Key Takeaways
Plan for backup systems to fail, not merely for the storm.
The warning signs were old. Memorial Medical Center stood in a flood-prone part of New Orleans. Its basement had flooded in 1926 and 1927, yet supplies and records were still stored there decades later. Before Katrina, staff knew that relatively little water could disable critical electrical equipment. The hospital had made some protective changes, but lacked the money for others.
The plan missed the worst case. Memorial’s disaster plans assumed generators would run for at least 72 hours, though they had not been tested that long. The plans did not explain how to evacuate patients through flooded streets or respond to total power loss. When the floodwaters reached vulnerable electrical components, the generators ultimately failed. The precise mechanical cause was disputed, but the hospital’s preparedness plainly did not match the danger.
What's striking is the gap between knowing a hazard exists and designing around its consequences. Organizations often prepare for familiar disruptions while treating a cascade of failures as too unlikely to plan for. Resilience engineering asks a harder question: what essential functions must continue when several protections fail together? That shifts planning from a binder of scenarios to practical tests of power, water, communications, transport, and staffing. Memorial's history shows how easily yesterday's disaster becomes background noise. A warning loses force when budgets, routines, and accreditation systems do not require anyone to act on it.
A DNR order refuses resuscitation, not rescue, treatment, or human concern.
One limit became a sorting rule. A Do Not Resuscitate order means that staff should not attempt CPR if a patient's breathing or heartbeat stops. It does not mean the person rejects evacuation, ordinary treatment, or help surviving a disaster. At Memorial, however, doctors decided that patients with DNR orders should be evacuated last, reasoning that they had less to lose. That interpretation was not in the orders or the hospital's disaster plans.
Families were kept in the dark. Some relatives were not told that a DNR status had delayed a patient's rescue. One daughter learned she could withdraw her mother's order only after nurses explained what the evacuation delay meant. She did so, and her mother was moved. The episode shows how a narrow medical instruction became a decision about who deserved access to rescue.
A DNR order is a preference about one intervention, not a verdict on the worth of a life. Conflating the two is a category error with grave consequences: a choice about CPR becomes a proxy for assumptions about prognosis, quality of life, or willingness to survive. The distinction also matters in ordinary care, where advance directives are often treated as shorthand rather than read precisely. Clear documentation and plain-language conversations can prevent that drift. In a disaster, when families may be absent, the case for precise interpretation becomes stronger, not weaker.
Never let a triage label become a permanent verdict on survival.
Triage sorts scarce help. At Memorial, staff eventually marked patients as 1, 2, or 3. Those considered able to sit or walk went first; sicker patients needing more assistance followed; the most ill and many DNR patients were placed last. Triage is the sorting of patients when need exceeds available resources. It can guide urgent choices, but predicting who will survive is uncertain, especially when patients are hot, dehydrated, and separated from their records.
The labels outlasted the assumptions. Patients assigned to the last group received less attention while they waited, and some families were removed from the floor. Yet help later arrived in larger numbers, and staff did evacuate Rodney Scott, a very heavy patient once judged difficult to move. The central danger was treating an early estimate as final instead of reassessing patients as conditions and resources changed.
Triage is not simply a medical calculation. It embeds values: save the most people, prioritize the most urgent, or give everyone an equal chance. The moment a category becomes fixed, however, a provisional forecast can quietly become a self-fulfilling outcome. Reduced attention worsens a patient's condition, which then seems to confirm the original judgment. Good decision systems therefore need repeated reassessment, recorded reasons, and a way to challenge classifications. A crisis may justify prioritizing care, but it does not make human prediction infallible.
Build one shared command system before every hospital becomes an island.
The rescuers were not seeing one picture. Memorial, LifeCare, Tenet, state agencies, FEMA, the Coast Guard, and local officials often relied on different contact lists, priorities, and assumptions. LifeCare staff believed federal officials had included their patients in an evacuation plan; Memorial's request to the Coast Guard covered Memorial patients, not LifeCare's. In some communications, Memorial was first on a list, in others it was near the bottom.
The confusion had consequences. Tenet had no established emergency command system or transport contracts before Katrina. Meanwhile, a Coast Guard helicopter arrived to continue rescues overnight, but hospital leaders declined the offer, in part because of darkness and exhaustion. Patients who depended on ventilators remained on the seventh floor. A clear chain of command, shared patient counts, named contacts, and agreed transport routes could not guarantee rescue, but their absence made coordinated action harder.
