly about his Katrina experiences until we had two lengthy interviews in the last year. Thiele told me that on Thursday morning, he saw Susan Mulderick walking out of the emergency room. ‘‘John, everybody has to be out of here tonight,’’ he said she told him. He said René Goux told him the same thing. Mulderick, through her lawyer, and Goux both say that they were not given a deadline to empty the hospital and that their goal was to focus their exhausted colleagues on the evacuation. ‘‘We’d experienced the helicopters’ stopping flying to us,’’ Goux told me, ‘‘and I didn’t want that to occur again.’’

Around a corner from where the patients lay on the second floor, Thiele and Fournier struggled to euthanize two cats whose owners brought them to the hospital and were forced to leave them behind. Thiele trained a needle toward the heart of a clawing cat held by Fournier, he told me later. While they were working, Thiele recalls Fournier telling him that Mulderick had spoken with her about something to the effect of putting patients ‘‘out of their misery’’ and that she did not want to participate. (Fournier declined to talk with me.) Thiele told her that he understood, and that he and others would handle it. Mulderick’s lawyer says that Mulderick did ask a physician about giving something to patients to ‘‘make them more comfortable,’’ but that, however, was not ‘‘code for euthanasia.’’

Thiele didn’t know Pou by name, but she looked to him like the physician in charge on the second floor. He told me that Pou told him that the Category 3 patients were not going to be moved. He said he thought they appeared close to death and would not have survived an evacuation. He was terrified, he said, of what would happen to them if they were left behind. He expected that the people firing guns into the chaos of New Orleans — ‘‘the animals,’’ he called them — would storm the hospital, looking for drugs after everyone else was gone. ‘‘I figured, What would they do, these crazy black people who think they’ve been oppressed for all these years by white people? I mean if they’re capable of shooting at somebody, why are they not capable of raping them or, or, you know, dismembering them? What’s to prevent them from doing things like that?’’

The laws of man had broken down, Thiele concluded, and only the laws of God applied.

‘‘Can I help you?’’ he says he asked Pou several times.

‘‘No,’’ she said, according to Thiele. ‘‘You don’t have to be here.’’

‘‘I want to be here,’’ Thiele insisted. ‘‘I want to help you.’’

Thiele practiced palliative-care medicine and was certified to teach it. He told me that he knew that what they were about to do, though it seemed right to him, was technically ‘‘a crime.’’ He said that ‘‘the goal was death; our goal was to let these people die.’’

Thiele saw that morphine, midazolam and syringes had been set up on a table near the A.T.M. There were about a dozen patients, and he took charge of the four closest to the windows — three elderly white women and a heavyset African-American man — starting IVs on those who didn’t have one. Apart from their breathing and the soft moans of one, the patients appeared ‘‘lifeless’’ and did not respond to him. Thiele saw Pou and several nurses working on patients lying near the hallway.

Thiele wavered for a moment. He turned to Karen Wynn, the I.C.U. nurse manager at Memorial who led the hospital’s ethics committee. ‘‘Can we do this?’’ he remembers asking the highly respected nurse.

Wynn felt that they needed to medicate the patients, she said when she described her experiences publicly for the first time in interviews with me over the past year. She acknowledged having heard rumors that patients were being euthanized, but she said no one had told her that that was what was happening to these patients and that her only aim was to make patients comfortable by sedating them. Wynn said she did not fear staying in the hospital after the 5 p.m. curfew announced by the State Police — she had already decided to ignore the evacuation deadline and stay at the hospital until everyone alive had been taken out. Instead, she said, she was motivated by how bad the patients looked.

Wynn described turning to an elderly woman who was unconscious with labored breathing. She then prepared a syringe with morphine and midazolam, pushed it slowly into the woman’s IV line and watched her breathing ease. The woman died a short time later, which didn’t disturb Wynn because she had appeared to be close to death. Wynn told me that at that point all the staff could offer was ‘‘comfort, peace and dignity.’’ She said: ‘‘We did the best we could do. It was the right thing to do under the circumstances.’’

She added: “But even if it had been euthanasia, it’s not something we don’t really do every day — it just goes under a different name.’’

