1. Introduction
Early warning systems, evacuations, and sheltering programs serve as a protective action for those at risk to various threats in coastal communities in the island territories, including high winds, storm surge inundation, rainfall flooding, and landslides. The extremely high levels of hurricane activity in 2020 and expected high levels of activity in 2021 (Klotzbach et al. 2021) were compounded by the coronavirus disease 2019 (COVID-19). Planning and coordination among emergency management, public health, and human services experts in Puerto Rico (PR) and the U.S. Virgin Islands (USVI) are needed to address the compounding threats during the pandemic. Threats are not only to individuals and families but to the health and social systems as a whole that are dedicated to protecting population health and safety.
Ensuring priorities and activities are realistic, well planned, and clearly communicated requires a whole community approach (Federal Emergency Management Agency 2011). Engaging local residents in the planning process provides jurisdictional government entities with the information needed to strengthen existing plans if confronted with a major hurricane and tailor these protective actions to meet community needs.
Formative research conducted among populations at risk provides data needed to identify barriers to accessing a safe space, establish interventions to reduce morbidity and mortality, and ensure public messaging and education positively influence decision-making processes. This study takes a bottom-up approach that places emphasis on local government and community actions for disaster risk reduction (Burger and Gochfeld 2020). Baseline knowledge of risk perception of compounding threats during a pandemic, the first of its kind in the Puerto Rico–U.S. Virgin Islands (PRVI) region, aims to support evidence-informed decisions and capacity building for public health preparedness planning. Slovic (1994) points out that perceptions are so often at odds with what experts say people should be concerned about and emphasizes that we must learn to treat perceptions as legitimate. Understanding these risk perceptions related to COVID-19 and hurricane evacuations will help inform policy. As Phillips et al. (2020) note, pandemics are likely to occur more often in our globalized economy. Therefore, such studies are needed for further planning that takes into consideration the intersectionality of social constructs and the compounding threats of natural hazards affecting PRVI (Chaplin et al. 2019).
The purpose of the research conducted in the PRVI region, at the beginning of the 2020 hurricane season prior to vaccine availability, is to further understand public perceptions of the compounding risks posed by hurricanes (and related hazards of landslides and storm surge) and a pandemic, and to examine the extent to which people risk their lives by sheltering in place rather than evacuating. The research questions addressed in this study include the following:
-
How does peoples’ risk perception of existing hazards associated with hurricanes and preparedness planning differ by gender, location, type of housing, political affiliation, economic earnings, level of education, transportation options, and access to a generator during the pandemic?
-
What percentage of people plan to evacuate due to a hurricane event during the COVID-19 pandemic?
-
What factors influence their evacuation behavior, and will they instead choose to shelter in place despite a recommendation to evacuate to a safer location or a government-operated shelter staffed by local authorities?
-
What are the perceived barriers and facilitators that trigger action decision-making under the threat of a hurricane during the COVID-19 pandemic?
-
Are individuals likely to utilize public shelters? Do people view public shelters as excessively risky due to concerns about the COVID-19 pandemic?
-
What are the most relied-on sources of information during a hurricane to reach residents in the PRVI region?
Prior research has not been conducted in the PRVI region into an individual’s risk perception of natural hazards while considering evacuating from a hurricane during a pandemic. Officials need to understand how evacuation plans change with COVID-19-encouraged social and physical distancing, which directly conflicts with the movement and congregation seen in hurricane evacuations. The results can be compared between studies of different islands and existing studies on the mainland to examine which policies and practices are best suited for each location.
2. Literature review
a. Tropical cyclones in the PRVI region
The PRVI region, situated in the Caribbean Sea, is subject to severe and frequent impacts from hurricanes. Our current knowledge of hurricane hazards focuses on landfalling events in the U.S. mainland, with less work on the lower latitudes (Elsner et al. 1999, 2000; Jury et al. 2012). More than 15% of all Atlantic hurricanes make landfall in the Caribbean Antilles Islands between August and September (Jury et al. 2019). It is expected that island territories will experience a substantial impact of flooding due to direct exposure to tropical storms and a strong dependency on coastal resources (Vitousek et al. 2017). Destructive impacts from hurricanes can lead to severe infrastructure loss, and environmental impacts such as power outages, damaged infrastructure, saltwater intrusion, and coastal erosion are unavoidable on small islands when exposed to high winds, flash floods, and landslides (Cox et al. 2019). Furthermore, the human impacts from a hurricane event can be severe in the PRVI region, as seen in Hurricane Maria, which is why hurricane evacuation planning in this region is of utmost importance. The PRVI region is ideal for studying severe hurricane hazards and intended evacuation behavior as the area often experiences storm surges, intense rain, and subsequent flooding from hurricane events with potential landslide risks and earthquakes occurring after a hurricane due to the mountainous terrain (Wang et al. 2014), hazards that may push residents to evacuate their homes. In addition to this, the COVID-19 pandemic brings additional concerns to the confluence of these hazards and implications for evacuation behavior.
b. Hurricane evacuation orders during a pandemic
While hurricane evacuation behavior and risk perception research have grown throughout the years, historically, research identified a correlation between evacuation rates and low-lying areas at high-risk along with factors such as housing and severity of the storm (Baker 1991, 1995; Gladwin et al. 2001; Whitehead et al. 2000). Additional studies have also explored how past experiences have influenced evacuation intentions and how individual assessments were vital in the decision-making process (Demuth et al. 2012; Dow and Cutter 2000). Over the years, there has been a heightened focus on the role of forecasting and risk communication to identify more effective channels of information sharing and synthesis for decision-making (Collins et al. 2017, 2018; Morss et al. 2016; Senkbeil et al. 2019; Sherman-Morris et al. 2011). Research on socioeconomic characteristics and social capital in correlation with evacuation behavior as a means of assessing social (in)equity for evacuation planning has provided a deeper understanding of vulnerability (Eisenman et al. 2007; Elder et al. 2007; Miller 2007; Moore et al. 2004). In recent years, emerging discussions centered around intersectionality highlight how social characteristics—such as gender, ethnicity, disabilities, age, people with chronic health conditions, and socioeconomic status—intersect, providing a more in-depth understanding of an individual’s priorities, needs, capacities, and overall experiences preparing for, coping with, and responding to natural hazards (Alvarez and Evans 2021; Chaplin et al. 2019; Chisty et al. 2021; Kuran et al. 2020; Versey 2021; Walker et al. 2019). Understanding social structures from historical, cultural, political, and economic contexts across diverse social identities delineates the root causes of social vulnerabilities and how inherent inequalities are exacerbated during compounding natural hazards when it comes to supporting evacuation decision-making and behaviors (Borowski and Stathopoulos 2020; Chaplin et al. 2019; Walker et al. 2019).
Minimal research on evacuation behavior has been conducted in the PRVI islands (Morrow and Gladwin 2014a,b; Shultz et al. 2019; Puerto Rico Department of Health 2020) with little understanding of how compounding risks will impact evacuation decisions and subsequent health and survival outcomes. Understandably, much of the U.S. population, including the U.S. territories, have no prior experience planning for hurricane evacuations during a pandemic, and public health planners need to rapidly assess the potential societal impacts to anticipate various worse case scenarios (Buckle 2006; Demuth et al. 2012). Therefore, this research provides insights into what could be expected about evacuation behavior for those in the PRVI region. Historically, government officials have opened shelters in advance of hurricane events in both Puerto Rico and the USVI. During Hurricanes Maria and Irma, there were a reported 10 692 occupants across 175 shelters in Puerto Rico and 558 occupants housed in 9 shelters in the USVI (Pan American Health Organization 2017). This does not account for the estimated population exodus from the islands before and after hurricane landfall, with an estimated loss of 475 779 from Puerto Rico (Acosta et al. 2020).
