Skip Navigation U.S. Department of Health and Human Services
Agency for Healthcare Research Quality
Archive print banner
Hospital Evacuation Decision Guide

Public Health Emergency Preparedness

This resource was part of AHRQ's Public Health Emergency Preparedness program, which was discontinued on June 30, 2011, in a realignment of Federal efforts.

This information is for reference purposes only. It was current when produced and may now be outdated. Archive material is no longer maintained, and some links may not work. Persons with disabilities having difficulty accessing this information should contact us at: Let us know the nature of the problem, the Web address of what you want, and your contact information.

Please go to for current information.

Chapter 3. Pre-Event Evacuation Decision Guide

A pre-event evacuation may be carried out in anticipation of an impending event, when the hospital structure and surrounding environment have not yet been compromised. A pre-event evacuation is appropriate when decision teams believe the effects of the impending disaster may either place patients and staff at unacceptable risk, or when an evacuation after the event is likely to be extremely dangerous or impossible.

Pre-event evacuations are an option in Advanced Warning Events—disasters that decision teams and emergency officials can anticipate and track, as they assess the possible consequences of the disaster on their hospital and the surrounding community. Hurricanes are the most common example of Advanced Warning Events, and decision teams may decide to evacuate prior to hurricane landfall. Wildfires, rising rivers, and bomb or other terrorist threats can also force decision teams to decide to preemptively evacuate their hospital. If decision teams elect not to preemptively evacuate—deciding instead to shelter-in-place—a post-event evacuation may become necessary, depending on the impact of the event on the hospital and surrounding area. (Post-event evacuations are discussed in Chapter 4.) Thus, an Advanced Warning Event frequently requires two evacuation decisions: one pre-event and another post-event. By contrast, an event with no advanced warning involves only the post-event decision.

Figure 1 shows a flowchart that illustrates both the pre-event and post-event evacuation decisions that an Advanced Warning Event may require. There are several possible "paths" through the Figure 1 flowchart, including ordering a pre-event evacuation following a wait and reassess period; deciding to shelter-in-place, with no subsequent evacuation required; and deciding to shelter-in-place following a wait and reassess period, and then subsequently ordering a post-event evacuation.

The flowchart begins with an initial consideration of the decision to order a pre-event evacuation. Typically, this would occur as soon as a disaster is identified that could potentially threaten a hospital. This is often days before the disaster "hits," such as when a hospital is inside a 3-day projected hurricane path. The flowchart highlights the three possible outcomes of this decision: wait and reassess, start an evacuation, or make an explicit decision to shelter-in-place during the event.

Return to Contents

Wait and Reassess

The wait-and-reassess option defers the decision of whether to start a pre-event evacuation and is typically the preferred option early in the tracking period, when the disaster is not yet an immediate threat. The wait-and-reassess option is predicated on decision teams' belief that after reassessment, there will still be ample time remaining for an evacuation, if it is needed. In this option, disaster tracking meetings are held regularly, and decision teams cycle through the flowchart loop of "Order Pre-Event Evacuation?" and "Wait and Reassess," possibly several times (go to Figure 1).

Decision teams rely on emergency management officials for accurate information about both the expected time and magnitude of the event, as well as explicit quantification of the uncertainty of estimates. Of course in some situations, such as a verified bomb threat, there may be no time to "wait and reassess."

In the wait-and-reassess option, the expected time until the event occurs should be compared to the time required to evacuate patients from the building and safely transport them to other facilities to determine if the decision to evacuate can be deferred. The evacuation time assumptions generated as part of the Pre-Disaster Self-Assessment (go to Chapter 2) provide estimates for the time required to safely evacuate. These assumptions should be revisited based on current conditions in the hospital and the expected impact of the event.

