Introduction

Disaster events are devastating to the communities they strike, bringing both financial and health-related impacts. In 2020, annual disaster event costs in the United States (U.S.) hit a record high at $95 billion [1]. To respond to disaster events, U.S. emergency management policies have been in place since the early 1800s [2]. Initially, these policies focused on providing funding to communities affected by disaster events [2]. With the creation of the Federal Emergency Management Agency (FEMA) in 1979, these policies shifted to also provide emergency management services, and eventually the development of the National Response Framework for management of disaster event response and recovery [2]. Today, most U.S. jurisdictions have a local emergency manager or emergency management team to develop a response to disaster events [3]. Disaster occurrence may necessitate evacuation, shelter-in-place mandates, or relocation efforts for individuals and their families [3], all of which can lead to reduced mental health and/or injury in people, as well as illness or loss of life in animals [4–7].

Although approximately 70% of households own pets, also known as companion animals [8], prior to Hurricane Katrina in 2005, few emergency management plans made accommodations for companion animals [4]. The poor health outcomes in people and pets during Hurricane Katrina in 2005 resulted in the passage of the Pets Evacuation and Transportation Standards Act (PETS) in 2006 [9], an amendment of the 1988 Stafford Disaster Relief and Emergency Assistance Act [9]. The PETS Act encourages state and local governments to consider animal welfare by making the inclusion of companion animals in their emergency planning a requirement to receive FEMA funding [9]. Although animal shelters and rescues are involved in disaster preparedness and provide harm-reduction services such as microchipping and vaccine clinics, there is still a disconnect between animal and human disaster planning [6].

Integrating animal welfare into human disaster planning is important, as scientific literature indicates that companion animal ownership is currently a risk factor for individuals experiencing disaster events [7, 10–12]. One study found that during Hurricane Irene, a lower percentage of pet owners evacuated (57% evacuated) compared to non-pet owners (63% evacuated) though the differences were not significant (X² = 1.2, ns) [13]. Among pet owners who did not evacuate, the slight majority (51%) cited pet-related reasons for not leaving their home [13]. Individuals may delay evacuation if their companion animal is unable to evacuate with them [5]. They may refuse evacuation to avoid abandoning a companion animal or enter hazardous situations, such as buildings with weakened infrastructure, to retrieve or care for a pet that was left behind [4]. Due to lack of pet-friendly housing, people may also choose to reside in unsafe living conditions post-disaster to keep their pets [5]. These choices can negatively impact first responders, who may be required to rescue or care for these individuals [14].

In addition to physical harm, people are also at risk of psychological harm when their companion animals are not accounted for in a disaster response. One survey found that over 63% of respondents considered their pet to be an important coping mechanism during stressful situations [15], so losing a companion animal during a disaster event could lead to increased levels of stress and anxiety. Following the 2008 Chaiten Eruption in southern Chile, 4,500 humans were evacuated from the area, but were prohibited from bringing their companion animals [16]. Following evacuation, negative psychosocial effects were identified [17]. Additionally, Chadwin found the median self-reported PTSD score 4.4 years after the Great East Japan Earthquakes of 2011 was 50% lower in pet owners compared to non-pet owners [18]. Accommodating pet evacuation along with human evacuation can improve mental health outcomes among affected populations.

Considering and planning for companion animals is front-and-center for emergency managers and state veterinary offices around the U.S. It is essential that companion animal welfare is included in disaster response plans [3, 14]. The most effective way to make sure that animal welfare is being protected is via collaboration between local governments, emergency management, animal shelter and rescue organizations, veterinarians, and pet owners themselves [4, 14, 19]. However, there is evidence that animals are not yet fully integrated into human disaster management. For example, over 3,000 animals were determined to have been killed during the 2010 Canterbury earthquake in Australia, 99% of which might have been preventable if animal welfare provision plans had included strategies on safe animal evacuation and how animals should be cared for once they were relocated to evacuation centers [4]. Animal care facilities can and do create independent, proactive actions to prepare for disaster. Relatively modifiable household factors such as having a cat carrier can facilitate pet owner evacuation [20]. Additionally, vaccine clinics prior to a disaster, encouraging companion animal microchipping, and evaluating the amount of space the shelter or rescue has for intakes were a disaster to occur could be aligned with the regional disaster patterns [6, 11, 21]. Although these are all actions being taken at jurisdictional levels, there is little documentation of this type of collaboration within the scientific literature or the integration of companion animal considerations into community-wide climate response.

