Two well documented events from the 2005 hurricane season bracket the hardest decision a long-term care administrator or hospital incident commander can face. At St. Rita’s Nursing Home in St. Bernard Parish, 35 residents died after the facility stayed in place for Katrina and the building flooded. Less than a month later, 23 elderly evacuees died when their bus caught fire on Interstate 45 during the Rita evacuation. This guide uses both events to cover the shelter-in-place decision, transport, and receiving facility coordination. It explains why neither staying nor leaving is a safe default for medically fragile people.

St. Rita’s, August 2005: a decision to stay and 35 deaths

St. Rita’s Nursing Home sat in St. Bernard Parish, east of New Orleans, in an area that Hurricane Katrina flooded catastrophically on August 29, 2005. The facility was owned and operated by Salvador and Mabel Mangano. The owners decided not to evacuate the residents ahead of the storm, and the building took on water after the storm surge and the failures of the levee and floodwall system that protected the parish. Accounts of the event describe the water rising inside the building fast enough that staff and the people who came to help could not move most of the residents who were bedbound or could not walk. Some residents were rescued, while 35 died. That figure, 35, is the one consistently reported for St. Rita’s in news coverage and in accounts of the prosecution that followed, and it is the number I use here.

Beyond the outcome, St. Rita’s matters for planning because of how the decision was framed at the time. Accounts of the owners’ defense describe a judgment that moving frail, elderly residents a long distance by road carried its own serious risk of death. Their lawyers pointed to earlier evacuations in which residents had suffered. That concern was not invented for the courtroom. As later sections show, evacuation does kill frail people, and the Rita bus fire weeks later made the point in the worst possible way. The flaw at St. Rita’s was that the decision to stay did not rest on a building and a site that could survive the hazard actually threatening it. A single-level facility in a parish that sits largely below sea level offers no vertical refuge if the water comes in. Accounts describe the building as single-story, and I would point readers to the trial coverage for the physical details.

I am not going to reconstruct the hour-by-hour sequence inside the building. Accounts differ on timing, and the people who were there gave their testimony in court. For a planner, the essential fact is simpler and is not in dispute. A shelter-in-place decision for a facility full of people who cannot walk is only as good as the facility’s ability to keep them alive through the worst plausible version of the hazard. At St. Rita’s, the worst plausible version was rapid flooding of a low building with no upper floor, and nothing in a sheltering plan can work around that once the water starts rising.

The Mangano prosecution: what the acquittal settled and what it left open

The office of Louisiana Attorney General Charles Foti charged Salvador and Mabel Mangano with negligent homicide and cruelty to the infirm. A jury acquitted both of them in September 2007. Accounts of the trial describe a defense built on several arguments. The first was that evacuation itself posed a mortal risk to the residents. The second was that the flooding resulted from the failure of a federally built flood protection system, which the owners could not have foreseen at the scale it reached. There was also a dispute over what the parish had ordered or offered before the storm. I have seen that last point reported in conflicting ways, so I am not stating what the parish did or did not offer. The trial record and the contemporary coverage are the places to read it.

The acquittal settled the criminal question for those two defendants and nothing more. It did not establish that staying was the right call. Nor did it establish that the deaths were unavoidable, or that the same decision would be defensible for another facility in another place. Readers sometimes treat an acquittal as a finding that nobody did anything wrong. A criminal jury, though, decides only whether the prosecution proved the elements of the charged offenses beyond a reasonable doubt. Civil litigation over Katrina nursing home deaths ran on separate tracks with different standards, and I am not summarizing its outcomes.

For emergency management, the prosecution’s lasting effect was on behavior. After St. Rita’s, administrators across the Gulf Coast understood that a decision to stay could be examined in a criminal court if residents died. Accounts from the period, including interviews with Louisiana nursing home administrators published in the medical literature, describe that pressure pushing facilities toward evacuation in later storms. That pressure has a cost of its own, and it brings this guide to Texas.

The wider Katrina count: why there is no single number, and Memorial

The 35 deaths at St. Rita’s form one count at one facility. The number of nursing home and hospital deaths connected to Katrina as a whole is a different question, and published figures for it vary. I am not giving a total, and I would caution anyone against blending the St. Rita’s figure into a larger number or quoting a larger number as if it had the same standing. The variation has identifiable causes. Some counts include only deaths inside facilities during the flooding. Others add residents who died during evacuation, and others add residents who died in the days and weeks after relocation from conditions the move may have worsened. Counts also differ on whether they cover Louisiana alone or Mississippi as well, and on whether assisted living and group homes are included with licensed nursing facilities.