Hospitals Closely Monitor Track of Hurricane Rita

The University of Texas Medical Branch (UTMB) initiated its hurricane preparedness procedures on Sunday, September 18, 2005, 5 days prior to landfall of what was then Tropical Storm Rita. When Rita was officially classified as a hurricane on Monday and was headed towards UTMB's location on the upper Texas coast, the medical center initiated census reduction efforts to discharge as many ambulatory patients as possible. On Tuesday, 3 days prior to landfall, the incident command center opened in accordance with UTMB's emergency operations plan. All unit-level emergency plans were also activated, biosafety level three and four labs were closed and decontaminated, and medical students and nonessential personnel were dismissed. Late Tuesday evening, a teleconference was held between UTMB and the Texas Department of Public Safety's Division of Emergency Management, during which the Department guaranteed UTMB adequate ground and air transportation for a full evacuation, if the evacuation decision was made by 7:00 the following morning. From Tuesday evening until the time of the final decision Wednesday, UTMB assessed and triaged patients, copied medical records, assembled patient medication lists, and contacted a large hospital network to determine the number of transfers they could accept. At 7:00 a.m. Wednesday, 2 days prior to landfall of Hurricane Rita, a hospital-wide evacuation was ordered.11

Specific items to consider in the reassessment of the time required to evacuate patients include the following:

  • Current patient census and mix. How does the current patient census differ from the assumptions used to estimate evacuation time and resource needs in the self-assessment?
  • Availability of ambulances, wheelchair vans, and buses. Are previous assumptions about the availability of transportation resources still valid? Are alternative sources of transportation resources available? Are other hospitals currently evacuating or planning to evacuate patients?
  • Location of facilities able to receive your hospital's patients. Are the intended receiving care sites still able to accept patients? What alternative receiving care sites are available to accept patients?

When there is time—particularly in the days prior to a hurricane—decision teams usually discharge any patients who can safely be released to their families and stop admitting new patients. This is called census reduction, and it reduces the demands on the hospital as the focus turns to evacuating patients who require ongoing care.2 Census reduction may take place before an official pre-event evacuation order is given, as in advance of hurricane; early discharges may also occur after a No Advanced Warning Event, such as an earthquake.3,5,46 Census reduction is a deliberate strategy to reduce the number of patients a hospital is responsible for transferring to other facilities.

Bomb Threat at Galion Community Hospital

Galion Community Hospital (Ohio) received a bomb threat at 9:30 a.m. one Wednesday morning in 1999. After the threat was announced over the hospital's intercom system, members of the Incident Command System team met to discuss a possible evacuation while the local police and fire department worked with the facilities engineers to search the building for a bomb, which they did not find. An hour and 10 minutes after the initial threat, the hospital received a second warning. Within the next 5 minutes, the evacuation decision was made based on recommendations from consultants, police, and fire department officials.5

Return to Contents

Pre-Event Evacuation or Shelter-in-Place?

The wait-and-reassess option is viable for only a limited period of time; as the event progresses and conditions deteriorate, patients will not be able to be evacuated safely due to, for example, hurricane force winds or impassable roads. At some point, decision teams must decide whether to evacuate the hospital or shelter-in-place during the event. In our review of the literature and expert interviews, by far the most common decision during the approach of an Advanced Warning Event is to shelter-in-place. We note, however, that in the years since Hurricane Katrina, decision teams may be more inclined to evacuate in advance of a major storm.

Pre-Event Evacuation Decisions

During a flood of the Red River in March 2009, administrators at Merit Care Hospital in Fargo, North Dakota, first reduced the patient census to 180 high-risk patients who would have the most difficulty evacuating. When the predicted height of the river's crest rose dramatically, and a nearby dyke was jeopardized, Merit Care administrators decided to fully evacuate. A key factor in this decision was the concern that a later evacuation would force the hospital to compete for available ambulances and buses with other evacuating groups.47

Rising water from the Cedar River, and uncertainty associated with the eventual crest, led to the evacuation of Mercy Medical Center in Cedar Rapids, Iowa, in June 2008.48 Administrators initiated an evacuation of all 176 patients upon determining that the facility was likely to lose power.49

