Introduction
Superstorm Sandy, which made landfall in October 2012, was the most destructive natural disaster ever to strike New York City (NYC; New York, USA), causing over US $19 billion in damage. 1 Sandy wreaked havoc on the health care infrastructure of NYC, causing massive floods that closed five acute care hospitals in Lower Manhattan.
The United States Department of Veterans Affairs (VA) New York Harbor Healthcare System’s (NYHHS) Manhattan campus and New York Downtown Hospital, two of the five hospitals in Lower Manhattan that suffered extensive flood damage, opted to evacuate in advance of the storm. The remaining three hospitals chose to shelter-in-place until forced to evacuate when flooding knocked out emergency power. In Sandy’s aftermath, authors of several commentaries questioned why hospitals within blocks of each other made very different decisions as the storm approached.Reference Powell, Hanfling and Gostin 2 , Reference Fink 3 Others found it alarming that administrators in hospitals with electrical switchgear, fuel sources, and pumps below ground level did not heed Hurricane Katrina’s (2005) lessonsReference Arendt and Hess 4 and anticipate a loss of generator power.Reference Evans, Carlson and Barr 5 , Reference Redlener and Reilly 6
Before Katrina, few hospitals considered evacuation in advance of a major storm, regardless of predicted intensity, and even fewer had evacuation plans in place.Reference Arendt and Hess 4 , Reference Gray and Hebert 7 In the past decade, an increasing frequency of weather disasters, particularly hurricanes and tornadoes,Reference Leaning and Guha-Sapir 8 have crippled hospitals and forced them to evacuate in the midst of disaster. Yet, a 2009 review of the research on hospital evacuation and sheltering concluded that there remained a “paucity of published, peer-reviewed data available on hospital evacuations.”Reference Bagaria, Heggie and Abrahams 9 Even less is understood about hospital-evacuation decision making.Reference Bagaria, Heggie and Abrahams 9 - Reference McGlown 13 Several case studies describe evacuations carried out after an event has compromised a hospital’s ability to provide care safely.Reference Schultz, Koenig and Auf der Heide 10 , Reference Childers and Taaffe 11 , Reference Sternberg, Lee and Huard 14 , Reference Taaffe, Kohl and Kimbler 15 In other studies, events that occurred suddenly (eg, earthquakes) or with little time to prepare (eg, tornadoes) caused dangerous conditions that led to swift evacuation decisions, often made by frontline staff.Reference Carlton and Bringle 16 , Reference Chavez and Binder 17 When buildings are on the verge of collapse, evacuation is the obvious choice, and sheltering-in-place may no longer be a viable option. Rarely, however, are hospital-evacuation decisions as straightforward in more protracted advance-warning events, the most common of which are hurricanes.Reference Sternberg, Lee and Huard 14 , Reference Taaffe, Kohl and Kimbler 15 , Reference Zane, Biddinger and Hassol 18
Deciding whether to evacuate a hospital or shelter-in-place before a hurricane is challenging, particularly due to uncertainty about the predicted storm and the risks and complex logistics of evacuating medically fragile patients.Reference Gray and Hebert 7 , Reference Bagaria, Heggie and Abrahams 9 , Reference Childers and Taaffe 11 , Reference Taaffe, Kohl and Kimbler 15 , Reference Chavez and Binder 17 Evacuation decisions are also time sensitive (ie, evacuating too early may put patients at risk unnecessarily, while waiting too long can result in evacuating under less-than-ideal conditions).Reference Childers and Taaffe 11 Yet, few studies have explored hospital-evacuation decision making in advance of a major storm. Downey and colleaguesReference Downey, Andress and Schultz 12 , Reference Downey, Andress and Schultz 19 conducted a retrospective study of seven hospitals impacted by Hurricane Rita (2005). Before landfall, three of the seven hospitals fully evacuated and four partially evacuated. They found that hospital-evacuation decisions were influenced most heavily by: predictions of the storm’s intensity, location, and timing; issuance of mandatory-evacuation orders; and loss of regional communications. The authors did not differentiate between evacuation decisions made before Hurricane Rita and those made after landfall. Thus, it is unclear which factors had the most influence on evacuation decision making before landfall, when uncertainty about the hurricane’s impact was greatest.