Thiele gave other patients a shot of morphine and midazolam at doses he said were higher than what he normally used in the I.C.U. He held their hands and reassured them, ‘‘It’s all right to go.’’ Most patients, Thiele told me, died within minutes of being medicated. But the heavyset African-American man didn’t.

His mouth was open, his breathing was labored and everyone could hear his awful death rattle. Thiele tried more morphine. He tried prayer. He put his hand on the man’s forehead; Wynn and another nurse manager took the man’s hands in theirs. Together they chanted: ‘‘Hail Mary, full of grace. The Lord is with thee.’’ They recited the Lord’s Prayer. They prayed for the man to die.

The man kept breathing, and Wynn says she and her colleagues took that as a sign. ‘‘God said, ‘O.K., but I’m not ready for him.’ Or he wasn’t ready.’’ She remembers passing him through the hole in the machine-room wall on his way to the evacuation helicopters.

Thiele has a different memory of what happened. ‘‘We covered his face with a towel’’ until he stopped breathing, Thiele told me.

He says that it took less than a minute for the man to die and that he didn’t suffer. ‘‘This was totally against every fiber in my body,’’ Thiele told me, but he also said he knew what he did was right. ‘‘We were abandoned by the government, we were abandoned by Tenet, and clearly nobody was going to take care of these people in their dying moments.’’ He added, ‘‘I did what I would have wanted done to me if the roles were reversed.’’

Both Thiele and Wynn recall that they, Pou and the other nurses covered the bodies of the dead and carried them into the chapel, filling it. Thiele said the remaining bodies were wrapped in sheets and placed on the floor in the corridor and in a nearby room.

‘‘It was very respectful,’’ Thiele told me. ‘‘It’s not like you would think.’’

That afternoon, Memorial’s pathologist and laboratory director walked though the hospital, floor by floor, to record the locations of the dead and make sure that nobody alive was left behind. They found Pou on the seventh floor with a nurse. Pou was working on the IV of a patient who seemed barely alive. The laboratory director told investigators that Pou asked for help moving the patient; the pathologist remembered it differently and said in a deposition that he offered Pou help with evacuating the patient, but Pou did not respond, and later, when he asked her again, she said she needed to speak with an anesthesiologist first.

Dr. John Walsh, a surgeon, told me that he was sitting on a bench, too tired to move, when Pou and the pathologist came downstairs. Pou looked upset. She sat down beside him. ‘‘What’s wrong?’’ he asked. He said she mentioned something about a patient, or patients, dying and about someone, or some people, questioning her.

Walsh had known Pou for about only a year, but he knew, he told me, that she was compassionate and dedicated to her patients. ‘‘I’m sure you did the right thing,’’ he remembers telling her. ‘‘It’ll work itself out. It’ll all turn out O.K.’’

Throughout the day, boats and helicopters drained the hospital of nearly all of its patients and visitors. At around 9 p.m., Rodney Scott, the obese I.C.U. patient who was recovering from surgery and heart trouble, at last felt himself being hoisted up the open metal steps to the helipad. Weighing more than 300 pounds and unable to walk, Scott was the last living patient to leave the hospital grounds. He felt relief. The four men surrounding him shouted, ‘‘Push! Push!’’ and rolled his heavy wheelchair into a Coast Guard helicopter. Evacuating someone as large as Scott had a cost — a nurse was briefly pinned against the helicopter, bruising his ribs and spleen — but it had been done.

Scott, Thiele and Wynn were flown separately to Louis Armstrong New Orleans International Airport, where their ordeals continued. Hundreds of hospital and nursing-home patients had been dropped there from across the disaster zone; they were met by federal disaster-management teams that were so understaffed and undersupplied that they couldn’t provide even basic nursing care to many patients. Reflecting on the scene at the airport, Thiele told me that if the patients he injected with drugs had made it there, ‘‘They wouldn’t have survived.’’

The Coroner's Dilemma

On Sunday, Sept. 11, 2005, 13 days after the storm hit, mortuary workers recovered 45 decomposing bodies from Memorial Medical Center. The next day the Louisiana attorney general, Charles Foti Jr., opened investigations into hospital and nursing-home deaths during Hurricane Katrina. The Justice Department’s phones were soon ringing with allegations of patient abandonment and euthanasia.