In October 2007, the rapid spread of wildfires in San Diego County caused the evacuation of 77 patients from Pomerado Hospital, as administrators closely monitored the conditions of the fire throughout the afternoon and evening.13 The decision to evacuate was made because the fire was visible from the hospital's grounds,50 and the fire department might not be able to protect the building due to the community-wide disaster.13

Deciding to Shelter-in-Place

Despite rising flood waters and disrupted roadway systems, Innovis Health administrators decided to shelter-in-place during the March 2009 flood in Fargo, North Dakota, and continue providing care, even though the hospital was in an area where officials had requested a complete evacuation. A key factor in this decision was the hospital's ability to remain operational for up to 10 days without city water, power, sewer, or other services—capabilities that had been intentionally designed when the hospital was constructed in 2000. In the end, the hospital was able to stay open throughout the incident.16

Sheltering-in-place was standard operating procedure at many of New Orleans' hospitals, and most did not consider preemptively evacuating prior to landfall of Hurricane Katrina.7,9,46 Administrators at Children's Hospital of New Orleans regularly updated the facility's adverse weather plan, "Code Gray," and coordinated with the State of Louisiana Emergency Operations Plan and the City of New Orleans Office of Emergency Preparations whenever there was advance warning of a category 3 or higher hurricane.46 CHNO had sufficient generators and fuel on site to maintain HVAC, and staff moved necessary equipment to the second floor in case of flooding.46 As Hurricane Katrina approached and strengthened, staff moved all patients to higher floors.46 CHNO evacuated following the subsequent flood, at great risk to its tiny and fragile patients. In the years since this event, CHNO administrators have taken steps to "harden" the hospital so that evacuation will never again be necessary. Investments in security, backup water sources, and other infrastructure should make sheltering-in-place a safe option during future hurricanes and floods.7

Deciding whether to preemptively evacuate or shelter-in-place requires consideration of two factors:

  • The nature of the event, including its expected arrival time, magnitude, area of impact, and duration; and
  • The anticipated effects on both the hospital and the community, given the nature of the event and the results of the Pre-Disaster Self-Assessment (go to Chapter 2).

Table 6 is intended to help decision teams facing this complex set of considerations. The first section of the table focuses on issues to consider and implications of different characteristics of the event. Decision teams will, of course, closely monitor impending disasters in order to gauge anticipated effects on the hospital and the surrounding area. Four generic disaster characteristics to be monitored include: arrival time, magnitude, geographic area affected, and duration. Perhaps more important than the estimate of these characteristics is the variability around that estimate and how likely the variability could potentially change. The most common example of variability is the width of the hurricane "cone" showing the projected path of the hurricane.

Local emergency management and other experts are the best source of information on event characteristics. At a minimum, hospital decision teams should educate themselves on disaster-specific characteristics, their variability, and what factors affect variability. For example, movement of wildfires is affected by three main factors: weather, fuel (e.g., ground material), and topography. In the case of river flooding, the areas that will be flooded at varying flood stages—in particular, the key roads to a hospital—should be documented and included in hospital evacuation plans.

The second part of Table 6 provides a framework for assessing the anticipated effects of the event on key resources needed to care for patients (water, heat, and electricity), the overall structural integrity of the building, and the surrounding community. The latter may include road conditions, community security, evacuation status of nearby health care facilities, the official evacuation orders, and the availability of local emergency response agencies. Specific questions are listed for each of these factors, the answers to which will highlight the risk of ordering a pre-event evacuation relative to the risk of sheltering-in-place.

Return to Contents

Evacuation Sequence in a Pre-Event Evacuation

If the decision is made to begin a pre-event evacuation, decision teams must also decide in what sequence to evacuate patients. While this is an operational issue, it is included in this Guide because decision teams that have considered this issue a part of evacuation planning and have trained staff according to the plan may be better prepared to carry out an evacuation.