Since Katrina, several hospital-evacuation planning tools have been developed by state and local public health agencies, 20 - 22 the US Department of Health and Human Services (Rockville, Maryland USA),Reference Zane, Biddinger and Hassol 18 as well as private sector entities, such as the California Hospital Association (Sacramento, California USA). 23 These documents contain planning tools such as checklists for gathering essential information and guidance regarding factors that hospital leaders should consider when deciding whether to preemptively evacuate or shelter-in-place. However, these documents provide little insight into the decision-making process; that is, how information and expertise is shared and acted upon. Moreover, challenges common in crisis decision making, such as balancing the needs of stakeholder groups and managing potential liability concerns, financial issues, and political pressures,Reference Parker, Nelson and Shelton 24 are rarely discussed.
These gaps in the literature on hospital-evacuation decision making in advance-warning events were the impetus for the current study. The VA NYHHS’s evacuation of its Manhattan tertiary care medical center in advance of Hurricane Irene (2011) and prior to Superstorm Sandy provided a rare opportunity to examine this issue in depth. The VA is the largest integrated health care system in the US, with 152 VA medical centers (VAMCs) throughout the country. This report focuses on NYHHS’s evacuation decision-making process and the factors that influenced decisions for both events from the perspective of senior leaders.
Methods
Semi-structured interviews were conducted with staff who participated in NYHHS’s response to either or both events. Potential participants from NYHHS and from the VA New York/New Jersey (USA) Veterans Integrated Service Network (VISN) 3 were identified through existing contacts in VA at local, regional, and national levels, and through VA documents (eg, After Action Reports and NYHHS online directory). Stratified, purposive sampling techniques were used to select participants who had filled key roles in NYHHS’s response to Irene and Sandy. This report is based on data from 11 interviews with senior leaders (executive managers and clinicians in senior administrative roles) to describe the evacuation decision-making process and the factors that influenced decisions for both events.
The interview guide was developed based on an instrument from a prior study examining VA nursing home evacuations after Hurricanes Katrina and Rita,Reference Claver, Dobalian and Fickel 25 a hospital-evacuation tool published by Schultz et al,Reference Schultz, Koenig and Auf der Heide 10 and the hospital-evacuation literature.Reference Gray and Hebert 7 , Reference Bagaria, Heggie and Abrahams 9 , Reference Sternberg, Lee and Huard 14 , Reference Chavez and Binder 17 Open-ended questions were constructed to elicit responses in a conversational style and to encourage participants to raise issues that the researchers may not have considered. Topic areas of the interview guide included prior disaster-response experience, participation in disaster-preparedness planning and exercises, preparations and planning in the days leading up to each event, the evacuation decision process, and operations and logistics of each evacuation. Participants were also asked about successes, challenges, and lessons learned. Interviews lasted 60-90 minutes each and were conducted by at least two members of the research team. All interviews were audio recorded and transcribed. This study was approved by the VA Greater Los Angeles Healthcare System’s Institutional Review Board (Los Angeles, California USA).
The Atlas.ti software program (Version 7.1.6, Scientific Software Development GmbH; Berlin, Germany) was used to manage the analysis of the transcripts. The first author developed a list of a priori codes based on the interview guide and the hospital-evacuation literature. The first two authors independently coded each transcript. When there was disagreement, the first two authors reconciled differences through discussion, sometimes with other members of the research team. A qualitative specialist also reviewed the codes and several coded transcripts for consistency. As themes emerged from the data, new codes were added, and codes were grouped by themes that were evident across the interviews.