One of the people who called was a LifeCare lawyer who relayed a report that nine of the company’s patients may have been given lethal doses of medicines by a Memorial doctor and nurses. State and federal investigators interviewed LifeCare witnesses and descended on the mold-ridden hospital to search for evidence. Separately, Foti’s staff asked the Orleans Parish coroner, Dr. Frank Minyard, to perform autopsies and drug tests on approximately 100 bodies that were recovered from more than a half-dozen hospitals and nursing homes in New Orleans.

The burden was unwelcome for Minyard, a 76-year-old obstetrician-gynecologist who was already struggling to oversee the autopsies and identification of hundreds of hurricane victims. Minyard was inspired by a Catholic nun to devote his life to public service. For 31 years as the city’s elected coroner, he peered into bodies in the basement office of the colonnaded criminal courthouse, emerging in cowboy boots and white suits to play jazz trumpet at city charity events. As New Orleans flooded, Minyard says, he got out of his car and swam to work. He was trapped there for four days.

After autopsies were done and specimens removed, workers at National Medical Services, a private laboratory in Pennsylvania, quickly detected morphine in nine bodies — the same nine patients LifeCare staff identified as potential victims.

The attorney general’s office hired a forensic pathologist, Cyril Wecht, who worked on the John F. Kennedy assassination case and the O.J. Simpson murder trial, to review evidence in the deaths of four patients whose full toxicology reports and medical records they obtained first: Emmett Everett, Rose Savoie and two other LifeCare patients. Wecht concluded that all four deaths were homicides, caused by human intervention.

After months of conducting interviews and collecting documents, investigators came to believe, they said, that doctors and nurses euthanized as many as two dozen patients at Memorial. But medical records were needed to substantiate the findings, and according to investigators, Tenet’s lawyers said that many of those belonging to Memorial patients were unavailable. (The Tenet spokesman said via e-mail that Tenet produced all records in its possession.) Armed with the testimony of LifeCare workers and the medical records of the four patients on the seventh floor, state prosecutors decided their strongest case was against Anna Pou, Cheri Landry and Lori Budo for those deaths.

At about 9 p.m. on July 17, 2006 — nearly a year after floodwaters from Katrina swamped Memorial hospital — Pou opened the door of her home to find state and federal agents, clad in body armor and carrying weapons. They told her they had a warrant for her arrest on four counts of principal to second-degree murder.

Pou was wearing rumpled surgical scrubs from several hours of surgery she performed earlier in the day. She knew she was a target of the investigation, but her lawyer thought he had assurance that she could surrender voluntarily. “What about my patients?” she asked reflexively. An agent suggested that Pou call a colleague to take over their care. She was allowed to freshen up and then was read her rights, handcuffed and ultimately driven to the Orleans Parish jail. On the way, she prayed silently. (Landry and Budo were arrested the same night.)

Pou was booked and released after midnight. The next day the attorney general, Foti, held a news conference carried on CNN, which had broken some of the initial reports of the investigation and the possible euthanasia at Memorial. “This is not euthanasia,” Foti said emphatically. “This is plain and simple homicide.”

At a news conference later that day, Pou’s lawyer blamed the storm — not Pou — for the deaths. He said his client was innocent and accused Foti, who was about to run for re-election, of orchestrating a media event with the arrests. He announced his intention to bring the results of his own investigation to the Orleans Parish district attorney, whose office had jurisdiction over the case after the arrests and would bring it before a grand jury.

As the government investigation progressed, Carrie Everett, Emmett Everett’s widow, spoke out on CNN. After Katrina she searched for her husband for two weeks before learning that he was dead. She filed wrongful death lawsuits against Tenet, LifeCare, Pou, Landry and Budo.

“Who gave them the right to play God?” Mrs. Everett demanded. “Who gave them the right?”

A successful murder prosecution in Orleans Parish typically requires a coroner’s medical determination of homicide — that a death was caused by the actions of another human being — without regard to fault or legal responsibility. It is a step toward a criminal finding of homicide, in which a Louisiana court assigns fault for a killing.