COVID-19 is the fifth pandemic to affect the world in the past 100 years [Centers for Disease Control and Prevention (CDC) 2017]. Since 1 December 2019, COVID-19 has spread globally with over 216 million cases and more than five million deaths worldwide as of 30 August 2021 (World Health Organization 2021). While guidance from hurricane experts reminds residents that they should prepare the same for every season, regardless of how much activity is predicted, preparing for hurricane seasons during the global pandemic is causing individuals to rethink their evacuation plans. Should a major hurricane occur in the PRVI region during the COVID-19 pandemic, residents will be making complex decisions as they balance their need to evacuate and the risk of infection. COVID-19 poses a unique threat to transportation and sheltering considerations during hurricane evacuations (Collins et al. 2021; Hill et al. 2021; Shultz et al. 2020a). The pandemic increases the complexity of planning for hurricanes and the related hazards (landslides/flooding) as social distancing is in direct conflict with human mobility and congregation in a disaster. Disease surveillance and addressing the needs of clients in the general population and special medical needs shelters pose additional challenges to public health preparedness and emergency management authorities in both U.S. territories (Schnall et al. 2019). In the USVI, there are also limited evacuation shelter options, especially when considering the demands of spaces conducive to social distancing.
Previous studies have shown that without the COVID-19 pandemic, natural disasters have already resulted in further spreading of infectious diseases attributed to the crowding of people, including evacuation shelters (Ivers and Ryan 2006; Lemonick 2011, Shukla et al. 2018). Given that social distancing is warranted with COVID-19, people may risk sheltering in place to avoid COVID-19 exposure despite being in an area vulnerable to storm surge and landslides. Shultz et al. (2020a) suggest that fear of contracting COVID-19 would prevent residents living in mandatory evacuation zones from leaving their homes for safety. Collins et al. (2021) found that approximately 74.3% of individuals viewed the risk of being in a shelter during the COVID-19 pandemic as more dangerous than enduring hurricane hazards when Florida residents were surveyed about their intended hurricane evacuation behavior and perceived risk in June 2020, the first hurricane season during the COVID-19 pandemic. Likewise, Botzen et al. (2022) noted that perceptions of COVID-19 risks at the start of the 2020 hurricane season exceeded those of flood risks. For example, they note that more people are “worried” or “strongly worried” about COVID-19 risks (63%) than about flood risks (33%). Shultz et al. (2020b) emphasize that messaging must overcome fear-dominated and paralyzing COVID-19 inertia that might prompt many to remain in vulnerable households in the path of oncoming storms. Shultz et al. (2020c, p. e507) also discuss the 2020 season and specific storm impacts noting that “populations in the path modified and contrived evacuation and sheltering protocols to minimize COVID-19 spread, tailored to the specific hazards encountered.” Phillips et al. (2020) note that the COVID-19 pandemic will likely combine with existing hazards to exacerbate mortality and morbidity.
As Blaikie et al. (2005) contend, social vulnerability and the various forms of risk are the root cause of disasters. They note that vulnerability manifests from numerous social factors, primarily economic imbalances, disparities in power among social groups, knowledge dissemination, and discrimination in welfare and social protection. Vulnerable populations such as racial minorities, those of low socioeconomic status, the homeless, and older adults will likely suffer more severe effects of COVID-19 due to compounding and systemic issues, such as preexisting disease, system barriers in health care systems; of note is that people of color have had an undue burden in the COVID-19 pandemic with more hospitalization and deaths than their white counterparts (Abuelgasim et al. 2020; CDC 2020). These groups also suffer from overlapping hurricane vulnerability caused by many factors, including a lack of financial means to properly prepare for a storm, potentially weak social support networks, limited mobility, and no access to reliable means of transportation (Aldrich and Benson 2008; Blaikie et al. 2005; Brown et al. 2012; Donner and Lavariega-Montforti 2018; Elder et al. 2007). The concept of social vulnerability during disasters is especially relevant in the PRVI region as it suffers from a lower socioeconomic status than the continental United States. In the USVI, approximately one-quarter of individuals live in poverty; in Puerto Rico, it is estimated that almost one-half of the population lives in poverty (Kaiser Family Foundation 2017; U.S. Census Bureau 2019). For comparison, the average poverty rate nationally in the United States is 10.5% as of 2019 (Semega et al. 2020).
Whytlaw et al. (2021) also note, when considering hurricane evacuation and sheltering during the pandemic, that vulnerabilities associated with underlying health conditions and socioeconomic disparities were of increased concern to evacuation and shelter planners. Public hurricane shelters tend to be utilized by the most vulnerable, those who lack the economic means for other evacuation options, and populations with underlying health conditions that place them at higher risk for pandemic complications. One of the most vulnerable populations in evacuating for weather-related disasters are older adults (Brunkard et al. 2013). In addition to the decreasing physical and cognitive abilities that are characteristic of aging, around 85% of older adults have chronic health conditions that make evacuating more complicated (National Institute on Aging 2017). Older adults may be more vulnerable to both the risks of staying at home during a hurricane as well as becoming very ill due to COVID-19 (Meng et al. 2020). Various studies support the conclusion that racial and ethnic minorities are at greater risk during pandemics (Kirby 2020; Tai et al. 2021). In many cases, they often have less capacity to implement preparedness strategies or tolerate its impact given disparities in underlying health status and social factors, such as socioeconomic disadvantages; cultural, educational, and linguistic barriers; and lack of access to health care and transportation (Hutchins et al. 2009).
3. Method
a. Study site description
The PRVI region is ideal for studying the hazard of a hurricane (with its related impacts of storm surges, intense rain, flooding, and landslides) and the vulnerability of the people due to geographic and topographic characteristics of the islands. In addition to this, the COVID-19 pandemic brings additional concerns to the confluence of these hazards and vulnerability. During the time of this study’s survey distribution (October 2020–February 2021), there were unprecedented COVID-19 infection rates within Puerto Rico as many cases were consistently reported, peaking at the end of November with over 1000 new cases being reported daily and a 7-day average of 847 cases (Puerto Rico Department of Health 2022; New York Times 2022a); the U.S. Virgin Islands reported a large wave of new cases in August, with a daily average of 34 cases per day preceding surveying, followed by another peak in cases mid-December of 25 cases per day (New York Times 2022b). The study sample represented over 120 zip codes (of 194 in this region) in total spread across the three primary islands in the USVI (Saint Croix, Saint John, and Saint Thomas) and 71 (of the 78 total) municipalities in PR (Fig. 1).
Distribution of respondents by zip code. The zip codes with the highest number of respondents in the USVI were 00802 (Saint Thomas) with 37 respondents and 00820 (Saint Croix) with 27 respondents. Puerto Rico had the highest concentrations in the zip codes of San Juan and Mayagüez, respectively.
Citation: Weather, Climate, and Society 14, 2; 10.1175/WCAS-D-21-0134.1
b. Data collection procedures
1) Sample selection and recruitment
All residents in PR and the USVI were included in the target sample criteria as the whole region is affected by hurricanes and may need to evacuate while being further exposed to COVID-19. Participants were excluded if they were under the age of 18 or 21 (in USVI and PR, respectively) or did not reside in the PRVI region. Barring these measures, no other exclusionary or inclusionary criteria were utilized. Subjects in this study were not compensated in any way for their participation.
The samples were obtained through convenience sampling utilizing an extensive network of personal and professional connections in PR and the USVI. Norris (2006) has noted the prevalent use of this method in her analysis of 225 disaster studies. The authors on this project, in addition to their extensive network, represented agencies such as the Puerto Rico Science, Technology, and Research Trust, the Virgin Islands Territorial Emergency Management Agency (VITEMA), Puerto Rico Emergency Management Bureau (PREMB), University of the Virgin Islands, University of Puerto Rico, the Caribbean Exploratory Research Center, and the Virgin Islands Department of Health, among others. Access to the large geographical study site and ultimately to the residents was gained by utilizing our extensive network with varied research and professional connections within the territories, which allowed for further dissemination of the digital link to their networks. Specifically, the research team contacted those who could act as partners and distribute to communities, such as emergency managers, professors, community leaders, faith-based leaders, government officials, and more. Connections were made by these professional contacts to distribute to trusted community members, local news stations, social media, and others within their network, who would then further distribute, to catch a wide variety of respondents; allowing access to many areas that were inaccessible to the core research team largely due to the limits of the COVID-19 pandemic. Our partners were provided with our outreach materials (twitter phrasing, graphics, and the survey links in Creole, Spanish, and English), and then they hosted this information on their website, distributed it to their networks via email, and/or printed out flyers for use at any in-person events.