Results
The interviews indicated that NYHHS’s evacuation decision-making process was, in many ways, similar for Hurricane Irene (August 2011) and for Superstorm Sandy (October 2012). Participants described a process of building situational awareness that included leadership from VISN 3, NYHHS, and facility directors throughout the VISN. In the days before both events, conference calls were held to discuss the latest storm predictions, the vulnerabilities of each facility to the storms’ predicted impact, and preparations to protect buildings and ensure the safety of patients and staff. During both storms, facilities throughout VISN 3 were at risk for power outages and wind and water damage, but NYHHS’s Manhattan facility was most at risk of flooding due to its location in Zone A, a low-lying area of NYC.
Confidence in Facility Leadership
Decision making in both cases was collaborative, and VISN leadership was described as supportive and involved in the decision-making process. Ultimately, though, the evacuation decision in both events rested with the facility director, because, as one participant asserted, “…medical center directors…know their facility better than anybody else, and what [the facility can] withstand, [or] not withstand in case of a natural disaster.” The VISN leaders described having “confidence in the leadership that’s on the ground” to make the final decision, and stated that once the decision to evacuate was made, the VISN’s role was to provide resources and logistical support to facilitate the evacuation.
Vulnerability Assessment
The interviews revealed that a few years before Irene, senior leaders in New York and VA officials in Washington, DC (USA) were aware of the Manhattan facility’s vulnerability to flooding and power loss. As Hurricane Irene approached in August 2011, plans were underway to construct a seawall to protect the facility against flooding, but there were no plans to move the hospital’s utilities and other critical infrastructure, which were located on the basement floor of the building.
Hurricane Irene
Risk Assessment
Hurricane Irene struck NYC on Saturday, August 27, 2011. Participants reported that in the days before Irene, there was significant media coverage of the storm’s predicted intensity and speculation that Mayor Bloomberg would issue a mandatory-evacuation order for hospitals in Zone A. On conference calls held on Thursday, August 25, NYHHS and VISN 3 leaders discussed whether they should evacuate the Manhattan facility before the Mayor issued a mandate. Participants reported that leaders weighed the risks of evacuation against the potential for flooding in the Manhattan facility, which if severe enough, could submerge utilities on the basement level.
Senior leaders who were associated with the Manhattan facility for more than 20 years had valuable experience with the types of storms that had caused flooding previously. The most notable was an unnamed Nor’easter in December 1992 that struck at high tide during a full moon. The facility’s lower level flooded quickly, leaving the building without power. The timing of Irene, however, was neither full moon nor high tide. Interviews revealed that leaders were not convinced of the need to evacuate before Irene, given their knowledge of what the facility could withstand and Irene’s predicted timing and intensity. According to one participant, “we thought that we could handle the water. We thought we were okay with that.” Another stated, “we were a little bit late in making that decision [to evacuate]… I’m not sure we agreed with… [the Mayor’s evacuation mandate].” There was also concern about the risks of evacuation to patients, particularly since leadership had no prior evacuation experience. One participant stated, “…the last thing you do is evacuate [a hospital] because you’re putting all these patients at risk, and at Manhattan, we have the sickest patients in the VISN.” Another senior leader echoed a similar concern, “…not having ever experienced an evacuation, that’s pretty dramatic… it’s a serious business to do it. Quite frankly, pulling the trigger… it’s a big deal, it is a very big deal.”
City’s Evacuation Mandate Tips the Balance
On Thursday evening, two days before Irene’s predicted landfall in NYC, Mayor Bloomberg notified hospitals in Zone A that they must evacuate by the following evening. One participant noted that the Manhattan VAMC was exempt from the evacuation order: “…technically, as a federal facility, we do not have to follow what the mayor says...” However, according to the same participant, the Mayor’s order put a “different tone” on the discussion. Comments from several others suggested that the Mayor’s mandate was the most influential factor in the decision to evacuate the Manhattan Campus in advance of Irene. According to one leader, “I don’t think we really had any choice…because the Mayor…wanted everybody in that zone to evacuate.” Another participant stated, “…we were marching to the city’s timeline…because the mayor said, you will evacuate and then essentially, we had no choice. Well, we did have a choice, but…who is going to take that risk?”
