The patient deaths at Memorial Medical Center during Hurricane Katrina occurred amid a failing emergency power system, a disrupted evacuation, and extreme difficulty maintaining care. Their significance lies not in a single settled explanation for every death, but in the collision between a known infrastructure risk, the deaths of patients across different hospital areas, and an investigation that had to distinguish suspected medication-related deaths from deaths whose causes remained uncertain.

A known vulnerability and a failing power system

Months before Katrina, plant operations director Eric Yancovich had assessed how floodwater could threaten Memorial. After studying storm-surge models with the U.S. Army Corps of Engineers, he sketched the elevations of hospital entrances and outdoor equipment against Magnolia Street. His calculations indicated that less than four feet of street flooding could carry water over the loading dock and into the hospital. The generators themselves stood on the second floor, but parts of the emergency power distribution system, including transfer switches, were at or near ground level. Yancovich warned that relatively little water could disable most of the medical center and predicted that patients might have to be moved to the newer Clara Wing. Although some defenses were improved, proposed work to raise switches and pumps was not expected to be approved for lack of capital.

During the crisis, Memorial’s three 750-kilowatt generators had already been operating for two days, although the disaster plan expected them to support the hospital for at least three. The staff on site were not generator specialists. Workers first restored a failed engine by adding water to its radiator; later they tried to address apparent oil-pressure and fuel problems, while power weakened in parts of the hospital. When another generator failed, engineers attempted to transfer some of its electrical load to a working unit. They waded through knee-deep basement water to reach the electrical system, where submerged transfer switches and distribution panels were consistent with the vulnerability Yancovich had identified.

The attempted workaround strained the remaining equipment. A gauge indicated that the working generator was drawing too much current, raising concern about a short circuit; workers tried to separate the loads and shut off circuits serving unused equipment. Another generator failed, and efforts to repair it with parts salvaged from the failed unit did not succeed. At about two in the morning on August 31, nearly forty-eight hours after Katrina made landfall near New Orleans, the last backup generator surged and stopped. The hospital’s alarms sounded, and the loss of power compounded an evacuation already unfolding under dangerous conditions. The technical failure does not by itself establish the cause of any particular patient’s death, but it formed the circumstances in which care, movement, and later assessments took place.

Deaths across the hospital

After the evacuation, Memorial pathologist John Skinner and laboratory director David Heikamp walked through the building to check that people had left and to record the dead and their locations. Medicine chairman Richard Deichmann and CEO René Goux had asked Skinner earlier in the week to keep track of deaths. The resulting notes and the men’s recollections later mattered because they helped investigators establish when and where patients had last been seen alive and when their deaths were recorded.

On the seventh floor, the records and witness accounts centered on nine LifeCare patients. LifeCare nurse Kristy Johnson later described accompanying a Memorial nurse into room 7305, where Rose Savoie, a ninety-year-old patient with acute bronchitis and a history of kidney problems, was injected. Savoie said that it burned, and Johnson saw swelling in her arm. A LifeCare nurse told investigators that Savoie and another patient had been alert and stable late that morning. By that afternoon Savoie and the other remaining LifeCare patients were dead. Johnson’s account became important to Virginia Rider and Butch Schafer’s investigation, while the hospital’s death records supplied another line of evidence about the timing.

The forensic review found morphine or midazolam, or both, in twenty-three of the forty-one Memorial and LifeCare bodies tested by toxicology laboratory director Robert Middleberg. That finding prompted close scrutiny, but it did not make interpretation straightforward. The bodies had remained in heat before sampling, which could have altered measured drug concentrations. Middleberg advised the investigators and experts to assess each result alongside the patient’s medical history. They considered whether drugs had been given in a single large amount or in repeated doses, whether kidney or liver disease might have affected drug accumulation, and whether respiratory effects could fit the available timelines. The tests therefore raised questions requiring case-by-case review rather than providing a uniform account of all the deaths.

