Trends in FEMA-Declared Weather-Related Disasters (1989-2019): Implications for Healthcare, Health Equity, and Extreme Weather Resilience

1. Introduction

As extreme weather events continue to increase in both frequency and severity, the resulting challenges to health system preparedness, health equity, and community resilience have become increasingly apparent. The National Oceanographic and Atmospheric Administration’s (NOAA) National Centers for Environmental Information (NCEI) has documented the escalating incidence of billion-dollar weather and climate disasters across the United States, highlighting the substantial economic losses associated with these events [1]. Beyond economic impacts, these disasters pose profound threats to population health and the functionality of the healthcare system. Exposure to extreme weather events is associated with elevated risks of cardiovascular, respiratory, and mental health conditions, which in turn lead to increased healthcare utilization and further strain on an already overburdened U.S. healthcare system [2,3]. Vulnerable populations, including older adults, individuals with chronic health conditions, low-income communities, and socially marginalized groups, are disproportionately affected, exacerbating existing health inequities [4]. These trends underscore the urgent need for comprehensive, evidence-based strategies to protect population health, enhance adaptive capacity, and strengthen the resilience of health systems. Interventions may include climate-informed clinical practices, trauma-responsive care, and cross-sector collaborations that integrate disaster preparedness, health equity, and community resilience into both public health policy and clinical practice frameworks. Without proactive planning and targeted investment, the cumulative burden of extreme weather events is likely to escalate, amplifying health disparities and challenging the sustainability of healthcare delivery in the United States [2,4]. These challenges also have direct implications for nursing education, as nurses must be prepared to respond to increasingly complex, climate-driven public health emergencies across diverse care settings [5].

As the largest group of health professionals in the U.S., nurses are inevitably involved with some aspect of disasters and public health emergency preparedness through assisting with mitigation, preparedness, response and/or recovery efforts[6]. However, many nurses are not aware of the risks in their own community. In a study designed to ascertain the current state of nursing education for a nuclear or radiation event and perceived risk of this type of event, Veenema et al. (2019) [7] found that 75% of survey respondents (N=679) taught either none or less than one hour of related content. Additionally, 53% (N=228) did not know their school was in a designated emergency planning zone (designated by the U.S. Nuclear Regulatory Commission for planning purposes) [7].

Despite nursing education[8], and professional code of ethics[9] related to disaster preparedness and duty to respond, and health care facility accreditation requirements for preparedness, there are still gaps in nursing education and workforce training for preparedness. The American Association of Colleges of Nursing, the professional organization that provides standards on baccalaureate and graduate nurse education and competencies includes competencies for preparedness under the domain of population health[8] (pp. 35-36). The American Nurses Association’s issue brief on nurses and disasters[9] describes nurse’s ethical duty to respond during a disaster and other considerations including legal and moral. The Joint Commission’s National Performance Goal #3 [10] describes the standards for hospitals to meet accreditation requirements. Identifying local and regional risks are an integral part of helping to ensure the healthcare workforce and systems are prepared.

The Federal Emergency Management Agency (FEMA), established in 1979, was created to centralize federal disaster response and civil defense programs that had previously operated in a fragmented manner. FEMA’s formation represented a major step toward improving coordination, efficiency, and accountability in federal disaster response efforts. The agency’s role was further formalized and significantly expanded by the Robert T. Stafford Disaster Relief and Emergency Assistance Act of 1988 (Stafford Act), which established a standardized legal framework for federal disaster declarations, emergency assistance programs, and hazard mitigation initiatives [11]. This legislation provided FEMA with the authority to mobilize resources rapidly, which is requested by a state governor or tribal leader when an event surpasses local and state response capabilities to coordinate across federal, state, and local levels, and implement recovery programs to support communities affected by disasters [6]. More importantly, the Stafford Act also enabled the systematic collection of disaster data, creating a rich longitudinal resource that allows for analysis of trends, patterns, and regional variation in disaster impacts. By linking these operational capabilities to evidence-based preparedness strategies, FEMA has become a cornerstone of the national disaster response infrastructure, shaping how communities, health systems, and policymakers approach risk reduction and resilience planning [12].