Hurricane Irene was a “Blip”
Inpatients at the Manhattan VAMC were fully evacuated by Friday evening, August 26, and the hospital was closed, as were other hospitals under the evacuation mandate. However, Irene’s impact on NYC was not as severe as had been expected, and by the following Tuesday, the Manhattan VAMC was fully reopened, having suffered little damage from the storm. As one leader described, “…Irene was a blip, we got the patients out…we had no impact on our facility, so the only thing we needed to do was bring the patients back.” Although the evacuation for Irene ultimately proved unnecessary, several participants claimed it was still the correct decision. One participant stated, “…with Irene we didn’t lose anything. We evacuated for really nothing. But I still think that was the right decision.”
Superstorm Sandy
Complacency as Sandy Approached
In October 2012, 14 months after Irene, another major storm was predicted for the New York area. Several participants felt that Irene’s less-than-predicted intensity contributed to a sense of complacency across NYC as Sandy approached. One participant stated, “I remember watching [the Mayor] on television…saying, ‘If you can see water from where you live, get out.’ You couldn’t have said it any blunter. But that false sense of security, ah, here we go again… [Sandy] is not going to be a problem.” Another participant attributed some of the complacency to the media,
…the media…loved Irene…they loved hyping it up. It was another story with Sandy. Irene did not play out the way…the Mayor’s office and media expected it to. And because of that, I think a lot of people thought that Sandy was going to be the same thing…you know, “what was the big deal”…
The same participant reported that there was scant media coverage of the impending storm in the days before Sandy. In fact, numerous attempts to contact the media and get the word out to Veterans about the evacuation and closure of the Manhattan VAMC before Sandy went unanswered.
Prior Evacuation Increased Confidence
As Sandy advanced, leaders from NYHHS and VISN 3 were again faced with a decision to evacuate the Manhattan facility before the storm or to shelter-in-place. According to participants, the decision-making process was essentially the same for Sandy as for Irene, and included the same group of leaders. However, several participants noted that as they deliberated about options before Sandy, there was an increased level of confidence that an evacuation could be accomplished safely, given the successful, albeit arduous, evacuation before Irene the year before. One participant expressed thanks that Irene had given them an opportunity to prepare for Sandy, “…thank God for …Irene because it got us ready for Sandy. [Irene] was a blessing because we knew the pitfalls of evacuation.”
Corporate Memory Drove the Decision
Unlike Irene, Sandy was predicted to strike NYC during a full moon, and the storm’s peak was expected at high tide, characteristics strikingly similar to a December 1992 Nor’easter that crippled the Manhattan facility. One participant described the impact that the 1992 storm had on the hospital: “…it wasn’t quite as bad as [Sandy], they did lose power but they…stayed in place, but it was brutal.” Another shared, “I had not been here [in] ‘92 …where cars were floating in the backyard [of the Manhattan facility], but I had heard the stories and I was in New York at the time so…I believed.”
A key NYHHS decision maker was a staff assistant at the Manhattan facility in 1992 and experienced first-hand the impact of the storm and its consequences:
…very few people remember [the 1992 storm] because it wasn’t what people would consider a hard-hit storm. It was a very rain-driven storm during a full moon, high tide. And that’s what does [the facility] in. It happened in an instant…we were completely shut down with seven or eight inches of water in our basement… [the flooding] had shut down all of the utilities, so we had no electric, no elevators, no phones, nothing, and we had a houseful of patients.