Communication is not administrative overhead in a disaster. It is part of the treatment system, as essential as oxygen or electricity because it connects patients to resources. The episode also illustrates a familiar coordination problem: each organization may act reasonably within its own picture while the combined response becomes irrational. Shared situational awareness requires more than frequent messages. Participants need compatible systems, explicit responsibility, confirmed handoffs, and a way to distinguish an offer of help from a completed plan. Without those safeguards, repeated updates can spread confusion faster.
Comfort care cannot quietly become permission to cause death.
The purpose of a drug matters. Morphine can relieve pain and the distress of breathlessness; sedatives can ease severe anxiety. In intensive care, carefully adjusted doses may also suppress breathing, a foreseeable side effect when the goal is comfort. This is often discussed through the principle of double effect: a treatment intended to relieve suffering may have a harmful side effect that is foreseen but not intended. Memorial's decisions were far less clear-cut.
The safeguards were missing. On September 1, Dr. Anna Pou prescribed large amounts of morphine for three patients on the second floor. Staff also administered morphine and midazolam to patients on LifeCare's seventh floor. Some believed the drugs were to ease suffering; others understood that death was intended or would be hastened. Patients' families had been sent away, records were limited, and treatment was not consistently documented or monitored. The dispute was not whether comfort mattered, but whether these acts were comfort care or intentional life-ending treatment.
The boundary between relieving suffering and intending death is ethically important precisely because the same medication can serve different purposes. Intent alone, though, is hard to verify after a patient dies, so practice needs observable safeguards: proportionate dosing, clinical reassessment, documentation, and consultation where possible. The emergency conditions at Memorial made those safeguards difficult, but not every ordinary process had disappeared. That makes the case especially unsettling. Compassion may explain why clinicians act, yet it cannot by itself establish that a choice was safe, authorized, or centered on the patient's wishes.
Make life-and-death judgments slowly enough to see the patient, not just the crisis.
Individual lives resisted the categories. LifeCare patient Emmett Everett was paralyzed and weighed about 380 pounds, so staff worried they could not carry him down the stairs. But he was conscious, ate breakfast, worried about his wife, and asked when evacuation would begin. A nurse remembered his plea not to be left behind. Some staff believed he could not be moved; others pointed out that alternative routes might exist.
The crisis narrowed attention. Doctors and nurses had little sleep, limited records, failing equipment, and frightening rumors about violence. Patients were sometimes evaluated without clinicians knowing their histories. The hospital ultimately evacuated Rodney Scott, another very large patient who had been designated to leave last. These cases do not prove that every patient could have been saved. They show why assumptions about prognosis, disability, or transport difficulty should prompt more inquiry, not end it.
Cognitive overload makes people rely on simple cues: a patient's size, diagnosis, age, or DNR status. Such shortcuts can help organize a chaotic scene, but they can also reproduce bias and conceal practical alternatives. The antidote is not to demand perfect judgment from exhausted clinicians. It is to structure the work so someone can ask whether the patient is conscious, what they want, what options remain, and whether the initial assessment still holds. A decision process should protect both urgency and curiosity.
Treat forensic evidence seriously without mistaking it for certainty about intent.
The evidence raised grave questions. In early testing, nine of 18 Memorial body samples were positive for morphine, and several also contained midazolam. The names in the initial morphine-positive group matched patients whom LifeCare staff had identified as suspicious deaths. Broader testing found morphine or midazolam in 23 of 41 bodies examined. Experts disagreed about how reliably drug concentrations in decomposed tissue could establish the dose or cause of death.
The legal process was contested. Prosecutors arrested Pou and two nurses, but a special grand jury declined to indict Pou. The jury did not hear from every key expert or from Kristy Johnson, a LifeCare employee who said she witnessed injections. A grand jury's refusal to indict is not a finding that nothing happened; toxicology findings, witness accounts, and disputed recollections also do not alone settle every person's intent.
This case demonstrates why evidence has layers. A substance detected in tissue can support a claim that a drug was administered, but interpreting its effect and the administrator's intention requires clinical context, reliable records, and testimony. Decomposition complicates toxicology; memory can shift under stress and after publicity. Yet uncertainty is not a reason to stop investigating. It is a reason to state clearly what is known, what is inferred, and what remains unresolved. Public argument often collapses those distinctions into either innocence or guilt.