After initial data were collected, target areas on the western side of PR and Saint Croix in the USVI were identified as a result of low initial response rate and higher social vulnerability (see Fig. 2). The CDC/Agency for Toxic Substances and Disease Registry (ATSDR) social vulnerability index (SVI) was employed to indicate which areas were of higher social vulnerability. The SVI is a widely utilized tool to help summarize many facets of social vulnerability in a community using 15 different census variables (ATSDR 2021). The factors that go into determining social vulnerability include those relating to four overall themes: 1) socioeconomic status, 2) household composition, 3) race/ethnicity/language, and finally 4) housing and transportation. Figure 2 was produced using the composite SVI score (one that takes all four themes into account) at the census tract level and then running a Getis-Ord hot-spot analysis that highlights areas of high and low social vulnerability (designated with red and blue, respectively) within Puerto Rico. Due to the high concentration of vulnerable census tracts (shown in red) on the west coast (and the relatively low social vulnerability on the east coast, shown in blue), special focus was given to ensure respondents from this highly vulnerable area were represented. Unfortunately, there are no publicly available SVI data for the USVI at the moment, an acknowledged weakness to this targeting method. Within USVI, the number of respondents on each major island was observed partway through data collection, and those areas with relatively underrepresented responses were chosen for targeting.
Hot-spot analysis of the social vulnerability index (CDC/ATSDR/Geospatial Research, Analysis, and Services Program 2018) for Puerto Rico, used in survey sampling.
Citation: Weather, Climate, and Society 14, 2; 10.1175/WCAS-D-21-0134.1
As a result, further targeted Facebook advertisements (in both Spanish and English) were placed to ensure more exposure in these identified regions. For recruitment via a social media approach, approximately 17 000 people viewed Facebook advertisements that targeted geographic areas needing an increased response (see Fig. 3). As a result of these targeted advertisements, there was an increased digital response to the survey, especially in the areas that were prioritized during the campaign.
Example of recruitment advertisements on Facebook.
Citation: Weather, Climate, and Society 14, 2; 10.1175/WCAS-D-21-0134.1
2) Sample demographics and other key information
In total, 457 unique survey respondents participated in this study: 321 of the surveys were completed in Spanish, and the remaining 136 were completed in English. Three hundred fifty-three respondents were from PR, and 95 were from USVI (an additional six identified as residing elsewhere and thus were not included in the analyses). Because of considerable drop-off in respondent participation partway through the survey, each survey question presented here may vary in the number of respondents.
The average age of the sample is 54.84 [standard deviation (SD) = 12.79], with a range of 21 to 91 years of age. The sample consisted of a majority of female respondents (76.4%). Approximately one-half identified as Hispanic or Latinx (47.4%), with the remaining respondents being distributed among other racial categories. Approximately two-thirds (66.2%) of the sample had earned a college degree. The average size of a respondent’s household was 2.61 (SD = 1.32) persons, with 35.6% of all households sampled including someone over the age of 65. On average, respondents made approximately $50,000–$59,999 for their annual household income; however, 28.1% did identify that they made less than $20,000 a year. Only one-half of the sample (48.9%) identified that they were employed full time. The average year the respondents’ homes were constructed is 1986 (±16.5 yr), with the majority of respondents living in a concrete block construction home. Of note is that only 0.8% of the sample resided in mobile or manufactured homes. Review Table 1 below for further individual-level respondent characteristics across the PRVI region.
Summary of demographics; M and SD are used to represent mean and standard deviation, respectively; N represents the number of individuals, and the percent is equivalent to the percent of respondents from the sample population as a whole.
c. Survey design
1) Pilot study and design rationale
This survey was modeled after a similar survey used in Collins et al. (2021), which acted as a pilot study for all subsequent research by this team on COVID-19 and hurricane evacuations. Conducted in June 2020, a separate online survey (in both English and Spanish) of 40 questions was disseminated to Florida residents, with 7102 people responding. Since this initial study, an additional set of surveys were deployed by this team for postdisaster behaviors following Hurricanes Laura and Sally. Using the prior experience from these earlier studies, several survey instruments were developed (based on the successful questions from these studies) for the geographic areas of PR and the USVI for both pre- and posthurricane usage. Question wording was kept consistent, when reasonable, to allow for comparisons to be made later between regional surveys.
Several meetings involving the research team members and community stakeholders were organized through virtual sessions to ensure the survey questions were suitable for the target audience. A strong emphasis was placed on collecting information that would be useful for practitioners in the PRVI region managing the COVID-19 pandemic and responsible for coordinating the response to hurricanes as well. Therefore, community stakeholders were involved in question design and pretesting, which occurred before subject recruitment. This was done to test the equivalence of translated instruments and determine a consensus on alternative wording. In general, the questions performed well with regard to participants’ understanding of the questions, perceived clarity, the accuracy of the translation, structure of the questions, and the technology used, with only minor changes needing to be made with the survey instrument. A draft survey instrument was circulated for comment among subject matter experts from communications, emergency management, public health, and geosciences. After further revisions, the survey instruments were submitted to the Ponce Health Sciences Human Research Protection Program, where it was deemed exempt from the Institutional Review Board (IRB).
2) Survey details
A digital survey available through Qualtrics was disseminated within the U.S. territories of the PRVI region between October 2020 and February 2021. Survey instruments were available in English, Spanish, or French Creole and were tested by our team and stakeholders fluent in these languages (note that about 8.6% of the USVI population are French Creole, also known as Patois). This survey collected information about risk perception with evacuation and public sheltering during the COVID-19 pandemic. The survey instrument, consisting of approximately 70 questions, was developed through previous research and pilot testing. These questions addressed the respondent’s demographics, characteristics of their home (i.e., located in a physically vulnerable area, and year built), special needs shelter use, and preexisting health conditions. The survey also included various Likert-scale questions that assess an individual’s perception of risk with regard to sheltering in place versus evacuating or staying at home when considering compound hazards. For further information on the wording of questions, see the online supplemental material included for this paper or the data availability statement.
d. Data analyses
Data were cleaned and organized logically from their original Qualtrics output. After these procedures, data were analyzed using the IBM statistical software program SPSS, version 26. Mapping was produced using ArcGIS Pro, version 2.6.0. These efforts included visualizing the respondent distribution in addition to the social vulnerability maps (presented in Fig. 2 and described above) that were used to target communities that were considered most vulnerable. Due to many of the questions featured in the survey being categorical or ordinal in nature, analytical techniques consisted of nonparametric testing, including chi-square tests, McNemar’s tests, and Spearman’s rho. When analyzing the few continuous variables from this survey, t tests, and ANOVAs were also performed. Any free-response boxes, such as “other” hazards, and health conditions, were cleaned and categorized to allow for analysis on frequency and comparisons between demographic groups. Although analysis was performed comparing many demographic groups with different variables of interest, only those of statistical significance are presented in the following segment.
4. Results
a. Risk perception and preparedness planning
The majority of the sample self-identified that they did not live in an area that is typically advised to evacuate during a hurricane (72.1%) or in an area that has experienced flooding during past hurricanes (70.0%). When looking at homeownership and the built environment, 70.8% of respondents were homeowners, and 23.2% were renters. When asked how concerned they were about general hurricane hazards at their residence, 27.4% said “extremely concerned,” 19.3% said “very concerned,” 23.0% said “concerned,” 20.1% said “somewhat concerned,” and 10.2% said “not concerned.” This was further explored by asking how concerned individuals were on a five-point scale ranging from “extremely concerned” to “not concerned” about a wide range of hurricane-specific hazards. The factors that sparked the most concern were the intensity/category of the storm (mean M = 3.13; SD = 1.10), the wind speed (differentiated from wind gusts) (M = 2.98; SD = 1.13), and size of the storm (M = 2.82; SD = 1.21). Of least concern to the participants were storm surge (M = 1.08; SD = 1.44), tornadoes (M = 1.58; SD = 1.49), landslides (M = 1.62; SD = 1.46), and tsunamis (M = 1.53; SD = 1.59), see Table 2.
Average respondent score for the question, “Please rate how concerned you are about the following hazards at your residence”; M and SD are used to represent mean and standard deviation. Mean score is based on a scale from 0 to 4, with 0 representing “not concerned” and 4 representing “extremely concerned.”
When asked to list any additional concerns or hazards in their area of residence, 41% of respondents identified a water hazard and 14.4% identified landslides. In further breaking down those who identified water hazards, it is seen that 43.9% identified concerns about an inland body of water, with 31.6% concerned about coastal waters and 12.3% concerned about general flooding.