Empowered as the final decision maker as Sandy approached, this leader shared thoughts about the impact that the 1992 experience had on the decision to evacuate the Manhattan facility before Sandy:
It points out in leadership…you can have all the training and take all the courses you want, but it’s…what experiences have you had and how does your judgment play into those experiences? And it’s putting it all together that gets you to the point to say, okay, this is what I think the right thing is to do and then moving on it. And I have to say I didn’t…second guess and [think] what is so-and-so going to think… My concern truly was I didn’t want to live through ’92 again… I knew what could happen in that facility, and we’re just getting out.
Interviews revealed that the key decision maker’s corporate memory, that is, historical knowledge about the 1992 storm and its impact on the facility, “…drove [the] decision to evacuate Manhattan.” Moreover, this decision maker conceded that had it not been for knowledge of the earlier storm that was similar to Sandy, then the unnecessary evacuation for Irene might have led to a different decision:
I probably would have had that false sense of security from Irene…who knows, but …I may have, especially knowing that neither NYU nor Bellevue, which are literally up the street from us...were not evacuating.
Another leader shared:
…if there were another [key decision-maker] that hadn’t lived through that [1992 flooding], maybe they would not have [evacuated]. If there was no history about how bad it could be, that it did flood once.
The same participant recognized the value of the key decision maker’s “corporate memory” of the 1992 storm, and the importance of preserving such information:
We probably should have codified that some place. What if we were all gone? Where’s the corporate memory to say, you know what, if you have one of these, that decision needs to be made, even if it’s a wrong decision, at least the patients will be safe.
Later in the interview, the same participant acknowledged that VA does not have a mechanism for sharing lessons learned with others in VA, “we don’t do well telling stories. I mean, Katrina happened and I don’t know that I can remember a full-out briefing.”
Discussion
The preemptive evacuations of the Manhattan VAMC in advance of Hurricane Irene and of Superstorm Sandy provided a rare opportunity to examine hospital-evacuation decision making at the same facility twice within a span of 14 months. This study focused on factors that most heavily influenced each decision. Results suggest that factors rarely addressed in the literature (corporate memory and the political environment) play a significant role in decision making.
A key finding from this study is that corporate memory (ie, historical knowledge about the facility’s vulnerability to a storm with similar characteristics) was critical in the decision to evacuate the Manhattan VAMC before Sandy. This finding is an important contribution to the hospital-evacuation literature, which is replete with case studies that do not address the role that corporate memory of past disasters plays in evacuation decision making. As Sandy approached, NYHHS’s key decision maker was, incidentally, also the individual with corporate memory of the 1992 storm and its impact on the Manhattan facility. Although it was fortunate that this critical information was known to a key member of the decision team, study participants recognized that this important knowledge was held by one individual and could have easily been lost had that individual left NYHHS. Yet, there is no systematic mechanism for preserving corporate memory at the facility level, across VA, or more broadly to organizations outside of VA.
As Sandy demonstrated, the current approach to collecting, documenting, and sharing lessons learned from past disasters has been ineffective, and the degree to which hospitals share critical information about pending threats is unclear. This raises important policy questions about how current approaches to collecting, documenting, and sharing critical information and lessons learned from past disasters could be improved. Although the health care market is highly competitive, greater shared situational awareness among at-risk hospitals, along with a more coordinated approach to evacuation decision making, could reduce pressure on hospitals to make these high-stakes decisions in isolation. Moreover, a mechanism for systematically capturing and recording hospital-evacuation data is sorely needed.Reference Gray and Hebert 7 , Reference Bagaria, Heggie and Abrahams 9 , Reference Schultz, Koenig and Auf der Heide 10 , Reference Downey, Andress and Schultz 12 , Reference Downey, Andress and Schultz 19 A shared repository of knowledge and guidelines is one possible way to address this need.