Given the importance of integrating human and animal disaster plans, combined with the fact that communities are currently experiencing adverse consequences associated with climate change, including extreme temperatures, more frequent and severe storms, drought, and rising sea levels [22] an organized, One Health disaster response that addresses human and animal needs is essential for the continued welfare of companion animals and their owners. To support the development of animal-specific adaptation plans, we used disaster event occurrence as our exposure variable. We analyzed the changes in animal shelter intake and outtake rates for cats and dogs during and after two disasters—fire and hurricane—to quantify the fluctuation in animals after one of these events, and therefore determine the extent to which an animal care facility is impacted. We were particularly interested in fire and hurricanes because of their widespread impact. As our data are at a county level and not a facility level, using a disaster with a geographically large impact increases the chances that a facility was impacted by a disaster reported by the county. Hurricanes were also of interest due to a high prevalence of literature addressing companion animal welfare post-hurricane. Findings are interpreted with respect to the expected changes that animal care facilities experience due to natural fluctuations in cat and dog populations outside of disaster events.

Methods

Data from the Federal Emergency Management Agency (FEMA) and Shelter Animals Count (SAC) were used in a repeated cross-sectional study design. The FEMA dataset is publicly available on FEMAs website (https://www.fema.gov/about/openfema/data-sets) and includes month, year, state, county, and classification for reported disaster events from 1953–2020. Between 2013 and 2018 the FEMA dataset documents 272 fires, resulting in 702 counties requesting federal aid due to fire, and 72 hurricanes, resulting in 1,015 total counties requesting federal aid due to a hurricane.

SAC is an independent, non-profit organization which has been compiling voluntarily shared animal shelter and rescue data since 2011 [23]. A portion of these data are publicly available from SAC (https://www.shelteranimalscount.org/data-dashboards/), however, the entire dataset was purchased from the organization in 2019 and is not published due to SAC’s data use agreement [23]. This dataset includes animal intake and outtake counts by species (dog or cat), age, organization type, location, year, and month of each record (N = 120,852). Each entry includes the signalment (e.g., young, old, or unknown age), number of stray animals collected by the facility, numbers of animals adopted, euthanized, or transferred between facilities, as well as beginning and end of month animal totals. SAC classifies organizations as Animal Rescue with a Government Contract, Animal Shelter with a Government Contract, Rescue without a Government Contract, Shelter without a Government Contract, and Government Animal Services, which we categorized into: (1) government contract (GC) by combining both shelters and animal rescues which had government contracts, (2) non-government contract (NGC) again by combining shelters and animal rescues, and (3) government animal services (GAS). We maintained classification of GC, NGC, and GAS organizations due to expected discrepancies in euthanasia rates. Government sponsored facilities could include animal control, resulting in increased euthanasia. As part of our analysis involved comparing euthanasia rates across different time periods, analysis for GC, NGC, and GAS organizations were kept separate. We use the term “Animal Care Facilities” to include animal shelters, which are typically government funded, and animal rescues, which are generally donation funded and rarely euthanize.

Data were restricted to the years 2013–2018 for the most robust dataset. Data prior to 2013 were excluded as fewer facilities were providing data in these years. By purchase date, the dataset extended through the first half of 2019, so the analysis was restricted to full year data. Data since 2019 and through the COVID-19 pandemic are not included as adoptions increased about 250% during that period [24] and would bias the association with hurricane and fire we sought to assess. Only organizations with data on dogs and/or cats were included in the analysis (N = 109,481).