Many respondents identified that they had disaster kits ready (77.1%) and 69.1% of the sample owned generators, but approximately one-half (45.9%) identified not having a hurricane evacuation plan. It was found that individuals residing in Puerto Rico were more likely to have a hurricane plan than their counterparts in the USVI [X2(2) = 8.737, with p = 0.013]. Interestingly, disaster kits were more likely to be owned by individuals with a lower household income (ranging from less than $20,000 to $99,999) than their wealthier counterparts (those earning between $100,000 and greater than $150,000) [X2(4) = 15.308, with p = 0.004]. Of those who did own a generator, 37.2% said that owning one influenced their evacuation decisions.
b. Evacuation decision-making (prior and predicted decisions)
When asked about their predicted evacuation response if a severe hurricane threatened their island, only 20.4% stated that they would evacuate the island. Another 59.4% stated they would not evacuate, and an additional 20.2% stated that they did not know what they would do (Table 3). Individuals in Puerto Rico were more likely to state that they would not evacuate (60.3%) than their counterparts in the USVI (56.1%). Additionally, when asked if they would utilize a public shelter during the 2020 hurricane season, only 14.2% of respondents definitely or probably felt they would. Finally, when asked if they were capable and could evacuate if a hurricane were to impact their area in their present situation, 41.8% said they could, 27.3% said that they could not, 10.2% said maybe, and 20.7% said that they did not know. When asked the question, “How likely is it that you and members of your household will evacuate for the next hurricane that is expected to have a major impact on your area?,” only 15.7% stated that this was “very likely,” 44.1% stated that this is “somewhat likely,” and 40.3% stated that this is “not at all likely.” There was a correlation with past evacuation experience (those who had been ordered to evacuate in the past but did not necessarily evacuate themselves) and their likelihood to evacuate in the future, with those who had been ordered in the past more likely to evacuate in the future [X2(2) = 9.974, with p = 0.007].
Summary of evacuation choice in response to “Would you evacuate the island if you thought the hurricane threat was severe?” (N represents the number of individuals).
For alternate shelter options, the majority of individuals (81.9%) identified that they would stay with family, friends, or a potential combination of the two in the case of a hurricane season as a means of alternative shelter. Of those residing in PR, 52.6% could find alternate shelter from a friend or family member residing in their municipality, 68.8% outside of their municipality, and 60.6% outside of PR. In USVI, 53.2% stated that they could find shelter from a family or friend on their island, 22.1% on another island, and 68.8% outside of the USVI. For future evacuation expectations, many individuals found that the phrasing of emergency management communications was critical. When asked if they were to be ordered to evacuate for a storm instead of being advised, 64.4% agreed that they would be “more likely to evacuate than stay at home.” Notably, 11.3% of the sample has been advised in the past to evacuate for a tropical storm or hurricane. Of this percentage, approximately one-half evacuated when they were recommended to (52.1%).
c. Facilitators and barriers to hurricane evacuation during COVID-19
When asked what would make it easier to evacuate, the two most common responses were access to transportation to a shelter or outside of the area of effect (14.3% of responses) and better communications on shelter and evacuation information, especially if that included COVID-19 specific information (20.1%) (Table 4). Other notable comments include having access to resources and places to evacuate to, having a safe shelter or family home, and for shelters to have adequate resources. For example, 24.6% of the sample stated that it was very difficult to leave their home if they needed to evacuate themselves and household members. However, approximately one-half stated that this would be “somewhat difficult” of a task (47.4%), and 28.0% found it to be “not difficult at all.”
Summary of facilitators and barriers to evacuation, with key phrases in boldface type. These categories were determined from short-answer responses to the questions, “What would make it easier for you to evacuate?” and “What would make it difficult for you to evacuate?” (N represents the number of individuals).
When asked about their barriers to evacuation, many addressed a lack of information on the evacuation process or shelters (14.4%), limits with their transportation (11.5%), and limitations with their pets (14.8%) (Table 4). This is corroborated by those who identified as having a pet (62.0%); among this group, 66.1% identified that their pets influence their evacuation decision. Other notable barriers to evacuation include limitations by their household members, poor shelter conditions, uncertainty in their evacuation decisions, and terrain-related barriers (such as poor road conditions or landslides making areas inaccessible). Of note is that very few (0.8%) identified COVID-19 as a barrier to evacuating. Based on survey results, 14.1% of respondents would require assistance with transportation to leave an evacuation area, with an additional 8.1% unsure if they would need help evacuating. Most people (87.6%) would rely on personal transportation to get them to a safer place during an evacuation.
d. Public shelter perceptions
Individuals’ likelihood of utilizing public shelters has decreased in light of the COVID-19 pandemic. When respondents were presented with the statement, “Prior to COVID-19, if I needed to evacuate to a shelter during the 2020 hurricane season, I would most likely have gone to a shelter,” 8.9% found this to be “definitely true,” 23.9% found this to be “probably true,” 25.4% found this to be “probably false,” and 41.8% found this to be “definitely false.” When posed the statement, “Considering the current situation with COVID-19, I would still go to a shelter if I needed to during a hurricane evacuation advisory in 2020,” 4.1% found this to be “definitely true,” 10.1% “probably true,” 32.2% “probably false,” and 53.6% “definitely false.”
A McNemar’s test was used to determine if the decision to go to public shelters was affected in any way by COVID-19. Grouping was done to produce dichotomous responses by combining responses that found the statements to be true or false (both definitely and probably) into one category. The results of this test showed statistical significance between an individual’s decision to use a public shelter before and during the COVID-19 pandemic (p < 0.001). This reflects that public shelter usage can potentially decrease when the decision is coupled with COVID-19 threats.
Individuals are additionally more likely to have a negative shelter perception, with 51.5% answering “definitely true,” 35.2% “probably true,” 6.1% “probably false,” and 7.3% “definitely false” when presented with the statement, “If my only option was to evacuate to a shelter in my area, I would rather shelter-in-place than risk being exposed to the potentially large group inside a shelter.” There was a significant difference between the territories on responses to this question [X2(1) = 5.727, with p = 0.017]; individuals from the USVI were statistically more likely to answer “true” to this statement. This indicates that those residing in the USVI may be more apprehensive and less trusting of their evacuation shelters as they would rather shelter-in-place to avoid COVID-19 risks. When presented with, “If I was advised to leave my house during a hurricane evacuation, I think the risks of being in a disaster shelter during the COVID-19 pandemic would be worse than staying at home and enduring the risks of a hurricane,” 40.9% found it to be “definitely true,” 34.8% “probably true,” 15.8% “probably false,” and 8.5% “definitely false.” Individuals who viewed this statement as true rated social networking higher as a source of information they relied upon during a hurricane [t(113.525) = 2.046, with p = 0.043].
Additionally, individuals did not have much trust in shelters’ ability to maintain social distancing, with 13.8% answering “definitely true,” 37.2% “probably true,” 27.0% “probably false,” and 22.0% “definitely false” when presented with the statement “I think if I went to a disaster shelter during a hurricane, the authorities will have adequate safeguards in place to prevent the spread of COVID-19 in the facility, such as being able to social distance at least 6 ft. in place.” The results to this question were found to be significantly different between men and women [X2(1) = 6.197, with p = 0.013], with women more likely to answer “false” to this statement, indicating an increased distrust in authorities to protect them. A difference was also found on the political spectrum [X2(6) = 26.853, with p < 0.001]. Those who identified as “very conservative” or “conservative” were more likely to agree with this statement (72% and 64%, respectively), while “moderates” (58%), “liberals” (37.5%), and “very liberal” individuals (50.0%) were less likely to agree in comparison with their “conservative” counterparts. The Likert-scale response results presented in this section are summarized in Table 5.