Findings also suggest that in advance-warning events, hospital-evacuation or shelter-in-place decisions are complex decisions that are influenced by external factors such as the political environment, an issue rarely addressed in the scientific literature.Reference Schultz, Koenig and Auf der Heide 10 , Reference Parker, Nelson and Shelton 24 As Hurricane Irene advanced toward NYC, then Mayor Michael Bloomberg ordered a mandatory evacuation of hospitals in low-lying areas of the city. As a federal facility, NYHHS’s Manhattan Campus was exempt from the city’s evacuation mandate. Yet, the Mayor’s order was the most influential factor in NYHHS’s decision to evacuate, essentially overriding senior leaders’ assessment that the facility could withstand Irene’s impact.
Results also illustrate that media coverage of impending storms and their aftermaths contribute to the “high-stakes” nature of hospital-evacuation decisions. As Hurricane Irene approached, participants reported that there was extensive media coverage of the impending storm, particularly once the Mayor issued an unprecedented mandatory-evacuation order for hospitals in Zone A. Irene turned out to be less damaging than expected, which several participants claimed contributed to a sense of complacency across the city in the days leading up to Sandy. Unlike Irene, media coverage of the impending storm was minimal, but rapidly expanded as Sandy’s massive storm surge caused extensive flooding and widespread power outages. Failed generators forced major hospitals to evacuate patients down darkened stairwells, sparking extensive media coverage and close scrutiny of hospitals that had chosen not to evacuate in advance.
Finally, this study illustrates the importance of comprehensive evacuation drills and exercises. Although, in hindsight, the Manhattan evacuation for Irene was unnecessary, it proved to be a valuable “practice run” for NYHHS. As Sandy approached, leaders were confident that the Manhattan facility could be safely evacuated before the storm, whereas before Irene, leaders hesitated to evacuate, having never done it before. Eliminating uncertainty about the capability to evacuate safely within a certain period of time could help hospital leaders make more informed decisions. Yet, few hospitals without disaster-evacuation experience have moved beyond written evacuation plans and tabletop exercises to demonstrate that capability. These and other concerns have led the Center for Medicare and Medicaid Services (Baltimore, Maryland USA) to propose national emergency preparedness requirements for Medicare- and Medicaid-participating providers. 26 Under the proposed rule, hospitals would be required to plan adequately for both natural and man-made disasters so as to ensure that they are able to meet the needs of patients during emergency situations.
Limitations
Given the longstanding presence of VA providers in disaster-prone areas of the country, VA has extensive experience in responding to disasters, making it an ideal “laboratory” for hospital preparedness and response research. Since 1971, three VAMCs have been destroyed by earthquakes and two were ruined beyond repair by Hurricane Katrina. Nonetheless, this study has limitations. Data reflect experience from a single facility, which limits generalizability.
However, the results were about two preemptive evacuation decisions for that facility, which provides a unique addition to the literature. Most of the decision-making factors identified do not appear to be unique to VA based on the prior literature. Nevertheless, as a large, federally-funded, integrated health care system, VA has greater acute care and long-term care surge capacity than does much of the private sector. As such, when deciding whether to evacuate a VA facility or shelter-in-place, locating receiving beds for evacuating patients may not be as challenging as it is for some private sector hospitals. The main findings, though, should be applicable to other hospitals in the private and governmental sectors. In particular, the finding that corporate memory of past disasters plays a role in evacuation decision making would seem to be widely applicable and not limited to VA. Finally, results are based on a limited number of interviews. However, these 11 participants included all of the senior leaders who were involved in the evacuation decision making.
Conclusions
Deciding whether to evacuate a hospital or shelter-in-place in advance of a predicted event is a complex, time-sensitive decision that is fraught with uncertainty and the potential for significant adverse outcomes. Moreover, potential legal liability issues, financial concerns, and political considerations can further complicate these decisions.
This study helps inform the current policy debate surrounding hospital-evacuation decision making. Ideally, hospitals should be capable of withstanding the impacts from extreme events, such as Superstorm Sandy, and safely shelter-in-place. Yet, given the aging health care infrastructure, it will take many years and substantial financial commitments to approach the ideal. In the meantime, increasing the knowledge base on hospital-evacuation decision making takes on a new urgency.