Minyard, the coroner, brought together Cyril Wecht, Michael Baden — another well-known forensic pathologist — and Robert Middleberg, the director of the toxicology laboratory where the autopsy samples were tested, to discuss the toxicology findings. Minyard’s flood-ravaged offices still hadn’t been repaired, so they met for three days in his temporary quarters in a vacant funeral home.

Records showed that more than half of the 41 bodies from Memorial that were analyzed by Middleberg’s lab tested positive for morphine or midazolam, or both. Middleberg had handled thousands of cases in his career, and the high drug concentrations found in many of the patients stuck out “like a sore thumb,” he told me.

The group considered the 90-year-old pneumonia patient Alice Hutzler, whom the LifeCare nurse Gina Isbell had promised to care for during the hurricane. Morphine and midazolam were found in her liver, brain and muscle tissue, but neither drug had been prescribed, according to her chart, which contained notes until the night before her death on Sept. 1. That chart showed that she was “resting calmly” the previous afternoon, and during the evening her nurses did not document any complaints of pain or distress that indicated she needed the drugs.

Hutzler was one of the nine LifeCare patients found on the seventh floor with one or both drugs in their systems. All were seen alive the morning of Sept. 1, and all were listed as dead by Memorial’s pathologist that afternoon.

“Homicide,” Wecht wrote on a sheet of paper with Hutzler’s name on top, underlining it twice. “Homicide,” he wrote for seven of the eight other seventh-floor patients, including Emmett Everett, Wilda McManus and Rose Savoie. The last patient, whose records indicated she was close to death, he marked as undetermined. Baden said he thought all nine were homicides.

The group considered one death on the eighth floor in the I.C.U.: Jannie Burgess was the comatose patient who was found by Ewing Cook when he climbed the stairs in the heat on Wednesday, Aug. 31. Burgess’s medical chart showed that she was given 15 milligrams of morphine seven times on Wednesday between 2:10 p.m. and 3:35 p.m. on spoken orders from Cook. This was more than seven times the maximum dose she was receiving for comfort care. But because she had already been receiving morphine and because of her advanced cancer, she was “not a clear, strong case,” Wecht wrote in his notes. He marked her death as undetermined.

Besides the nine patients who remained on the LifeCare floor and Burgess, the group also reviewed 13 Memorial and LifeCare patients whose deaths were recorded by Memorial’s pathologist on the second-floor lobby near the A.T.M. and elsewhere. (Other deaths struck investigators as suspicious, but because not all bodies were tested for drugs after autopsy, they were not considered.) Of those 13, 9 tested positive for midazolam and 4 for morphine, too. Investigators searching the hospital found prescriptions for large amounts of morphine for three of them, including Carrie Hall, the woman who fought hard to survive on Wednesday night. The prescriptions were dated Thursday, Sept. 1, and were signed by Dr. Anna Pou.

Despite Wecht and Baden’s strong opinions that the LifeCare deaths were the result of drug injections, Minyard wanted additional information to help him make his decision. He sent the patients’ medical, autopsy and toxicology records to three other experts for an independent review.

“Homicide,” Dr. Frank Brescia, an oncologist and specialist in palliative care, concluded in each of the nine cases. “Homicide,” wrote Dr. James Young, the former chief coroner of Ontario, Canada, who was then president of the American Academy of Forensic Sciences. “All these patients survived the adverse events of the previous days, and for every patient on a floor to have died in one three-and-a-half-hour period with drug toxicity is beyond coincidence.”

A local internal-medicine specialist concluded that while medical records and autopsies for several of the patients revealed medical issues that could reasonably have led to their deaths, most of the patients’ records did not. In his report to Minyard, he wrote that it was "evident" that Emmett Everett was "in stable medical status with no clear evidence that death was imminent or impending." (Pou’s lawyer says that Everett almost certainly died of an enlarged heart, not an overdose of medication.)

As Minyard deliberated, he continued his own inquiry, inviting several LifeCare administrators to his office for interviews. Their stories focused on Anna Pou. Minyard had never met Pou, but two months after her arrest he watched her defend herself and her nurse colleagues with passion on "60 Minutes." "I want everybody to know that I am not a murderer," she told Morley Safer. "I do not believe in euthanasia."