SAC and FEMA data were merged based on year, month, and county. To verify the data, we compared reported beginning and end number of animals to the calculated values by summing reported intake, adoptions, returns to owner and euthanasia. If the difference between the reported and calculated totals was greater than five, the observation was considered unreliable and excluded from analysis (N = 85,517 for dogs; N = 72,079 for cats). Only one randomly selected observation was kept when duplicates were observed in the dataset, which might have occurred if an organization entered information multiple times in one month. The final dataset used for analysis included 100,656 observations, with 83,910 observations for dogs and 70,538 observations for cats.

The organizations were compared when they reported data for both time points (month of and either year or month after the event) and if they reported from a county that experienced a hurricane or fire based on the FEMA database. Comparing the month of an event with the month after an event allowed us to evaluate short-term impacts of the disaster, while comparing the month of an event with the same month one year later allowed us to evaluate longer-term impacts of the disaster. We independently evaluated organizations in counties that experienced fires and hurricanes. Matched observations were the two time points for animal shelter or rescue organizations that had experienced a disaster event in a specific year and month; the data reported the month when the event occurred were considered cases. The subsequent month or the same month the following year was treated as the control for each animal shelter or rescue organization, provided there was not an additional disaster reported in the subsequent month.

For each organization we compared two metrics: a) the total number of animals in the facility at the end of the month, and b) the total number of animals euthanized. While the total end number of animals is influenced by an organization’s capacity, the statistic we used for comparisons uses each organization as its own control and thus accounts for size of facility. The total end number of animals is the metric that facilities use and provides an important metric of changes in capacity demand (i.e., how many animals need housing) for animal care facilities. Second, we evaluated the monthly total number of euthanized animals. The number of animals euthanized provides another context to the total number of animals in a facility, for example, a facility might report no change in end numbers between time points if they were at capacity, but this might have led to more animals being euthanized. Animal shelter and rescue organizations in the dataset with no records of euthanizing animals were considered no-kill shelters and were excluded from analyses comparing euthanasia numbers. There were 17,982 observations of potential no-kill shelters for dogs that were excluded and 11,109 observations for cats.

The non-parametric paired Wilcoxon Signed Rank test was selected for analysis because the data were not normally distributed, and to allow each organization to be compared to itself at the two different time points. For each comparison, a positive value indicates that more animals were in the facility at the end of the month or euthanized in the month of the disaster compared to subsequent month or year. Analyses were stratified by animal type (dog or cat), organization type (government contract, non-governmental contract, or government animal services), and completed independently for each disaster type (hurricane or fire).

All data cleaning and analysis steps were performed in STATA 16.1 (College Station, TX). We report significance at the p<0.05, <0.01 and <0.001 levels.

Results

Dogs—Hurricane comparison

GAS organizations in counties with a FEMA documented hurricane tended toward fewer dogs in the month of the event compared with the numbers of dogs they reported the month and year after (Fig 1). On average, more GAS reported end numbers lower in the month of a hurricane (median: 70 dogs) compared to the month after a hurricane (median: 74 dogs; Z Score = 2.69; p<0.05; Fig 1). For the end number of animals, GAS organizations also reported significantly fewer animals in the month of a hurricane (median: 49 dogs) compared to the same month one year after a hurricane event (median: 95 dogs; Z Score = 2.11; p<0.05; Fig 1). Note: Figs 1–4 show the number of facilities that have demonstrated a median positive, negative, or no change in the end number of animals or the end number of euthanasias from a month with a disaster event to the month or year after a disaster event. The median end numbers of animal counts are in S1 Table.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Aggregated difference in median of reported end number of dogs during and after disaster.