Excess mortality after relocation is the hardest category to count, because a 90-year-old with heart failure who dies three weeks after an evacuation is not recorded on a death certificate as a hurricane death. Researchers led by David Dosa at Brown University have studied mortality among Gulf Coast nursing home residents after the 2005 storms and later hurricanes. As I read their work, they found evacuation associated with higher mortality and hospitalization in the following weeks and months compared with residents who sheltered in place, though the size of the effect depends on the storm and the methods. Anyone who needs figures should read those papers directly rather than take a number from me. The Louisiana Department of Health and the HHS Office of Inspector General are the agencies to consult for official counts and for the federal review of nursing home preparedness after the 2005 hurricanes.

Hospitals had their own catastrophe. At Memorial Medical Center in New Orleans, which also housed a separately operated long-term acute care unit run by LifeCare Hospitals, the building lost power and became isolated by floodwater. Patients and staff were trapped for days in extreme heat while rescue proceeded slowly. Forty-five bodies were widely reported recovered from the hospital afterward. Dr. Anna Pou and two nurses were arrested in 2006 on allegations related to some of those deaths. In 2007 a New Orleans grand jury declined to indict Pou, and the charges against the nurses were dropped. Sheri Fink’s reporting, collected in her 2013 book Five Days at Memorial, is the most thorough public account. I am not characterizing what happened to individual patients beyond what that legal outcome records.

The Rita bus fire near Wilmer: 23 deaths on the way out

Hurricane Rita approached the Texas and Louisiana coast in late September 2005, less than four weeks after Katrina. The Houston area produced an enormous evacuation, much of it by people outside the formal evacuation zones, with pictures from New Orleans fresh in their minds. Among those evacuating were residents of an assisted living facility in Bellaire, in the Houston area, commonly identified as Brighton Gardens. They were traveling by chartered motorcoach toward the Dallas area. Early on September 23, 2005, the coach caught fire on Interstate 45 near Wilmer, south of Dallas. According to the National Transportation Safety Board, 23 passengers died.

The NTSB investigated and issued a final report containing the board’s probable cause determination, its findings and its safety recommendations. I am not restating or paraphrasing the probable cause here, because the exact wording matters and a secondhand summary tends to drift from it, so anyone who needs it should read the report on the NTSB website and the supporting material in the docket. What can be described without reaching into causation is the situation on board. The passengers were elderly residents, many of them with limited mobility and some needing staff help to stand, and medical oxygen cylinders were carried in the cabin for residents who used oxygen. The report addresses the oxygen on board, the difficulty of getting passengers with limited mobility off the coach, and federal oversight of the motor carrier, which coverage commonly identifies as Global Limo, Inc. I list those topics so that a planner knows what to look for in the report, and the list is not a summary of what the board concluded about any of them.

I am not assigning fault for the night of the fire, because the NTSB report is the record for that. The planning point stands apart from causation. The residents were moved because their operators decided, during a storm that was frightening everyone on the Gulf Coast, that leaving was safer, and once that decision was made they depended on a commercial motorcoach whose maintenance nobody standing at the curb could evaluate. Passengers who needed help to stand would have had to get down a coach stairwell in an emergency, and pressurized oxygen rode in the cabin with them. The choice of vehicle and carrier, the mobility of the passengers and the oxygen on board were all knowable before departure, which is why each of them belongs in the evacuation plan of any facility that intends to move frail people by road.

How the move itself hurts: transfer, heat, oxygen and lost records

Clinicians who care for frail elderly patients recognize the harms of relocation even without a fire or a crash. A resident who lies for many hours on a stretcher or sits in a bus seat with no repositioning is at risk of pressure injury, and residents with dementia often become acutely confused in unfamiliar surroundings. Fluids, meals and scheduled medications tend to be interrupted, and dehydration develops quickly in heat. Accounts of the Rita evacuation describe trips from Houston to Dallas that normally take a few hours stretching far longer in gridlocked traffic. Reports from that evacuation attribute deaths to heat in stalled traffic. I am not giving a figure, because tallies of Rita evacuation deaths differ depending on who counted and what they included.