Summary of Likert-scale responses to statements about public shelters during the COVID-19 pandemic.
e. COVID-19 and health
When this 2020 survey was conducted, 57.5% of respondents were under stay-at-home conditions, and 13.1% had limited business restrictions. Of those surveyed, 91.0% of respondents said they would be able to provide a mask for all family members evacuating with them if they went to a shelter. More than one-half of the sample considered themselves at greater risk for COVID-19 illness due to existing health risks (56.4%). 78.1% of respondents identified that they had one or more health conditions from a list of conditions that were identified as high-risk comorbidities for COVID-19 (CDC 2019) or required electricity for their health condition (since power loss from a hurricane might occur). Additionally, it was found that there was no significant difference between evacuation decision (an individuals’ likelihood to evacuate and their predicted evacuation behavior) and their self-identified health status (both in regard to COVID-19 vulnerability and their preexisting health conditions), indicating that health conditions did not play a role in the decision to evacuate or stay at home.
f. Sources of information
Questions addressing their most relied upon sources of information during a hurricane were also presented with a Likert-type response option. Overall, the most relied-on information sources were the national media, electronic media, and radio broadcasts. Whereas the least relied on were print media, social networking (e.g., Facebook), and friends a distance away from the respondent (see Table 6). However, when comparing the results of the question, “How likely is it that you and members of your household will evacuate for the next hurricane that is expected to have a major impact on your area?” with sources of information on which the three groups most relied (those who were very, somewhat, and not at all likely to evacuate), it was found that there was a significant difference in the reliability self-rating between them about friends—both those nearby [X2(8) = 28.936, with p < 0.001] and those a distance away [X2(8) = 21.506, with p = 0.006]—as an information source. For both of these information sources (friends near and far away), those who replied “somewhat likely” had the highest mean rating for communication from a friend on their reliance during a hurricane; this is then followed by those who answered “very likely” and then finally those who answered, “not at all likely.”
Sources of information that individuals relied upon when making prior hurricane evacuation decisions; M and SD are used to represent mean and standard deviation; mean score is based on a scale from 1 to 5, with 5 being “more likely to rely upon” when making prior hurricane evacuation decisions.
5. Discussion
More than one-half of the sample considered themselves vulnerable to COVID-19. When asked about their actions in the 2020 hurricane season, individuals who would have previously evacuated to a shelter would choose not to during the pandemic, reflecting that public shelter usage has the potential to decrease when the decision is coupled with COVID-19 threats. Additionally, individuals were shown to have a negative perception of public shelter options, with approximately one-half of the respondents having little faith in shelters’ ability to maintain social distancing and other preventative actions. This distrust in the shelters’ ability to ensure their safety in the 2020 hurricane season was corroborated in a recent community assessment for public health emergency response (CASPER) conducted by the Puerto Rico Department of Health (2020), with 53% of individuals lacking confidence in staff to ensure their safety in shelter. This distrust was seen to be statistically higher among women, which is supported by prior research highlighting that women tend to be more risk-averse, more apt to worry, and may potentially overestimate their personal disaster risks relative to their male counterparts (Brody 1984; Brown et al. 2021; Gustafson 1998; Kellens et al. 2011; Lindell and Hwang 2008).
Further, more than one-half the sample stated that they would “definitely” shelter-in-place to minimize exposure to large groups in the shelter, and 75.5% of the sample thought that the risks of being in a shelter were higher than those that they would have to endure if they were to stay home during a hurricane. These results are comparable to the responses seen in Florida (Collins et al. 2021), where 74.3% of individuals felt shelters were riskier than their homes during the pandemic and that more than one-half agreed that they would aim to shelter-in-place rather than risk exposure. So, although, as stated earlier, very few respondents noted COVID-19 was a barrier to evacuation, this is not the case when one considers specifically going to a shelter rather than other evacuation choices where people may not be concerned about congregating with many people. Our survey made a clear distinction between going to a shelter and evacuation, which included going to a friend or family member’s home, or going to a hotel.
The majority of the sample (59.4%) would not evacuate for an impending severe hurricane, and only 14.2% would utilize a public shelter. Of note is that no differences in the choice to evacuate were found by demographic groups (such as gender, income, education, and race), which is in conflict with the literature previously mentioned; even further, race did not affect any of the key variables (shelter perception, sources of information, disaster preparedness, COVID-19 vulnerability, etc.). This is drastically different from decision-making on evacuation revealed by prior research in the PRVI region. Morrow and Gladwin (2014a,b) found that 68% of Puerto Rico residents and 61% of residents in the USVI would evacuate for a major storm and that 22% of people in Puerto Rico and 29% of people in the USVI would utilize public sheltering. This change may indicate the difference that COVID-19 can play in evacuation decision-making during a hurricane event. However, in the more recent CASPER (Puerto Rico Department of Health 2020), it was found that 47% of those surveyed would go to a disaster shelter during 2020, showing the need for more analysis of the role of COVID-19 on sheltering decisions.
Fortunately, many individuals did identify that they had taken protective actions at their households, including preparing disaster kits, owning generators, and having hurricane evacuation plans established with their families. Interestingly, contrary to prior literature that states that individuals with more financial means are more likely to have taken disaster preparedness measures (Elder et al. 2007; Donner and Lavariega-Montforti 2018; Fothergill and Peek 2004), those from a low household income were more likely to have a disaster kit than those who made over $100,000 a year. Additionally, 41.8% did identify that they think they would be able to evacuate in the future, indicating that despite choosing to stay at home, many individuals would likely be able to evacuate if absolutely necessary. The most commonly mentioned aids to evacuation were increased options for transportation to shelters and better communication of shelter and evacuation information. Understanding geographic location in relation to evacuation decisions will allow public health preparedness planners to identify municipalities in those areas with the highest need.
Almost one-quarter of respondents stated that they would have found it very difficult to evacuate their households if needed, showing some potential room for improvement in ensuring equitable hurricane preparedness for all. The most commonly mentioned barriers to evacuation were a lack of easily obtained evacuation and shelter information, and limitations on transportation and pets. Potential communication methods should focus on national media, electronic media, and radio as these were the most relied upon sources of information used by survey respondents during a hurricane. Additionally, communications should continue to emphasize that the intensity of a storm is not the only predictor of its risk and emphasize the risks of water-based hazards, which have historically posed the largest threat to loss of life, as it was shown that survey participants ranked intensity as their top hurricane concern while also placing storm surge in last.
Limitations
Some limitations of this study are acknowledged. Generalization of results may not be applicable to the entire population of PR and the USVI due to the sample not being representative of the population, as noted when comparing with census data. The sample featured in this research has an underrepresentation of USVI (21.1%) as compared with PR (78.8%). Additionally, the sample is differing from the 2019 PR population estimates (U.S. Census Bureau 2019), specifically for the high percentage of female respondents (76.4% as compared with the 2019 estimate of 52.5%), older age (M = 54.84, SD = 12.79; 2019 estimates predict only 21.3% of the population is 65 years of age or older), higher socioeconomic status (with the sample identifying a median income between $50,000 and $59,999, whereas the median household income reported in the 2019 estimates was $20,539), and more diverse racial sample (with only 47.4% of the sample identifying as Hispanic or Latinx as compared with 98.7% in the 2019 census estimates).
Additionally, since PR is overrepresented in this sample, the USVI could have unique differences not well captured through the initial sample obtained in this study. As of 2010, the Creole-speaking population in USVI was 8.6% of the population; 2.6% of the population who speak Creole have an English-speaking proficiency that is less than “very well” (U.S. Census Bureau 2010). Although this survey was available in French Creole and some connections were established with this community, we had no respondents. As such, this population in the USVI is not represented in this sample. These discrepancies could be due to the digital format that the survey was presented in, as those without access to necessary technology or internet (such as those from a lower socioeconomic status) would not be able to take the survey. Because of the complicated nature of the COVID-19 pandemic and associated university closures, it was not possible to supplement the study with in-person interviews as planned for this initial survey. A final sample limitation is that the majority of the sample self-identified that they did not live in an area that is typically advised to evacuate during a hurricane (72.1%) or in an area that has experienced flooding during past hurricanes (70.0%), potentially causing a bias in the results as compared with a group that would be evacuated for a hurricane.
6. Conclusions
This study captures a unique time of the COVID-19 pandemic when vaccinations were not available, and many emergency managers were struggling to balance the need to protect people from potential natural hazards while preventing the spread of COVID-19. As seen in the results, the COVID-19 pandemic has had an effect on how people make their decisions to evacuate, particularly in regard to shelter usage. Public shelters are typically used by those who do not have economic means and are part of the most vulnerable segments of our population. Most telling is that three-quarters of the sample would rather endure the risks of a hurricane than risk staying in a public shelter. Due to the compounding hazards presented during the COVID-19 pandemic, utilization of these shelters was predicted to decrease as people have a negative perception of shelters, causing potentially harmful and dangerous scenarios for those who both feel uncomfortable with large group gatherings at shelters and lack the means to enact an alternate evacuation plan. Additionally, the majority of the sample indicated that they would not evacuate their homes for any future hurricanes. These numbers are historically different from past years, highlighting that many people are taking to heart the messaging put out by emergency management in 2020 to stay at home if they feel they can. However, many people did identify that they had taken preparedness measures. Approximately one-quarter of participants found that they would have a difficult time evacuating if needed. The two key factors affecting their ease of evacuation were communications and transportation access, which should both be improved upon to ensure equitability in evacuation. Communications during hurricane events, especially those that mention the COVID-19 pandemic and accurately portray water-based hurricane hazards, need to improve so as not to act as a barrier to evacuation in any future pandemic-disaster compounding events.