NCG = non-government contract organization; GC = government contract organization; GAS = government animal services, N = total number of organizations with data for analysis. Statistical significance is noted with *p-value<0.05; **p-value<0.01; ***p-value<0.001.

https://doi.org/10.1371/journal.pclm.0000066.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Aggregated difference in median of reported number of dog euthanasia’s performed during and after disaster.

NCG = non-government contract organization; GC = government contract organization; GAS = government animal services, N = total number of organizations with data for analysis. Statistical significance is noted with *p-value<0.05; **p-value<0.01; ***p-value<0.001.

https://doi.org/10.1371/journal.pclm.0000066.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Aggregated difference in median of reported end number of cats during and after disaster.

NCG = non-government contract organization; GC = government contract organization; GAS = government animal services, N = total number of organizations with data for analysis. Statistical significance is noted with *p-value<0.05; **p-value<0.01; ***p-value<0.001.

https://doi.org/10.1371/journal.pclm.0000066.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Aggregated difference in median of reported cat euthanasia’s performed during and after disaster.

NCG = non-government contract organization; GC = government contract organization; GAS = government animal services, N = total number of organizations with data for analysis. Statistical significance is noted with *p-value<0.05; **p-value<0.01; ***p-value<0.001.

https://doi.org/10.1371/journal.pclm.0000066.g004

Most of the organizations show no change in the number of dogs euthanized during a month with a hurricane event compared with the subsequent month or the same month one year later (Fig 2). Among NGC organizations that did report a change, more organizations report euthanizing fewer dogs the year after a hurricane compared to euthanasia’s in the month of a hurricane though the difference (Z Score = -1.73; p<0.05; Fig 2). A similar trend was observed among GC organizations, with more organizations reporting fewer dogs being euthanized in the month of a hurricane (median: 4 dogs euthanized) compared to the month after a hurricane (median: 6 dogs euthanized; Z Score = -1.84; p<0.1; Fig 2).

Discussion

We found significant changes in the demand for animal care services in both the immediate and long-term aftermath of the disaster events we evaluated. These changes vary based across cats and dogs, hurricanes and fires, and the type of animal care facility. Variability in demand for animal care services after disaster events may be explained by animal behavior, the type and extent of damage that a disaster incurs, and the specific services offered by each type of animal care facility.

For dogs, we see an increased demand for housing within animal care facilities in the time periods following hurricanes, as evidenced in the increase in GAS facilities reporting higher end number of animals (a median increase of 4 the month after and almost 40 more dogs when comparing facilities a year later). Canines show an opposite trend within GAS organizations post-fire, although results are not statistically significant, indicating that fires and hurricanes may affect dog interaction with animal care facilities differently. The increases in dog intakes observed in the month and year following a hurricane may be due to the large influx of stray animals that has been noted post-disaster, resulting in an increased need for animal housing [12]. During a disaster, animals that are already stray tend to emerge as their usual food sources have been altered, and with the shift in human population, many animals also become stray [12]. This increase in dogs may also be due to pet-owners temporarily leaving their companion animals in the care of an animal shelter or rescue while experiencing the impacts of a disaster [25] as it is more often animals being housed in-shelter, rather than at a personal residence, that are more likely to become loose [12].

In our analysis, fires and hurricanes do not result in changes in the reported number of dogs euthanized at animal care facilities in affected counties. However, this could be indicative of a greater demand on the facilities outside of the disaster area where animals are relocated or the surge capacity of “pop-up” animal disaster response. Even the capacity for response can be driven by the disaster itself, be it a hurricane which may be observed, tracked and, to an extent, planned for, versus a fire, which are less predictable and typically rapid onset. For example, Louisiana, one of the states most affected by hurricanes [26], lists household pet emergency planning guides on their Department of Agriculture and Forestry website (https://www.ldaf.state.la.us/animal-health/emergency-programs/) [27] and their Louisiana State Animal Response Team (LSART) [28] works within the state/federal system to respond during disaster events. Many animal care facilities, depending on the extent of damage by the event, will host reunification events after a disaster to reunite lost companion animals with their owners, which may also explain fewer euthanasia when more animals are returned to their owners [29].