Disaster heroism does not absolve institutions of preventable failures.
Many people worked heroically. Nurses carried patients down dark stairwells, staff improvised ventilation for premature babies, and volunteers used boats to reach the hospital. But individual courage unfolded inside systems that had not secured power distribution, arranged sufficient evacuation transport, or prepared corporate leaders to coordinate a response.
History shaped who was vulnerable. Memorial began as a segregated hospital and resisted federal programs before eventually accepting Black patients and Medicare. Jannie Burgess, a Black nurse who had once cared for patients at hospitals where she could not be treated, lay in its ICU during Katrina. After the storm, rumors about armed, violent outsiders influenced decisions to secure the hospital and turn people away. The story asks readers to hold both truths together: caregivers faced extreme conditions, and the institution's failures helped create them.
Hero narratives can honor sacrifice while inadvertently letting organizations escape scrutiny. In safety-critical fields, individual effort often compensates for weak infrastructure until a crisis exceeds what people can improvise. The same dynamic appears in aviation, engineering, and public health: frontline competence can hide systemic underinvestment. Historical exclusion adds another dimension, since decisions about whose danger is believed or whose life is prioritized occur within unequal institutions. Recognizing structural responsibility does not erase personal agency. It makes accountability more complete.
Write rationing rules with the public, and test whether they save lives.
A plan can make an unjust choice look scientific. After Katrina, New York developed rules for allocating scarce ventilators during a severe pandemic, partly using the SOFA score, a measure of organ failure. Researchers later asked doctors to apply such protocols to patients from the milder H1N1 pandemic. Many patients who would have been marked unlikely to survive and denied support actually recovered. Triage judgments also differed among clinicians.
The public can deliberate. Maryland invited residents to discuss which values should guide scarce-care decisions before experts settled on a framework. Participants debated age, family responsibilities, fairness, and social consequences. Their discussions showed that difficult ethical questions are not beyond public understanding. Rules need evidence that they improve outcomes, transparency about whose values they reflect, and ways to revise them when conditions change.
Formal rules can reduce arbitrary decisions, but precision in a score should not be confused with accuracy in predicting a human outcome. A protocol that excludes patients based on weak evidence can make discrimination more consistent rather than less likely. Public deliberation offers a check on professional blind spots, especially when disabled, older, and historically underserved people may bear disproportionate costs. Still, participation alone cannot resolve conflicts or create resources. Its value is to make tradeoffs visible, contestable, and tied to evidence.
Fund practical backups before scarcity becomes a moral test.
Improvisation can preserve options. At public Charity Hospital, staff siphoned fuel from cars to power portable generators, keeping ventilators and monitors running. They used meetings, routine shifts, and humor to steady the hospital and avoided designating the sickest patients as beyond rescue. During Superstorm Sandy, Bellevue staff passed fuel up 13 flights of stairs to keep generators operating. The hospital evacuated patients without having to choose which of six outlets would keep someone alive.
Resources change the ethical problem. At Memorial, some patients were medicated on Thursday morning just as more rescue aircraft were becoming available. Later examples in the book show how low-tech equipment, reserve fuel, and alternative transport can prevent scarcity from becoming absolute. Preparedness is not only about better rules for rationing. It is also about investing in flexible power, supplies, communications, and transport so fewer people must be rationed at all.
The strongest ethical intervention may be engineering rather than moral philosophy: preserve enough capacity that clinicians are not forced to make irreversible choices under extreme uncertainty. Redundancy can seem wasteful in ordinary budgets, but disasters reveal the hidden cost of efficiency built around just-in-time supply and single points of failure. Improvisation matters too, though it should supplement preparation rather than become an excuse for underinvestment. The practical test of a rationing plan is not how elegant its categories look, but whether better preparation can make those categories unnecessary.
Analysis
Five Days at Memorial is investigative history built around a moral question, but its deepest subject is the system that made the question urgent. The narrative moves from Memorial's long history of flood exposure to five days of escalating failures, then into the investigation, legal conflict, and later attempts to write emergency rules. That structure matters: the disputed injections cannot be understood apart from the decisions that left patients in a powerless hospital, the fragmented rescue effort, and triage rules that confused a DNR order with diminished claim to rescue.