After the "60 Minutes" story, some of Minyard’s longtime colleagues questioned why he was even investigating the case. The day after the CBS broadcast, the American Medical Association released a statement: "The A.M.A. is very proud of the many heroic physicians and other health care professionals who sacrificed and distinguished themselves in the aftermath of Hurricane Katrina."

Minyard told me that after Pou appeared on national television, he had an urge to meet her, to chat over a cup of coffee and try to “get a handle on her.” He had done this before with people accused of crimes. “Science is great, but there is a point where you have to go beyond science; you have to go by your gut feeling, whatever you do.” He invited Pou’s lawyer to bring her to his office for a visit.

Pou sat across from Minyard, “a very ladylike lady, real Southern charming lady.” On his desk was a Bible, on his wall a crucifix, and all around them were framed pictures of life in their native city. Soon they were discovering mutual friends and chatting about several members of Pou’s large Catholic family with whom Minyard was close. They reminisced about Pou’s deceased father, a family doctor who had been especially kind to Minyard and had referred patients to him when Minyard opened his ob-gyn practice.

They talked for about an hour. She told him that she had been trying to alleviate pain and suffering. Given that Pou’s lawyer was there, Minyard was careful not to put her on the spot with direct questions about what she had done. The conditions she described at Memorial took him back to the days he spent trapped in the courthouse after Katrina. How precious food and water had seemed. How impossible it was to sleep at night with gunshots echoing all around.

Minyard told me that his feelings were less sympathetic than he let Pou know. He believed he would have at least tried to save Emmett Everett. There must have been a way to get the 380-pound man downstairs, he said he thought. It also bothered Minyard that documentation suggested that few of the elderly patients who died were being treated for pain.

Minyard reached out to the noted University of Pennsylvania bioethicist Arthur Caplan for more advice. Caplan reviewed the records and concluded that all nine LifeCare patients on the seventh floor were euthanized, and that the way the drugs were given was “not consistent with the ethical standards of palliative care that prevail in the United States.” Those standards are clear, Caplan wrote, in that the death of a patient cannot be the goal of a doctor’s treatment.

Despite all the expert determinations of homicide, Minyard was still struggling with what to tell the grand jury. He consulted one more pathologist, Dr. Steven Karch. Karch had staked his career on advancing the argument that the level of drugs found in a cadaver may have no relationship to the levels just before death.

Karch flew to New Orleans, examined the evidence and concluded that it was absurd to try to determine causes of death in bodies that had sat at 100 degrees for 10 days. In all of the cases, he advised, the medical cause of death should remain undetermined.

The coroner said he believed that if the case went to trial, the defense would bring in someone like Karch to provide reasonable doubt. “We’d lose the case,” Minyard told me. “It would not be good for the city, for the recovery. It’s just a bigger picture that I had to consider than just that pure basic scientific thing.”

Minyard agonized. Willfully taking a life was “a very bad, bad thing,” he thought. “Only God knows when you’re going to die.” The case occupied Minyard’s life, his thoughts and the dreams that awoke him in the middle of the night. He called his experts again and again for support and advice.

The Grand Jury's Decision

In March 2007, the grand jurors who would consider Anna Pou’s fate were sworn in. That spring, they began meeting about once a week at a secret location. Normally prosecutors are advocates for indictment, calling their strongest witnesses to testify and granting immunity in exchange for critical information. But the assistant district attorney, Michael Morales, whose office received condemnatory letters every day for bringing a case against Pou, told me that he and the Orleans Parish district attorney, Eddie Jordan, “weren’t gung-ho” about prosecuting the case. “We were going to give some deference to the defendant,” he said, because Pou wasn’t the usual career criminal accused of murder. At the same time, because a judge had signed a warrant to arrest Pou and multiple witnesses were willing to testify, “we weren’t going to shirk our duties and tank it.” He said that he personally “didn’t care one way or the other” about the outcome.

Rather than presenting the evidence to the jurors and seeking an indictment, as he typically did, he said he invited the jurors, in conjunction with the district attorney’s office, to act as investigators and decide what evidence they wanted to consider. This didn’t sit well with the attorney general and his staff. Foti told me that he repeatedly asked the district attorney’s office to present all the evidence and the experts.