The effect of disaster events on cats in animal shelters and rescues differs from that on dogs, with more GC and NGC organizations reporting more cats in the month of a hurricane compared with the month after a hurricane. While fire in the county had limited effect on dog occupancy across facility type, more organizations reported higher numbers of cats the month of a fire compared with the month after. Perhaps related to the higher occupancy, more organizations also reported higher cat euthanasia rates the month and year after a fire, though responders and animal care facilities are averse to euthanizing any animal whose owner might be searching for it. Additionally, animals brought into a shelter post-disaster may have sustained severe enough of injuries to warrant humane euthanasia, this may be particularly relevant for fire-related injury. At this point, services such as search and rescue to locate lost animals, and vaccination and other mitigation practices to reduce disease spread become more essential [10, 11, 30].

Our findings suggest companion animals and the facilities that house them are greatly affected by the impacts of climate change, as evidenced by the shifts in demand on animal housing and euthanasia in the periods following a disaster. Results across other bodies of scientific literature indicate the importance of companion animal welfare post-disaster, and the negative public health outcomes that can occur when companion animals are not accounted for. Additionally, adoption of policies that account for and protect companion animals during disaster events may shift animal ownership from being a risk factor to being a protective factor [10, 31]. Companion animals could provide an incentive to their owners to follow evacuation procedures as owners are inclined to keep their pets safe. Unfortunately, although there are indications that disaster response plans are in place for companion animals, there is little evidence that animals are being integrated into community climate adaptation.

Currently, several U.S. communities are developing climate adaptation plans to better predict and prepare for the effects of climate change. To facilitate this, the Centers for Disease Control and Prevention (CDC) developed the Building Resilience Against Climate Effects (BRACE) framework [31]. The BRACE framework consists of five sequential steps, including (1) “forecasting climate impacts and assessing vulnerabilities”, (2) “projecting the disease burden”, (3) “assessing public health interventions”, (4) “developing and implementing a climate and health adaptation plan”, and (5) “evaluating impact and improving quality of activities” [31]. Across the U.S., states and cities have been funded by the CDC’s Climate-Ready States and Climates Initiative (CRSCI) to locally implement the BRACE framework to combat the effects of climate change [32]. BRACE projects across the nation address multiple climate hazards, and while some projects address mental health, none were located that include companion animals as part of the human adaptation planning [33, 34]. We suggest that companion animals provide a climate hazard when not properly accounted for, due to the negative mental and physical health outcomes that can occur in pet owners, as well as the injuries, death, and loss of home that occur in companion animal populations. Therefore, companion animal welfare should be integrated into the BRACE framework. This is particularly relevant in step 4, “developing and implementing a climate and health adaptation plan”, by ensuring that animals are included into that climate and health adaptation plan, as well as in step 5, “evaluating impact and improving quality of activities”, given the aforementioned benefits that people experience from pet ownership.

Due to the relationship between humans, animals, and the environmental impacts of climate change, we suggest that a One Health approach to climate adaptation is key. Ideally, this includes collaboration between animal shelters and disaster management, to offer services such as transportation for people with pets–which can be difficult if a pet owner lacks individual transportation or resources such as a cat carrier–as well as pet-friendly emergency shelters for evacuees, so that evacuation does not necessitate the separation of a pet-owner from their companion animal. Animal shelters might also consider providing resources for pet owners under duress, including food or supplies, as well as information on how to react to different disaster situations. Incorporating the needs of both humans and their companion animals into climate adaptation plans has the potential to improve disaster response among humans, improve mental health outcomes within affected communities, and increase overall community resilience.