Differing assessments of medication evidence

Alice Hutzler, a ninety-year-old LifeCare patient being treated for pneumonia, was a particularly consequential case. Morphine and midazolam were found in her liver, brain, and muscle tissue, although her chart showed that neither had been prescribed. Her records described her as resting comfortably on Wednesday afternoon and documented no later complaints of pain or distress indicating a need for those drugs. She was among the nine patients on the seventh floor who were seen alive on the morning of September 1 and listed as dead that afternoon.

Pathologist Cyril Wecht classified Hutzler’s death as homicide in his notes, as he did the deaths of seven of the eight other seventh-floor patients, including Emmett Everett, Wilda McManus, and Rose Savoie. He marked the remaining patient, Elaine Nelson, whose records suggested that she was close to death on Thursday morning, as undetermined. Forensic pathologist Michael Baden considered all nine deaths homicides. These were expert assessments based on the available evidence, not findings that erased the differences among the patients’ conditions or the limits of the records.

The investigation had initially focused its arrest allegations on four LifeCare patients: Everett, Savoie, Ireatha Watson, and Hollis Alford. Watson, eighty-nine, had dementia and had been treated for gangrenous toes; Alford, sixty-six, had schizophrenia in his history and was gravely ill with a blood infection. Toxicology detected morphine and midazolam in the four patients named in the affidavit, and Wecht concluded that they had died from lethal amounts of the drugs. The affidavit also relied on witness accounts concerning the intended evacuation of the LifeCare patients, medication, and what staff observed. Anna Pou’s attorney, Rick Simmons, later disputed one reported statement attributed to Pou, underscoring that parts of the account were contested.

The eighth-floor intensive care unit presented a less conclusive picture. Jannie Burgess, a seventy-nine-year-old nurse with advanced metastatic uterine cancer, received seven documented 15-milligram morphine doses over an hour and a half on August 31, on spoken orders from her physician, Ewing Cook. The total, 105 milligrams, exceeded five times the largest amount she had previously received in a full day during her hospitalization. Yet Burgess had already been receiving morphine, had advanced cancer, and was not, in Wecht’s view, a clear case. He marked her death undetermined and advised against including it in a legal case. Two other ICU patients left behind also had morphine in their systems, but at lower concentrations; Wecht did not consider their deaths homicides.

The second floor and the limits of the record

Experts also reviewed nine Memorial and LifeCare patients whose deaths Skinner recorded in the second-floor lobby during the September 1 evacuation, as well as Merle Lagasse. All tested positive for midazolam, and five also tested positive for morphine. Among them were Carrie Mae Hall, Wilmer Cooley, and Donna Cotham, three patients whose charts included prescriptions for large amounts of morphine signed by Pou. Two of those three tested positive for midazolam but not morphine. Cotham’s autopsy report noted a recent needle puncture at the bend of her left arm, while decomposition made it difficult to assess her severe liver disease.

Essie Cavalier, a stroke patient whose former nurse had been troubled by what she saw being done to her on the second floor, also had what Wecht considered high levels of morphine and midazolam in her tissues. He marked her death and many other second-floor deaths as likely homicides pending review of the hospital records. That qualification mattered: investigators had not obtained many of the relevant Memorial charts. Tenet said it had been unable to locate some records, and suggested they might have been removed with the bodies. A mortuary worker who helped remove the bodies told Rider that the records had not been taken with them.

The review also documented findings that argued against treating every hospital death as part of one pattern. A longtime patient of Horace Baltz had no morphine or midazolam in her system; the available toxicology suggested a natural, if unexpected, death. Experts believed that about five of the forty-five adult bodies recovered had been patients who died just before the disaster. Seven apparently stillborn specimens were recovered but were not included in their review. Five bodies had no toxicology results. At least one was a Memorial patient found dead near the second-floor ATM on Thursday afternoon after an infectious-disease physician had seen her alive that morning; investigators could not establish whether she had received the drugs.

The investigation thus produced both grave allegations and boundaries to what could be known. Witness accounts, medication records, toxicology, clinical histories, and the timing of death did not align into a complete record for every patient. The deaths became a central criminal and forensic controversy precisely because the emergency conditions were severe, the evidence varied from case to case, and the distinction between suspected lethal medication and other causes could not be resolved with equal certainty for everyone.