Despite recognition of the growing disaster burden, prior research has largely examined disaster frequency, hazard trends, or health outcomes in isolation [13,14]. Few studies have systematically integrated longitudinal disaster trend analysis with healthcare system preparedness, workforce capacity, and social vulnerability, particularly in ways that inform nursing education and workforce development [15]. This gap limits understanding of how evolving disaster patterns influence healthcare delivery, resource allocation, and population health outcomes, particularly among socially vulnerable populations. Communities with limited economic resources, reduced access to healthcare, higher baseline health risks, or pre-existing infrastructure deficits often face disproportionate health burdens during disaster events. These inequities can exacerbate morbidity, delay recovery, and place additional stress on healthcare systems, creating cycles of vulnerability that are amplified with successive disasters. Moreover, the health impacts of disasters extend beyond physical injury to include mental health challenges, chronic disease exacerbation, and trauma-related outcomes, underscoring the importance of integrating trauma-responsive and climate-informed approaches into healthcare planning and disaster preparedness initiatives.

Recent evidence demonstrates that climate-related disasters substantially disrupt healthcare delivery, workforce capacity, and patient outcomes. A mixed-methods study of Federally Qualified Health Centers (FQHCs) in Puerto Rico and the U.S. Virgin Islands and a systematic review of 46 studies on climate disasters and oncology care report consistent impacts across settings, including facility closures, supply chain disruptions, infrastructure failures, and reduced patient access [16,17]. Workforce preparedness remains limited, with many healthcare workers reporting low confidence and insufficient disaster training. Both studies also highlight the dual burden on clinicians, who must manage personal disaster-related impacts while continuing to provide care. Adverse effects are disproportionately experienced by vulnerable populations, including individuals with chronic conditions and cancer, exacerbating existing health inequities through treatment interruptions and reduced access to services. Despite these insights, current literature largely focuses on downstream healthcare impacts and does not systematically link temporal and geographic disaster patterns to healthcare system capacity, underscoring the need for integrated, population-level analyses.

To address these gaps, this study examines FEMA disaster data from 1989 to 2019, a period reflecting the agency’s modern operational and legislative framework under the Stafford Act. Data after 2019 are excluded to avoid confounding effects introduced by the COVID-19 pandemic, which caused unprecedented disruptions in disaster declarations, resource allocation, and healthcare system demands [18,19]. Using descriptive statistics, generalized linear mixed models, and clustering techniques, this study quantifies temporal and geographic patterns in disaster declarations, evaluates variation across hazard types and regions. The study examines implications for healthcare system capacity, workforce preparedness, and health equity to inform nursing workforce education in the context of extreme weather-related disasters. Future integration of disaster trend analysis with considerations of healthcare readiness and social vulnerability will provide actionable insights for workforce planning, infrastructure investment, policy development, and clinical preparedness strategies. Ultimately, the findings from this phase of the study aims to inform the development of healthcare systems that are resilient, adaptable, and capable of delivering equitable care in the face of increasing disaster frequency, climate change, and evolving public health challenges.

2. Materials and Methods

2.1. Data Source

Data on federally declared disasters from 1989 through 2019 were obtained from FEMA. The analysis included both Major Disaster Declarations (DR) and Emergency Declarations (EM). A Major Disaster Declaration provides a mechanism for comprehensive, long-term federal recovery programs and financial aid for rebuilding infrastructure and assisting individuals after severe damage has occurred. An Emergency Declaration provides immediate, short-term assistance to protect life, property, and public health before or during a crisis [20]. FEMA assigns each declaration a disaster number and declaration type, and declarations were issued for specific states. Therefore, a single physical disaster event affecting multiple states is represented by separate FEMA declaration identifiers for each affected state. In this study, the unit of disaster counting was the state-level FEMA declaration, rather than the underlying physical disaster event. Accordingly, multi-state events were counted separately for each affected state. Besides, disaster occurrences were analyzed by calendar year, with each declaration assigned to the year in which it was issued. Events spanning multiple calendar years were counted only once, in the year of declaration. For this study, weather-related disasters were defined as natural hazard events attributable to extreme or anomalous meteorological conditions that resulted in substantial loss of life, property damage, or community disruption and promoted a FEMA declaration. Disasters classified as weather-related included coastal storms, wildfires, floods, freezing events, hurricanes, landslides, severe ice storms, severe convective storms, snowstorms, tornadoes, tsunamis, and droughts.