Many hospital evacuation plans focus on keeping patients together with the staff who know them best, and evacuating entire floors or units together.18 This is more difficult if many hospital personnel are absent, as may happen when staff evacuate a city with their families in advance of a hurricane, flood, or fire.2,12,33 Safe transportation must be arranged for non-ambulatory patients, and it may not be possible to evacuate all ICU patients together with their staff when there are insufficient ambulances available, or when air evacuation is necessary.7,35 ICU and NICU patients will likely require more staff assistance and equipment during evacuation, and decision teams must decide whether to send out these precious resources with the first wave of evacuees (leaving fewer behind for the remaining patients) or wait to move the most resource-intensive patients last.9 Finally, in the hours before a hurricane, flood, or wildfire, decision teams must decide whether the most fragile patients are at more risk from an evacuation than from sheltering-in-place.

Every hospital has an evacuation plan, and although most acknowledge that specific circumstances may alter evacuation decisions, some plans/protocols do not acknowledge this need for flexibility.51 A one-size-fits-all evacuation plan may become obsolete in the midst of a disaster, especially in responding to a No Advanced Warning Event, such as an earthquake. Lessons learned from the experience of others in many different and challenging disaster scenarios will help decision teams adjust their plans to suit specific disaster circumstances.

After census reduction has occurred and a pre-event evacuation has been ordered prior to an Advanced Warning Event, decision teams must decide whether to evacuate their most medically unstable patients (e.g., those requiring powered life-support equipment) or keep these patients in the threatened hospital and hope that essential services will not be disrupted.7 The risks of moving medically unstable patients are high, and physicians and decision teams must weigh the risks of moving these patients against the risks of sheltering-in-place.7

The deliberate strategy of evacuating the most resource-intensive patients first in a pre-event evacuation emerged following Hurricane Katrina.9 The experience of waiting too long, and then being stranded in hospitals without water or power, was a powerful lesson for decision teams and staff. Those who went through this ordeal advise that it is preferable to preemptively evacuate resource-intensive patients, so as to avoid having to evacuate them in even more treacherous conditions after the storm hits.viii

This advice was followed by several decision teams a few weeks after Hurricane Katrina, as Hurricane Rita approached the Gulf Coast. At eight hospitals that were evacuated (partially or fully) prior to landfall of Hurricane Rita, decision teams chose to evacuate their most medically unstable patients first, as soon as appropriate transportation teams and equipment were available. For example, the University of Texas Medical Branch (UTMB) evacuated critically-ill patients first because this was deemed safer than having them shelter-in-place.11 Seven other hospitals within one medical system first evacuated the NICU and then ICU patients, as appropriate transportation became available.12

Staff and transportation shortages, or fear of them, may also prompt pre-event evacuations and affect the sequence of patient transfers.12 In the case of a community-wide evacuation order, hospital workers may be dispersed, leaving hospitals with insufficient staff to shelter-in-place or without enough able bodied people to assist during an evacuation.2 The evacuation process drastically reduces the number of staff available to stay in the hospital and care for patients, as some staff must join transport teams. Medically unstable patients are particularly resource-intensive and may need to be transferred with several care givers (to provide manual ventilation, monitor cardiac status, and provide other services in the absence of electricity) on specialized vehicles. Evacuating resource-intensive patients well before disaster strikes allows at least some opportunity for transport staff to return to the hospital to care for those sheltering-in-place or to evacuate additional patients. In addition, an early pre-emptive evacuation may allow time for more staff to arrive as replacements for the departing transport teams.9 This strategy also prevents patients from potentially being transferred in post-event conditions, such as a flood, which may preclude the use of ambulances and other specialized transport equipment.

Unlike other hospitals that triaged patients by transportation needs and acuity, Pomerado Hospital (California) evacuated all patients simultaneously in response to San Diego wildfires in 2007.13 In this case, appropriate transportation, including buses and ambulances, had been pre-staged near the facility. There was no shortage of evacuation teams or equipment, and therefore it was not necessary to triage patients as is often the case prior to hurricanes.50

viii Advice from the technical expert panel on hospital evacuations convened by Abt Associates on January 14, 2009 (go to Appendix A).

Return to Contents
Proceed to Next Section


The information on this page is archived and provided for reference purposes only.


AHRQ Advancing Excellence in Health Care