This study adds to the existing body of knowledge on this subject and benefits emergency planners by providing information that could benefit their efforts of communicating risk and targeted messaging to their constituents. Findings help guide existing plans across the PRVI region based on intended future actions of residents if confronted with a hurricane of major impact while providing knowledge of potential barriers, concerns, and misconceptions that may exist due to the pandemic. Understanding potential needs for emergency sheltering and transportation during the response along with public health services helps forecast future staffing needs and identify gaps in an already fatigued workforce due to the pandemic.
Adaptation to build the capacity of communities and local governments needs to occur at the community level (Keim 2008). Involving culturally diverse and vulnerable populations in evacuation and decision-making assessments ensures protective actions are designed for individuals and communities to realistically operationalize mitigation capabilities when needing to evacuate during the threat of a natural disaster during the pandemic (Arlikatti et al. 2018). Our study serves as a conceptual model for involving academic institutions, regional, territorial, and local government agencies, community organizations, and trusted leaders in interdisciplinary research designed to arm healthcare and service providers with more information to protect vulnerable populations and ensure access to services are maintained before, during, and after the impact of cascading disasters. Integrating public health methods and epidemiological analyses ensures emergency operations plans take into consideration members of vulnerable communities when planning emergency management and public health activities.
Future work
Future work includes additional analyses that will be generated from the wealth of data gathered from this survey. For example, conducting geospatial and hot-spot analysis on a survey question that was not addressed within this paper about to which municipalities individuals would consider evacuating can provide some indication of which municipalities would have the highest amounts of evacuees or shelter users, allowing emergency management to prepare accordingly in areas where there is an anticipated influx of individuals. Determining and prioritizing locations where there is a need for congregate shelters provides an opportunity to reduce the number of shelters in areas with little need while identifying facilities in close proximity to demand.
Furthermore, a second survey was created and disseminated prior to the start of the 2021 hurricane season in the PRVI region. These data will create a timeline of changes in attitudes on evacuations and COVID-19 as we are further into the pandemic. Additionally, this 2021 season survey collected information on an individual’s vaccination status and how that would affect their decision to evacuate. Future work, when in-person fieldwork is possible, could capture data from a more representative sample focusing on those without access to technology or the internet. Further analysis comparing the Florida survey data (Collins et al. 2021) and the two subsequent PRVI surveys (including this one presented here) will be conducted to see whether there were differences in evacuation and sheltering responses and risk perception of residents for Florida, Puerto Rico, and the U.S. Virgin Islands.
Acknowledgments.
We acknowledge our extensive stakeholder team for their efforts in survey question feedback, testing, and distribution. We also thank all of our translators, with special acknowledgement of Louiseul Azor, for assisting with the Creole translation. We acknowledge funding from the Natural Hazards Center and anonymous reviewers who contributed to the Natural Hazards Center report on which this paper is based. The Quick Response Research Award Program is based on work supported by the National Science Foundation (NSF; Award 1635593). This Special Call for Quick Response Research in U.S. Territories is made possible through supplemental funding from the Centers for Disease Control and Prevention (CDC). Any opinions, findings, conclusions, or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the NSF, CDC, or Natural Hazards Center.
Data availability statement.
The dataset is not available, but the summarized reports and survey instruments are available online (https://doi.org/10.17603/ds2-7czq-en54). In addition, a report provided to the Natural Hazards Center on this work is available online (https://hazards.colorado.edu/public-health-disaster-research/compound-hazards-evacuations-and-shelter-choices).
REFERENCES
Abuelgasim, E., L. J. Saw, M. Shirke, M. Zeinah, and A. Harky, 2020: COVID-19: Unique public health issues facing Black, Asian, and minority ethnic communities. Curr. Probl. Cardiol., 45, 100621, https://doi.org/10.1016/j.cpcardiol.2020.100621.
Acosta, R. J., N. Kishore, R. A. Irizarry, and C. O. Buckee, 2020: Quantifying the dynamics of migration after Hurricane Maria in Puerto Rico. Proc. Natl. Acad. Sci., 117, 32772–32778, https://doi.org/10.1073/pnas.2001671117.
Aldrich, N., and W. F. Benson, 2008: Disaster preparedness and the chronic disease needs of vulnerable older adults. Prev. Chronic Dis., 5, 1–7.
Alvarez, C. H., and C. R. Evans, 2021: Intersectional environmental justice and population health inequalities: A novel approach. Soc. Sci. Med., 269, 113559, https://doi.org/10.1016/j.socscimed.2020.113559.
Arlikatti, S., P. Maghelal, N. Agnimitra, and V. Chatterjee, 2018: Should I stay or should I go? Mitigation strategies for flash flooding in India. Int. J. Disaster Risk Reduct., 27, 48–56, https://doi.org/10.1016/j.ijdrr.2017.09.019.
ATSDR, 2021: CDC/ATSDR social vulnerability index. Accessed 14 January 2022, https://www.atsdr.cdc.gov/placeandhealth/svi/index.html.
Baker, E. J., 1991: Hurricane evacuation behavior. Int. J. Mass Emerg. Disasters, 9, 287–310.
Baker, E. J., 1995: Public response to hurricane probability forecasts. Prof. Geogr., 47, 137–147, https://doi.org/10.1111/j.0033-0124.1995.00137.x.
Blaikie, P., T. Cannon, I. Davis, and B. Wisner, 2005: At Risk: Natural Hazards, People’s Vulnerability and Disasters. Routledge, 464 pp.
Borowski, E., and A. Stathopoulos, 2020: On-demand ridesourcing for urban emergency evacuation events: An exploration of message content, emotionality, and intersectionality. Int. J. Disaster Risk Reduct., 44, 101406, https://doi.org/10.1016/j.ijdrr.2019.101406.
Botzen, W. J. W., J. M. Mol, P. J. Robinson, J. Zhang, and J. Czajkowski, 2022: Individual hurricane evacuation intentions during the COVID-19 pandemic: Insights for risk communication and emergency management policies. Nat. Hazards, 111, 507–522, https://doi.org/10.1007/s11069-021-05064-2.
Brody, C. J., 1984: Differences by sex in support for nuclear power. Soc. Forces, 63, 209–228, https://doi.org/10.2307/2578866.
Brown, G. D., A. Largey, and C. McMullan, 2021: The impact of gender on risk perception: Implications for EU member states’ national risk assessment processes. Int. J. Disaster Risk Reduct., 63, 102452, https://doi.org/10.1016/j.ijdrr.2021.102452.
Brown, L. M., D. M. Dosa, K. Thomas, K. Hyer, Z. Feng, and V. Mor, 2012: The effects of evacuation on nursing home residents with dementia. Amer. J. Alzheimers Dis. Other Dementias, 27, 406–412, https://doi.org/10.1177/1533317512454709.
Brunkard, J., G. Namulanda, and R. Ratard, 2013: Hurricane Katrina deaths, Louisiana, 2005. Disaster Med. Public Health Prep., 2, 215–223, https://doi.org/10.1097/DMP.0b013e31818aaf55.
Buckle, P., 2006: Assessing social resilience. Disaster Resilience: An Integrated Approach, D. Paxton and D. Johnston, Eds., Charles C. Thomas Publishers, 88–104.
Burger, J., and M. Gochfeld, 2020: Involving community members in preparedness and resiliency involves bi-directional and iterative communication and actions: A case study of vulnerable populations in New Jersey following Superstorm Sandy. J. Risk Res., 23, 541–556, https://doi.org/10.1080/13669877.2019.1593221.
CDC, 2017: Influenza (flu): Past pandemics. Accessed 14 January 2022, https://www.cdc.gov/flu/pandemic-resources/basics/past-pandemics.html.