2.2. Statistical Methods

National trends in annual weather-related disaster occurrences from 1989 to 2019 were first described through visual inspection. Subsequently, temporal trends in disaster occurrences across the 51 US states (including the District of Columbia) were evaluated using generalized linear mixed-effect models (GLMMs). Since the likelihood of weather-related disasters varies geographically, and disaster management practices differ by state, state-level disaster counts were utilized as the outcome. Time was the main predictor with random intercepts and random slopes included to account for state-level variation in baseline disaster frequency and temporal trajectories, respectively, and the presence of an excess of zero counts was accommodated. Model specification was guided by assessment of over-dispersion and the linearity of temporal trends. Over-dispersion was evaluated using the Pearson chi-square dispersion statistic, calculated as the sum of squared Pearson residuals divided by the residual degrees of freedom (df). Linearity was explored descriptively through bar plots and formally assessed using likelihood ratio tests (LRTs). To capture potential nonlinear temporal patterns, various GLMMs were compared using: (i) a linear year term, (ii) a basis spline for year (df = 3), that is, a smooth curve that allows the trend to bend slightly over time, and (iii) a combined linear and spline specification. Model fit was evaluated using the Akaike Information Criterion (AIC), the Bayesian Information Criterion (BIC), and LRTs, with P < 0.05 indicating significant improvement in model fit. Final model selection was based on convergence behavior, LRT results, and information criteria, with simpler models favored when added complexity did not materially improve fit.

Finally, hierarchical clustering was applied to model-derived estimates to identify groups of states exhibiting similar temporal patterns in disaster occurrence. Clustering was conducted using Euclidean distance, which considers both the overall magnitude and the shape of the temporal trajectories, and Ward’s minimum variance method [21,22], which tends to form compact and homogeneous clusters. The optimal number of clusters was determined by visually inspecting dendrograms and selecting the cut point that balanced between-cluster separation and within-cluster similarity.

All statistical analyses were performed using R 4.6.0 (R Core Team, Vienna Austria) and the “glmmTMB” (Version 1.1.14) package.

3. Results

3.1. National Trends in Weather-Related Disaster Declarations

Annual statewide disaster counts by type from 1989 to 2019 are illustrated in Figure 1, where it is evident that the counts fluctuated considerably from 1989 to 2019. In particular, the total number of weather-related disasters increased from 30 in 1989 to 65 in 2019, with notable peaks in 2005 (n=116) and 2011 (n=122). Severe storm occurrences rose from 3 in 1989 to 26 in 2019, peaking between 2007 and 2011. Flood occurrences followed a distinct V-shaped trajectory, with a 10-year low between 1999 and 2009.

Further, Figure S1 (Supplementary Information) illustrates the percent age of US states that experienced weather-related disasters categorized by annual count and type per year. Each disaster is counted once per state, regard less of the number of counties designated within that state.

To reduce short-term annual variability and better visualize broader temporal patterns, the annual statewide disaster counts were further aggregated into five-year intervals, revealing simplified and visually distinct nonlinear trajectories in comparison to the annual counts (Figure S2, Supplementary Information). Data from 1989 were excluded from this analysis to maintain equal interval lengths. In the figure, total weather-related disaster counts peaked in 2005-2009 (n=399), followed by a gradual decline, though remaining higher than levels observed before 2005. Fire counts increased sharply from 1995 to 2004, dipped in 2005-2009, and then gradually rose again in the two most recent periods. Flood counts displayed a pronounced V-shaped pattern, falling from 1990-1994 to 2005-2009, then rebounding markedly in 2010-2014 and 2015-2019. Severe storm counts rose steadily to a maximum in 2000-2004 (n=244) before decreasing in later intervals with the most recent period still exceeding the first. Tornado were most frequent in 1990-1994 (n=16), dropped to a low in 2005-2009 (n=3), and partially recovered thereafter (n=8 in 2015-2019). Hurricane counts fluctuated, with a low in 2000-2004 (n=26) and a peak in 2005-2009 (n=74) before a modest decline.

3.2. State-Level Trends in Weather-Related Disaster Occurrences

After evaluating model fit based on over-dispersion parameters (Table S1, Supplementary Information), the high proportion of zero counts (Figure S1, Supplementary Information), linearity, and model convergence (Table S2, Supplementary Information) for each disaster type, specific models were selected. A zero-inflated negative binomial (ZINB) model with a nonlinear spline term was employed for hurricanes; zero-inflated Poisson (ZIP) models with nonlinear spline terms were selected for total weather-related disasters, severe storms, and floods; and ZIP models with linear terms were utilized for fires and tornadoes.

Based on these model specifications, distinct temporal patterns across disaster types were revealed by the ZIP and ZINB mixed-effects models (Table 1) and the modeled state-specific temporal trajectories were shown in Figure 2.

In particular, in Table 1, while states exhibited significant heterogeneity in baseline disaster frequencies (indicated by substantial random intercept variance), the temporal trajectories were remarkably consistent across states. Zero-inflation was generally low for most disaster types, indicating few structural zeros, except for hurricanes, where the probability of structural zeros was notably higher.