Grand-jury hearings are conducted in secret, making it difficult to know exactly what jurors hear. Minyard told me that in the end, he decided that four of the nine deaths on the seventh floor were homicides, including Emmett Everett and Rose Savoie. Until now, he has never publicly revealed that conclusion. He also said of Pou, “I strongly do not believe she planned to kill anybody, but it looks like she did.”

The jury heard from Minyard but not from any of his forensic experts; nor from two family members who were present on the LifeCare floor during most of the ordeal; nor the main Justice Department investigator, who worked the case for a year and helped collect 50,000 pages of evidence. Only two of the main LifeCare witnesses were brought before the jury late in the process. Budo and Landry, who were compelled to testify after the district attorney decided not to prosecute them, had publicly expressed their support for Pou.

The grand jurors lived among the general public, which was firmly in Pou’s corner. Pou had one of New Orleans’s premier public-relations agencies representing her. A poll commissioned by her lawyer’s office to assess the potential jury pool found that few New Orleanians favored indictment.

Any grand jurors who might have turned on their radios or TVs, or opened The Times-Picayune, or surfed the Web would have heard samples of the community’s drumbeat of support. Nearly every day, New Orleans’s most popular talk-radio host, Garland Robinette, raised his bass voice on WWL’s “Think Tank” in outrage at “what’s being done to these three ... for trying to save lives.” On July 17, 2007, a support rally to mark the first anniversary of Pou’s arrest garnered top billing on Robinette’s show and on every local news program. Hundreds gathered in City Park. Speakers aimed their comments directly at the grand jury, warning that medical professionals, whose ranks had already been depleted by Katrina, would flee Louisiana in droves if a doctor was indicted after serving in a disaster.

The week of the rally, the grand jurors stopped hearing evidence. The district attorney’s office prepared a 10-count bill of indictment against Pou for the grand jury to consider — one count of second-degree murder in Emmett Everett’s case and nine counts of the lesser conspiracy to commit second-degree murder, one for each of the LifeCare patients on the seventh floor.

This meant that the grand jurors were being asked to decide whether the evidence they heard persuaded them that Pou had “a specific intent to kill” — part of Louisiana’s definition of second-degree murder.

On July 24, 2007, the jurors filed into Section E of Orleans Parish Criminal District Court, the building where Minyard survived Katrina. Judge Calvin Johnson read aloud the 10 counts of indictment. The grand jury did not indict Pou on any of them.

Four years after Katrina, it’s summer again in New Orleans, and the myrtle trees are in bloom. Rodney Scott, the patient whom Ewing Cook once took for dead, is still alive.

Scott is grateful to be with his family. A former nurse, he says he does not know whether euthanasia occurred at Memorial; but if it had, he wonders what the doctors and nurses could have been thinking. “How can you say euthanasia is better than evacuation?” he asked me not long ago. “If they have vital signs,” he said, “get ’em out. Let God make that decision.”

The debate among medical professionals about how to handle disasters is intensifying, with Pou and her version of the Memorial narrative often at the center. At a conference for hospital executives and state disaster planners a few months ago in Chicago, she did not mention that she injected patients, saying that helicopters arrived in the afternoon of Thursday, Sept. 1, and “we were able to evacuate the rest.”

Pou projected the booking photo from her arrest onto the screen as she argued for laws to shield health workers from civil and criminal liability in disasters.

Before delivering the keynote address, Pou participated in a panel on the “moral and ethical issues” that could arise if standards of care were altered in disasters. At one point, one of the panelists, Father John F. Tuohey, regional director of the Providence Center for Health Care Ethics in Portland, Ore., said that there are dangers whenever rules are set that would deny or remove certain groups of patients from access to lifesaving resources. The implication was that if people outside the medical community don’t know what the rules are or feel excluded from the process of making them or don’t understand why some people receive essential care and some don’t, their confidence in the people who care for them risks being eroded. “As bad as disasters are,” he said, “even worse is survivors who don’t trust each other.”

Please note that republication of this story is not permitted until September 29, 2009. At that point, all republications must indicate that the story was originally published in The New York Times Magazine.