The book's unusual strength is its refusal to make one explanation do all the work. Staff members acted with courage and with fear; some tried to comfort patients, some believed death was intended, and memories contradicted one another. Toxicology supported serious suspicion while its limits complicated conclusions about cause and intent. The grand jury's refusal to indict closed one legal path, not the underlying ethical inquiry. This distinction protects the narrative from pretending that either a prosecution or an acquittal could settle what happened to every patient.
The central institutional lesson is that catastrophic choices are often prepared in ordinary times. Underfunded electrical protection, plans written to satisfy accreditation, absent transport contracts, incompatible communications, and reliance on just-in-time supplies shaped the options available after Katrina. This does not remove responsibility from individuals. It locates individual judgment within the architecture that makes some actions easier and others nearly impossible.
The epilogue broadens the case into a warning about crisis standards of care. Written protocols can make decisions more consistent, yet weak predictive tools can wrongly exclude patients and give biased judgments a scientific appearance. Public participation, transparent reassessment, and investment in redundancy are not distractions from emergency medicine. They are ways to preserve trust and keep scarcity from becoming the default explanation for avoidable suffering.
Review Summary
Five Days at Memorial received mixed reviews. Many praised Fink's extensive research and compelling storytelling, highlighting the ethical dilemmas faced by medical staff during Hurricane Katrina. Some found it thought-provoking and eye-opening, while others criticized its length and complexity. The book raised important questions about disaster preparedness, triage decisions, and end-of-life care. Some reviewers felt Fink's portrayal was biased, particularly regarding Dr. Pou. Overall, it was seen as a harrowing account of a tragic event that sparked debate about healthcare ethics in crisis situations.
People Also Read
Glossary
DNR order
No CPR after arrestA Do Not Resuscitate order directs clinicians not to attempt cardiopulmonary resuscitation if a patient's breathing or heartbeat stops. It does not, by itself, refuse other treatment, comfort care, or evacuation. At Memorial, the distinction became crucial when some staff treated DNR status as a reason to move patients last.
triage
Sorting patients by priorityTriage is the process of ranking patients for care or evacuation when needs exceed available resources. Different systems prioritize urgency, likelihood of benefit, or other values. The book emphasizes that predictions can be wrong, circumstances change, and a category should be reassessed rather than treated as a final judgment.
reverse triage
Moving the more able firstIn the Memorial case, this term described prioritizing patients thought more likely to survive or easier to move ahead of the sickest. The term can also refer to theoretical emergency approaches that prioritize patients who could return to useful work. The book stresses that this was not the usual battlefield approach and that the label does not resolve whether the choices were justified.
double effect
Relief with foreseen harmThe double effect principle distinguishes an intended therapeutic benefit from a harmful side effect that is foreseen but not intended. In end-of-life care, a clinician may use medication to relieve pain or breathlessness while knowing it could suppress breathing. The distinction depends on purpose and practice, and it cannot be established simply by describing a drug as compassionate.
palliative sedation
Sedation to relieve severe symptomsPalliative sedation uses medication to reduce consciousness when severe symptoms cannot otherwise be relieved, with the aim of easing suffering rather than causing death. The book notes that accepted practice involves safeguards such as careful dosing, documentation, and consultation when possible. It distinguishes this from rapidly giving large doses without clear records or patient consent.
SARBOO
Search-and-rescue staging siteA Search-and-Rescue Base of Operations was a staging point outside the flooded area where helicopters could quickly drop evacuees and return for more. The plan depended on buses, drivers, supplies, and medical support at the site. At the New Orleans cloverleaf, many people were delivered without adequate onward transport or basic services.
SOFA score
Organ failure severity measureThe Sequential Organ Failure Assessment score tracks dysfunction across several organ systems in critically ill patients. New York's proposed pandemic ventilator rules used it to help guide access to scarce life support. The book describes important limits: the score was not designed to predict survival, and studies suggested that some patients it would classify as unlikely to benefit recovered with continued treatment.
Hurricane Pam
New Orleans disaster-planning exerciseHurricane Pam was a planning exercise that modeled catastrophic flooding in the New Orleans region and anticipated thousands of hospital patients needing help. It raised questions about evacuation order, transport, and staging sites. The exercise captured important risks, but officials had not completed the systems needed to carry out the modeled response.
Download PDF
Download EPUB
.epub digital book format is ideal for reading ebooks on phones, tablets, and e-readers.