Oxygen is a specific hazard that planners underweight. A facility with several residents on continuous oxygen will put multiple cylinders or concentrators into a vehicle. Cylinders in a passenger compartment add to fire load and can intensify a fire. Concentrators need power, and a bus may not supply it for the duration of a long trip. The supply has to be calculated for the worst trip duration rather than the normal one, with the cylinders secured. A medical professional also has to be assigned to manage them on board.

Records and identification fail in ways that harm residents for weeks. A resident who arrives at a receiving facility without a medication list, an advance directive, allergy information or even a reliable name band forces the receiving staff to rebuild a clinical picture from nothing. Residents with dementia cannot help them do it. After Katrina, families spent days or weeks locating relatives who had been moved, and some residents arrived at receiving facilities with no documentation that identified them. Paper copies of essential records traveling with each resident remain the most dependable approach, because electronic records depend on network access that the disaster may have removed.

Shelter or evacuate: the facts the decision actually rests on

The decision starts with the building and the hazard, and the hazard has to be specific. A facility that is structurally sound for wind may still be in a storm surge zone, and a building outside the surge zone may still flood from rainfall or a levee failure. The administrator needs to know where the facility sits on the surge inundation maps, and the local emergency manager can usually get that answer from the state emergency management agency or the National Hurricane Center products. The administrator also needs the elevation of the lowest occupied floor and whether the building has upper floors where residents can be moved. A two-story facility with a functioning elevator on emergency power has options that a single-story facility in the same zone does not.

Power is the next question. Many residents depend on electricity for oxygen concentrators, suction, feeding pumps, pressure-relief mattresses, and cooling in summer heat. The generator’s fuel supply, its location, and whether its transfer switch and fuel system sit above the flood level all determine whether sheltering in place can last. Memorial’s loss of power after flooding shows what happens when the electrical system is below the water. Sheltering also depends on staff who will stay, on enough supplies of food, water and medication for the expected period of isolation, and on a way to communicate when landlines and cellular service fail.

The residents themselves are part of the calculation. A facility caring mostly for ambulatory residents faces a very different evacuation from one with many bedbound, ventilator-dependent or bariatric residents. The higher the acuity, the greater the risk of the move and the more specialized the transport required. That pushes toward staying when the building can genuinely protect people. When the building cannot, it pushes toward leaving early, because the most vulnerable residents need the most time and the most specialized vehicles, and those vehicles run short first.

Timing ties the factors together. Evacuating frail residents takes far longer than evacuating the general public, so the decision has to be made before the general evacuation order, while roads are open, weather is tolerable and contracted vehicles are still available. Federal rules now require plans for both options. The Centers for Medicare and Medicaid Services published its emergency preparedness rule for participating providers in September 2016, with compliance required from November 2017. The rule requires covered facilities, including hospitals and long-term care facilities, to maintain plans and procedures that address both evacuation and sheltering in place. It has been revised since, so verify the current requirements with CMS and your state survey agency.

Transport: double-booked contracts, the wrong vehicles and the long ride

Most facility evacuation plans rest on a contract with a transportation provider, and the most common failure in that arrangement shows up only during a real regional event. A bus company or ambulance service may hold evacuation agreements with many facilities at once. Each agreement is valid on paper, but the provider has only so many vehicles and drivers. When every facility in a coastal region calls on the same day, the contracts cannot all be honored. The federal review of nursing home preparedness after the 2005 hurricanes, as I read it, identified failures of contracted transportation among the problems facilities reported. Drivers who live in the threatened area may also be evacuating their own families, which shrinks the available fleet further.

The vehicle has to match the residents. Ambulatory residents can ride a motorcoach. Wheelchair users need lift-equipped vehicles with securement points, and bedbound residents need stretcher-capable transport, which in most regions means ambulances or specially configured buses. The number of stretcher-capable vehicles is small relative to the number of bedbound people in a large metropolitan area. Ambulances also have to keep up with ordinary emergency calls while an evacuation is under way, so a plan that assumes ambulances for every nonambulatory resident is counting on a resource the whole region is drawing on at once.

Carrier vetting belongs at the contract stage, and Wilmer is the reason to take it seriously. In an emergency, facilities take whatever vehicle arrives, and nobody on the curb is in a position to judge a carrier’s maintenance program. That judgment has to be made beforehand, when the contract is signed, by checking the carrier’s federal safety record and asking how it maintains its fleet. Staff have to ride with residents and must be able to assist them out of a vehicle in an emergency. The plan should also state who on the vehicle is responsible for the oxygen, the medications and the resident list, so that those duties are not left to a driver who has no clinical training.