In Figure 2, fires and tornadoes showed no significant linear trends over the 1989–2019 period (β = –0.03, 95% Confidence Interval (CI) [-0.11,0.05]and β = -0.03, 95% CI [-0.07,0.01]; P = 0.417 and P = 0.257, respectively), while floods, severe storms, hurricanes, and all weather-related disasters combined exhibited significant nonlinear temporal trends, with multiple spline terms reaching statistical significance (P<0.05). The overall shapes of trajectories, derived from the combined contribution of all spline basis functions and their coefficients, revealed complex non-monotonic dynamics. For floods, two negative and highly significant spline terms (β = –2.09, 95% CI [-3.23, -0.95] and β =–1.98, 95% CI [-2.80, -1.16]; P = 0.0003 and P <0.0001, respectively) indicated an initial decline followed by a rebound. Severe storms, hurricanes, and total disasters all showed significant nonlinear effects (all P < 0.05). Although spline coefficients were positive, differences in their magnitude led to more complex and non-monotonic trajectories, with disaster counts rising sharply from the early 1990s to the late 2000s before showing a modest decline after approximately 2005–2010. To present these temporal trends on an interpretable scale, weather-related disasters rose steadily from the early 1990s, increasing from about 0.3–1.2 to roughly 0.5–2.5 around 2010, before tapering to approximately 0.5–1.8 by 2020. Thus, the increase through the 2000s was substantial, followed by a moderate decline. Fire counts generally trended upward, although the magnitude of change varied. For most groups, increases were modest (on the order of 0.05–0.2 over the full period), whereas one trajectory rose more sharply from about 0.3 to over 1.2. Floods showed a distinct pattern, declining from around 0.3–0.8 in 1990 to a low near 0.1–0.2 in the early 2000s, then rebounding to roughly 0.2–1.25 by 2020; in many cases, the later increase offset the earlier decline. Severe storms exhibited one of the largest changes, increasing from about 0.05–0.3 to approximately 0.2–1.3 by the late 2000s, followed by a decline to roughly 0.2–1.0 by 2020. Tornado counts declined over time, from about 0.03–0.20 in 1990 to approximately 0.02–0.04 by 2020, with no evident rebound. Hurricanes showed a similar pattern to severe storms, increasing from 0-0.15 to around 0.1–1.6 by about 2010, then declining to roughly 0.1–0.9 by 2020.

3.3. Geographic Clustering of US States Based on Temporal Disaster Trends

Hierarchical clustering was applied to the model predicted disaster occurrence trajectories for floods, severe storms, hurricanes, and all weather-related disasters combined. Predictions were obtained from the final ZIP or ZINB models described above. The optimal number of clusters was determined by visually inspecting the dendrograms and selecting the cut point that balanced between-cluster separation and within-cluster similarity (Figures S3, Supplementary Information). Distinct temporal pat-terns of the four disaster types are illustrated in Figures S4–S7 (Supplementary Information) and their geographic distributions are mapped in Figure 3.

4. Discussion

This study examined FEMA emergency and major disaster declarations for weather-related hazards across U.S. states from 1989–2019, identifying important temporal, hazard-specific, and geographic patterns with direct relevance to public health preparedness and the extreme weather-related readiness of healthcare systems. Rather than a simple monotonic increase, the national pattern was nonlinear, characterized by substantial interannual variability and pronounced peaks (e.g., mid-2000s and early 2010s), followed by a decline that nonetheless remained elevated relative to early study years. These findings reinforce that the U.S. disaster environment has shifted toward more frequent and complex weather-related emergencies over the past three decades, with implications for surge capacity, continuity of care, and longer-term recovery needs.

4.1. Interpreting National and Hazard-Specific Trends

Across the study period, overall weather-related declarations and severe storms demonstrated statistically significant upward trends in annual state-level counts, while tornado-related declarations declined in the national annual analyses. When examined in five-year intervals and with mixed-effects models allowing for nonlinearity, several hazard categories, particularly floods, severe storms, hurricanes, and total weather-related disasters, showed clear nonlinear trajectories. These patterns likely reflect a combination of changing hazard exposure, climate variability, land use and development, and differences in mitigation, preparedness investments, and response capacity. From a public health standpoint, the key message is that health systems must plan not only for more events, but for shifting mixes of hazards with different clinical and operational consequences (e.g., injury and displacement, interruptions in dialysis and oxygen supply, medication access, heat-related illness, respiratory impacts from wildfire smoke, and post-event mental health burden).