This study has several limitations. As a cross-sectional study, these data are limited, and it is difficult to demonstrate a temporal relationship. Additionally, as the SAC data are self-reported by animal care facilities, the data may be incomplete. Although the SAC data used has 83,910 observations for dogs and 70,538 observations for cats, we see small sample sizes with respect to organizations in counties experiencing fires, especially with dogs, so these results are less robust than those for hurricanes. For example, when looking at the sample sizes for dogs, we see an N of 509 dogs for end number monthly comparison with hurricanes, and an N of 283 for monthly comparison with fires. Likewise, when comparing dog end numbers at the yearly level, we see a total N of 308 for the hurricane sample, and a total N of 135 for the fire sample. This smaller sample size associated with fires may be because fires have a smaller geographical impact area than hurricanes, so fewer people are impacted, and because fires tend to have a quick onset time, so dogs which respond to being called are more likely to be successfully evacuated than their cat counterparts which need a carrier to be transported and may need to be retrieved from a hiding spot prior to transportation [35]. Additionally, we merged county observations from FEMA to individual facility observations in the SAC dataset based on which county an organization is located. At this aggregation, we have no way of knowing if a facility within a county was directly impacted by the disaster. This may be especially problematic in western states, where counties are larger, increasing the likelihood that a disaster may have affected only part of a county, but not the part of the county where an animal care facility is located and our analysis assumes that there is no between county effect, e.g., animals sent to the nearest facility even if it is in a different county. Additionally, the FEMA data only includes a disaster if a county requested federal assistance for managing that disaster. Therefore, there may be disasters that have occurred but are not reflected in our data. Lastly, although collaboration between disaster response teams and animal sheltering facilities is occurring, as this is required by the PETS Act to receive FEMA funding, it may not be documented in the literature on how animal care facilities are utilized in climate change adaptation or during disaster events and we are likely underestimating the integration of companion animals in response planning. Some pet owners may utilize options outside of animal shelters and rescues to house their animals and are therefore excluded from this analysis. However, we are familiar with the climate adaptation literature and explicitly considering pets is all but absent.

Companion animals are an integral part of families across the country. Animal shelter and rescue organizations are directly affected by the occurrence of disaster events and are therefore important stakeholders when developing community or state level climate change adaptation plans. Currently, there are organizations including NARSCA (National Animal Rescue & Shelter Coalition), NACA (National Animal Care & Control Association), NASAAEP (National Alliance of State Animal and Agricultural Emergency Programs), and AVMA (American Veterinary Medical Association) that actively partner across federal, state, and local jurisdictions to protect companion animals during disaster events. Although there are extensive disaster management plans in place for human and animal populations, little research has been done on how human and animal disaster planning can be integrated. Given the close bond between people and pets, and the poor health outcomes observed in both human and animal populations when the two are separated, integrating human and animal disaster response is essential to preserving community resilience in the face of climate change. Increasing research on disaster management from the perspective of animal care facilities will inform animal shelters and rescues of specific scenarios that they can prepare for–for example, our results show an increase in housing demand for dogs post-hurricane, whereas further research may demonstrate different needs in response to different disaster situations. This will also lay the foundation for applying a One Health approach to incorporate companion animal welfare not only in human disaster preparedness, but in the greater discussion of climate change adaptation and planning. Our study demonstrates that there is an increased need for dog capacity post hurricane. Further research can illustrate what type of response will be most needed after different disaster scenarios, allowing animal care facilities to be prepared for a variety of situations.

Supporting information

Acknowledgments

We are deeply grateful to the two practicing veterinarians who read early drafts of this manuscript and provided helpful feedback. Their comments helped to shape the manuscript and to ground the manuscript in how veterinary disaster response really happens. In particular, we are grateful to Tim Krone, DVM MPH, our longtime collaborator and friend, who tragically passed while co-authoring this document.

The information contained herein was derived from data supplied by Shelter Animals Count. Shelter Animals County specifically disclaims all responsibility for any analysis, interpretations, conclusions and opinions contained in the information presented.