Receiving facilities: beds, records, tracking and the state’s role

Evacuation succeeds only if there is somewhere to go that can care for the residents on arrival. Many facility plans name a sister facility or a receiving agreement. In a regional event, though, the receiving facility may sit inside the same threatened area, may already be full, or may have promised the same beds to several other facilities. A receiving agreement that has not been reconfirmed recently is unreliable, and an agreement with a facility within the same hazard footprint provides no protection at all. Planners should look for receiving sites far enough away to be outside the event while near enough that the trip does not become its own hazard.

Receiving facilities face licensing, staffing and reimbursement constraints when they take residents beyond their normal capacity. Section 1135 of the Social Security Act allows the federal government to waive certain Medicare and Medicaid requirements during a declared emergency, and states can issue their own licensing waivers. The emergency manager does not grant either. Under Emergency Support Function 8, which covers public health and medical services, the state health department typically coordinates health care facility evacuation, patient tracking and bed availability. In most states, the regional health care coalitions funded through the federal Hospital Preparedness Program are where hospitals, nursing homes, EMS and emergency management work out these arrangements in advance. Verify with your state health department how its patient tracking and bed reporting systems work, because they differ by state.

Tracking is the piece families experience most directly, and it is the piece that broke down most visibly after Katrina. Each resident needs an identifier that travels with them and records showing where they went. Someone at the sending facility has to keep the list current as residents move to hospitals, receiving facilities or shelters. When the sending facility is destroyed or loses communications, that list has to exist somewhere else, which is why the state system and the coalition matter. The receiving end also needs staff, so plans should provide for sending facility staff to accompany residents and help care for them.

What to do at your agency

• The county emergency manager requests from each nursing home and assisted living administrator in the jurisdiction a copy of the transportation agreement in the facility’s evacuation plan. The manager compiles a one-page list showing which vendors appear in more than one facility’s plan, and that list goes to the regional health care coalition coordinator.
• The EMS chief or operations officer counts the stretcher-capable and wheelchair-capable vehicles actually available in the county on a weekday afternoon, including private services. That count is written into the health and medical annex of the emergency operations plan next to the number of bedbound residents the facilities report.
• The emergency manager confirms with each facility administrator whether the facility sits in a mapped storm surge or flood inundation zone. The manager also records the elevation of its lowest occupied floor and whether its generator and transfer switch are above that flood level.
• The fire marshal or the officer who reviews facility preplans adds a line to each long-term care preplan. The line records the number of residents on continuous oxygen and where the facility stores its cylinders, so that crews know before an evacuation or fire what pressurized oxygen is in the building.
• The ESF-8 lead or public health emergency preparedness coordinator asks the state health department how its patient tracking system records residents who leave a facility during an evacuation. The coordinator then writes one paragraph into the county health and medical annex naming that system and the local contact who reports into it.
• The emergency manager adds a review of facility receiving agreements to the agenda of the next health care coalition meeting already on the calendar. Each facility is asked to state whether its receiving site lies outside the same hazard footprint and when the agreement was last reconfirmed.

Takeaways

• Thirty-five residents of St. Rita’s Nursing Home in St. Bernard Parish died after the facility sheltered in place for Katrina and the building flooded in August 2005.
• The owners of St. Rita’s were charged by the Louisiana Attorney General’s office and acquitted by a jury in September 2007, which resolved the criminal case without establishing that staying was the right decision.
• Totals for Katrina nursing home and hospital deaths vary by source because counts differ on location, timing and facility type, and they should never be blended with the St. Rita’s figure.
• According to the NTSB, 23 elderly evacuees died when their motorcoach caught fire near Wilmer, Texas, on September 23, 2005, during the Rita evacuation, and the board’s final report is the place to read its probable cause and recommendations.
• Researchers including David Dosa’s group at Brown University have associated nursing home evacuation with increased mortality in the following weeks, so evacuation carries real risk even when the trip goes smoothly.
• Sheltering in place is defensible only when the building, its elevation, its power supply and its staffing can carry residents through the worst plausible version of the specific hazard.
• Transportation contracts fail in regional events when several facilities rely on the same vendor, and stretcher-capable vehicles run short before anything else.
• Receiving agreements, patient tracking and records that travel with each resident determine whether an evacuation ends with residents in safe care or lost in the system.