CDC, 2019: COVID-19: People with certain medical conditions. Accessed 14 January 2022, https://www.cdc.gov/coronavirus/2019-ncov/need-extra-precautions/people-with-medical-conditions.html.
CDC, 2020: COVID-19: Risk for COVID-19 infection, hospitalization, and death by race/ethnicity. Accessed 14 January 2022, https://www.cdc.gov/coronavirus/2019-ncov/covid-data/investigations-discovery/hospitalization-death-by-race-ethnicity.html.
CDC/ATSDR/Geospatial Research, Analysis, and Services Program, 2018: CDC/ATSDR social vulnerability index 2018 database: Puerto Rico. Accessed 14 January 2022, https://www.atsdr.cdc.gov/placeandhealth/svi/data_documentation_download.html.
Chaplin, D., J. Twigg, and E. Lovell, 2019: Intersectional approaches to vulnerability reduction and resilience-building. Resilience Intel, No. 12, BRACED, 35 pp., https://cdn.odi.org/media/documents/12651.pdf.
Chisty, M. A., S. E. A. Dola, N. A. Khan, and M. M. Rahman, 2021: Intersectionality, vulnerability and resilience: Why it is important to review the diversifications within groups at risk to achieve a resilient community. Continuity Resilience Rev., 3, 119–131, https://doi.org/10.1108/CRR-03-2021-0007.
Collins, J., R. Ersing, and A. Polen, 2017: Evacuation decision-making during Hurricane Matthew: An assessment of the effects of social connections. Wea. Climate Soc., 9, 769–776, https://doi.org/10.1175/WCAS-D-17-0047.1.
Collins, J., R. Ersing, A. Polen, M. Saunders, and J. Senkbeil, 2018: The effects of social connections on evacuation decision making during Hurricane Irma. Wea. Climate Soc., 10, 459–469, https://doi.org/10.1175/WCAS-D-17-0119.1.
Collins, J., A. Polen, K. McSweeney, D. Colón-Burgos, and I. Jernigan, 2021: Hurricane risk perceptions and evacuation decision-making in the age of COVID-19. Bull. Amer. Meteor. Soc., 102, E836–E848, https://doi.org/10.1175/BAMS-D-20-0229.1.
Cox, D., and Coauthors, 2019: Hurricanes Irma and Maria post-event survey in US Virgin Islands. Coastal Eng. J., 61, 121–134, https://doi.org/10.1080/21664250.2018.1558920.
Demuth, J. L., R. E. Morss, B. H. Morrow, and J. K. Lazo, 2012: Creation and communication of hurricane risk information. Bull. Amer. Meteor. Soc., 93, 1133–1145, https://doi.org/10.1175/BAMS-D-11-00150.1.
Donner, W. R., and J. Lavariega-Montforti, 2018: Ethnicity, income and disaster preparedness in deep south Texas, United States. Disasters, 42, 719–733, https://doi.org/10.1111/disa.12277.
Dow, K., and S. L. Cutter, 2000: Public orders and personal opinions: Household strategies for hurricane risk assessment. Global Environ. Change, 2B, 143–155, https://doi.org/10.1016/S1464-2867(01)00014-6.
Eisenman, D. P., K. M. Cordasco, S. Asch, J. F. Golden, and D. Glik, 2007: Disaster planning and risk communication with vulnerable communities: Lessons From Hurricane Katrina. Amer. J. Public Health, 97 (Suppl. 1), S109–S115, https://doi.org/10.2105/AJPH.2005.084335.
Elder, K., S. Xirasagar, N. Miller, S. A. Bowen, S. Glover, and C. Piper, 2007: African Americans’ decisions not to evacuate New Orleans before Hurricane Katrina: A qualitative study. Amer. J. Public Health, 97 (Suppl. 1), S124–S129, https://doi.org/10.2105/AJPH.2006.100867.
Elsner, J. B., A. B. Kara, and M. A. Owens, 1999: Fluctuations in North Atlantic hurricane frequency. J. Climate, 12, 427–437, https://doi.org/10.1175/1520-0442(1999)012<0427:FINAHF>2.0.CO;2.
Elsner, J. B., T. H. Jagger, and X. Niu, 2000: Changes in the rates of North Atlantic major hurricane activity during the 20th century. Geophys. Res. Lett., 27, 1743–1746, https://doi.org/10.1029/2000GL011453.
Federal Emergency Management Agency, 2011: A whole community approach to emergency management: Principles, themes, and pathways for action. FEMA FDOC 104-008-1, 28 pp., https://www.fema.gov/sites/default/files/2020-07/whole_community_dec2011__2.pdf.
Fothergill, A., and L. Peek, 2004: Poverty and disasters in the United States: A review of recent sociological findings. Nat. Hazards, 32, 89–110, https://doi.org/10.1023/B:NHAZ.0000026792.76181.d9.
Gladwin, C., H. Gladwin, and W. Peacock, 2001: Modeling hurricane evacuation decisions with ethnographic methods. Int. J. Mass Emerg. Disasters, 19, 117–143.
Gustafson, P. E., 1998: Gender differences in risk perception: Theoretical and methodological perspectives. Risk Anal., 18, 805–811, https://doi.org/10.1023/B:RIAN.0000005926.03250.c0.
Hill, S., N. S. Hutton, J. L. Whytlaw, J. E. Yusuf, J. G. Behr, E. Landaeta, and R. Diaz, 2021: Changing logistics of evacuation transportation in hazardous settings during COVID-19. Nat. Hazards Rev., 22, 04021029, https://doi.org/10.1061/(ASCE)NH.1527-6996.0000506.
Hutchins, S. S., K. Fiscella, R. S. Levine, D. C. Ompad, and M. McDonald, 2009: Protection of racial/ethnic minority populations during an influenza pandemic. Amer. J. Public Health, 99 (S2), S261–S270, https://doi.org/10.2105/AJPH.2009.161505.
Ivers, L. C., and E. T. Ryan, 2006: Infectious diseases of severe weather-related and flood-related natural disasters. Curr. Opin. Infect. Dis., 19, 408–414, https://doi.org/10.1097/01.qco.0000244044.85393.9e.
Jury, M. R., R. Rios-Berrios, and E. García, 2012: Caribbean hurricanes: Changes of intensity and track prediction. Theor. Appl. Climatol., 107, 297–311, https://doi.org/10.1007/s00704-011-0461-5.
Jury, M. R., S. Chiao, and R. Cécé, 2019: The intensification of Hurricane Maria 2017 in the Antilles. Atmosphere, 10, 590–612, https://doi.org/10.3390/atmos10100590.
Kaiser Family Foundation, 2017: U.S. Virgin Islands: Fast facts. Accessed 14 January 2022, https://www.kff.org/racial-equity-and-health-policy/fact-sheet/u-s-virgin-islands-fast-facts/.
Keim, M. E., 2008: Building human resilience: The role of public health preparedness and response as an adaptation to climate change. Amer. J. Prev. Med., 35, 508–516, https://doi.org/10.1016/j.amepre.2008.08.022.
Kellens, W., R. Zaalberg, T. Neutens, W. Vanneuville, and P. De Maeyer, 2011: An analysis of the public perception of flood risk on the Belgian coast. Risk Anal., 31, 1055–1068, https://doi.org/10.1111/j.1539-6924.2010.01571.x.
Kirby, T., 2020: Evidence mounts on the disproportionate effect of COVID-19 on ethnic minorities. Lancet, 8, 547–548, https://doi.org/10.1016/S2213-2600(20)30228-9.
Klotzbach, P. J., M. M. Bell, and J. Jhordanne, 2021: Extended range forecast of Atlantic seasonal hurricane activity and landfall strike probability for 2021. Colorado State University Rep., 34 pp., https://tropical.colostate.edu/Forecast/2021-04.pdf.
Kuran, C. H. A., and Coauthors, 2020: Vulnerability and vulnerable groups from an intersectionality perspective. Int. J. Disaster Risk Reduct., 50, 101826, https://doi.org/10.1016/j.ijdrr.2020.101826.
Lemonick, D. M., 2011: Epidemics after natural disasters. Amer. J. Clin. Med., 8, 144–152.
Lindell, M. K., and S. N. Hwang, 2008: Households’ perceived personal risk and responses in a multihazard environment. Risk Anal., 28, 539–556, https://doi.org/10.1111/j.1539-6924.2008.01032.x.
Meng, Y., and Coauthors, 2020: Sex-specific clinical characteristics and prognosis of coronavirus disease-19 infection in Wuhan, China: A retrospective study of 168 severe patients. PLOS Pathog., 16, e1008520, https://doi.org/10.1371/journal.ppat.1008520.
Miller, L. M., 2007: Collective disaster responses to Katrina and Rita: Exploring therapeutic community, social capital, and social control. J. Rural Soc. Sci., 22, 45–63.
Moore, S., M. Daniel, L. Linnan, M. Campbell, S. Benedict, and A. Meier, 2004: After Hurricane Floyd passed: Investigating the social determinants of disaster preparedness and recovery. Fam. Commun. Health, 27, 204–217, https://doi.org/10.1097/00003727-200407000-00007.
Morrow, B., and H. Gladwin, 2014a: Puerto Rico hurricane evacuation study: Behavioral analysis. FEMA Final Rep., 46 pp.
Morrow, B., and H. Gladwin, 2014b: U.S. Virgin Islands hurricane evacuation study: Behavioral analysis. FEMA Final Rep., 52 pp.
Morss, R. E., J. L. Demuth, J. K. Lazo, K. Dickinson, H. Lazrus, and B. H. Morrow, 2016: Understanding public hurricane evacuation decisions and responses to forecast and warning messages. Wea. Forecasting, 31, 395–417, https://doi.org/10.1175/WAF-D-15-0066.1.
National Institute on Aging, 2017: Supporting older patients with chronic conditions. Accessed 27 August 2021, https://www.nia.nih.gov/health/supporting-older-patients-chronic-conditions.
New York Times, 2022a: Tracking coronavirus in Puerto Rico: Latest map and case count. Accessed 14 January 2022, https://www.nytimes.com/interactive/2021/us/puerto-rico-covid-cases.html.
New York Times, 2022b: Tracking coronavirus in the US Virgin Islands: Latest map and case count. Accessed 14 January 2022, https://www.nytimes.com/interactive/2021/us/virgin-islands-covid-cases.html.
Norris, F. H., 2006: Disaster research methods: Past progress and future directions. J. Trauma. Stress, 19, 173–184, https://doi.org/10.1002/jts.20109.
Pan American Health Organization, 2017: Hurricanes Irma and Maria. PAHO Situation Rep. 8, 3 pp., accessed 15 January 2022, https://www.paho.org/disasters/dmdocuments/PAHOWDC_SituationReport8_Hurricane%202017_22Sept2017.pdf.
Phillips, C. A., and Coauthors, 2020: Compound climate risks in the COVID-19 pandemic. Nat. Climate Change, 10, 586–588, https://doi.org/10.1038/s41558-020-0804-2.
Puerto Rico Department of Health, 2020: A rapid community initial assessment to estimate the seroprevalence of COVID-19. Puerto Rico Science, Technology, and Research Trust Rep., 51 pp.
Puerto Rico Department of Health, 2022: COVID-19 en cifras en Puerto Rico. Accessed 14 January 2022, https://covid19datos.salud.gov.pr/#casos.
Schnall, A. H., J. Roth, L. L. Ekpo, I. Guendel, M. Davis, and E. M. Ellis, 2019: Disaster-related surveillance among US Virgin Islands (USVI) shelters during the Hurricanes Irma and Maria response. Disaster Med. Public Health Prep., 13, 38–43, https://doi.org/10.1017/dmp.2018.146.
Semega, J., M. Kollar, E. A. Shrider, and J. Creamer, 2020: Income and poverty in the United States: 2019. U.S. Census Bureau Rep. P60-270, accessed 14 January 2022, https://www.census.gov/library/publications/2020/demo/p60-270.html.
Senkbeil, J., J. Collins, and J. Reed, 2019: Evacuee perception of geophysical hazards for Hurricane Irma. Wea. Climate Soc., 11, 217–227, https://doi.org/10.1175/WCAS-D-18-0019.1.
Sherman-Morris, K., J. Senkbeil, and R. Carver, 2011: Who’s Googling what? What internet searches reveal about hurricane information seeking. Bull. Amer. Meteor. Soc., 92, 975–985, https://doi.org/10.1175/2011BAMS3053.1.
Shukla, M. A., L. Woc-Colburn, and J. E. Weatherhead, 2018: Infectious diseases in the aftermath of hurricanes in the United States. Curr. Trop. Med. Rep., 5, 217–223, https://doi.org/10.1007/s40475-018-0162-6.
Shultz, J. M., J. P. Kossin, J. M. Shepherd, J. M. Ransdell, R. Walshe, I. Kelman, and S. Galea, 2019: Risks, health consequences, and response challenges for small-island-based populations: Observations from the 2017 Atlantic hurricane season. Disaster Med. Public Health Prep., 13, 5–17, https://doi.org/10.1017/dmp.2018.28.
Shultz, J. M., and Coauthors, 2020a: Mitigating the twin threats of climate-driven Atlantic hurricanes and COVID-19 transmission. Disaster Med. Public Health Prep., 14, 494–503, https://doi.org/10.1017/dmp.2020.243.
Shultz, J. M., C. Fugate, and S. Galea, 2020b: Cascading risks of COVID-19 resurgence during an active 2020 Atlantic hurricane season. JAMA, 324, 935–936, https://doi.org/10.1001/jama.2020.15398.
Shultz, J. M., J. P. Kossin, A. Ali, V. Borowy, C. Fugate, Z. Espinel, and S. Galea, 2020c: Superimposed threats to population health from tropical cyclones in the prevaccine era of COVID-19. Lancet, 4, e506–e508, https://doi.org/10.1016/S2542-5196(20)30250-3.
Slovic, P., 1994: Perceptions of risk: Paradox and challenge. Future Risks and Risk Management, Springer, 63–78.
Tai, D. B. G., A. Shah, C. A. Doubeni, I. G. Sia, and M. L. Wieland, 2021: The disproportionate impact of COVID-19 on racial and ethnic minorities in the United States. Clin. Infect. Dis., 72, 703–706, https://doi.org/10.1093/cid/ciaa815.
U.S. Census Bureau, 2010: U.S. Virgin Islands demographics. Accessed 27 August 2021, https://www.census.gov/data/datasets/2010/dec/virgin-islands.html.
U.S. Census Bureau, 2019: Quick facts: Puerto Rico 2010–2019. Accessed 27 August 2021, https://www.census.gov/quickfacts/PR.
Versey, H. S., 2021: Missing pieces in the discussion on climate change and risk: Intersectionality and compounded vulnerability. Policy Insights Behav. Brain Sci., 8, 67–75, https://doi.org/10.1177/2372732220982628.
Vitousek, S., P. L. Barnard, C. H. Fletcher, N. Frazer, L. Erikson, and C. D. Storlazzi, 2017: Doubling of coastal flooding frequency within decades due to sea-level rise. Sci. Rep., 7, 1399, https://doi.org/10.1038/s41598-017-01362-7.
Walker, H. M., A. Culham, A. J. Fletcher, and M. G. Reed, 2019: Social dimensions of climate hazards in rural communities of the global north: An intersectionality framework. J. Rural Stud., 72, 1–10, https://doi.org/10.1016/j.jrurstud.2019.09.012.
Wang, G., T. J. Kearns, J. Yu, and G. Sanchez, 2014: A stable reference frame for landslide monitoring using GPS in the Puerto Rico and Virgin Islands region. Landslides, 11, 119–129, https://doi.org/10.1007/s10346-013-0428-y.
Whitehead, J. C., B. Edwards, M. Van Willigen, J. R. Maiolo, K. Wilson, and K. T. Smith, 2000: Heading for higher ground: Factors affecting real and hypothetical hurricane evacuation behavior. Global Environ. Change, 2B, 133–142, https://doi.org/10.1016/S1464-2867(01)00013-4.
Whytlaw, J. L., N. Hutton, J. E. W. Yusuf, T. Richardson, S. Hill, T. Olanrewaju-Lasisi, and R. Diaz, 2021: Changing vulnerability for hurricane evacuation during a pandemic: Issues and anticipated responses in the early days of the COVID-19 pandemic. Int. J. Disaster Risk Reduct., 61, 102386, https://doi.org/10.1016/j.ijdrr.2021.102386.
World Health Organization, 2021: WHO coronavirus (COVID-19) dashboard. Accessed 31 August 2021, https://covid19.who.int/.