4.2. Geographic Heterogeneity and Implications for Preparedness

State-level modeling highlighted substantial between-state variation in baseline disaster declaration frequency (random intercept variance), consistent with geographic differences in hazard exposure and vulnerability. The clustering analyses further suggest that states can be grouped by similar temporal trajectories, creating an opportunity for regionally tailored preparedness strategies, mutual aid planning, and shared training approaches. For health systems, this means preparedness should be calibrated to local and regional risk profiles rather than relying on a uniform national planning assumption. Workforce models, supply chain planning, evacuation and shelter coordination, and continuity-of-operations planning should be informed by the hazards most likely to affect a given state cluster and by expected changes in frequency over time.

4.3. Health Equity: Disaster Trends AS a Multiplier of Vulnerability

Although this study does not directly measure social vulnerability, the observed disaster patterns have important implications for populations known to be at higher risk. A consistent implication of these findings is that the health impacts of disasters are not evenly distributed. Socially vulnerable communities, those with fewer financial resources, constrained transportation, limited access to healthcare, higher baseline chronic disease burden, and under-resourced infrastructure, face disproportionate exposure and barriers to response and recovery. Repeated or closely spaced disasters can compound risk through housing instability, interruption of routine care, delayed access to medications and medical devices, and cumulative stress and trauma. Integrating social vulnerability considerations into hazard planning is therefore essential: disaster trend analyses should be paired with equity-focused planning tools to prioritize high-risk communities for mitigation resources, mobile/alternate care strategies, culturally responsive risk communication, and post-disaster care continuity. For example, populations with limited access to primary care or specialty services may experience worsening of chronic conditions during disaster events, while those in regions with weakened infrastructure may face greater barriers to evacuation, emergency care, and post-disaster support. These disparities contribute to long-term morbidity, mental health challenges, and cumulative health risks, highlighting the critical need for targeted interventions and policies designed to protect high-risk populations. Addressing these inequities requires integrating social vulnerability assessments into disaster planning, prioritizing resource allocation for underserved populations, and implementing strategies that reduce barriers to healthcare access during and after emergencies.

4.4. Implications for Nursing Education and the Public Health Workforce

These results have direct implications for the intersection of nursing education, public health, and climate change. As disasters become more frequent and hazard patterns evolve, nurses increasingly practice in conditions where climate-related emergencies are routine rather than exceptional. Nursing curricula and continuing professional education should explicitly incorporate:

• • Climate-informed and disaster-ready clinical practice, including recognition of disaster-associated exacerbations of chronic disease, environmental exposures (e.g., smoke, heat), and management of care disruptions.
• • Population health and equity competencies, including applying social vulnerability information to triage planning, outreach, resource navigation, and advocacy during recovery.
• • Trauma-responsive approaches for individuals and communities experiencing displacement, loss, and recurrent events.
• • Data literacy and systems thinking, enabling nurses to interpret trend data (such as FEMA declarations and regional clustering), translate it into local preparedness priorities, and participate meaningfully in interagency planning.

Positioning nurses as core partners in climate and disaster preparedness strengthens public health capacity and supports more equitable outcomes during response and recovery. A publicly available dashboard, with hazard risk information by county, would be a foundational component for education, training, mitigating, and preparing in addition to response and recovery.

This is innovative because it will integrate hazard data and data about population vulnerabilities that would affect healthcare workers’ treatment approach or ability to treat and potential education needs that regional and disasters.

4.5. Practice, Policy, and Planning Relevance

At the policy and systems level, these findings support using longitudinal disaster trend information to guide investment in resilient infrastructure, cross-sector coordination, and workforce development. Health system leaders and public health agencies can use trend and cluster outputs to prioritize preparedness investments (e.g., backup power and water planning, supply chain redundancy, staffing surge protocols, and alternate care site planning). Policymakers can use these patterns to support targeted mitigation and recovery resources for high-risk regions and communities, with explicit attention to reducing inequities in exposure and access to care.

4.6. Limitations and Future Directions

Future work should integrate FEMA declaration trends with measures of health outcomes (e.g., hospital utilization, mortality, chronic disease disruptions, mental health indicators), community vulnerability, and health system capacity to clarify pathways from hazard patterns to inequitable health impacts. Evaluating how educational interventions and preparedness training translate into measurable improvements in response performance and recovery outcomes would directly advance the evidence base for nursing education in the context of climate change.

5. Conclusions

Analysis of FEMA weather-related disaster declarations from 1989–2019 demonstrates that the U.S. disaster landscape has evolved over the past three decades, with increasing and nonlinear patterns in overall declarations and severe storms, meaningful variation by hazard type, and substantial geographic heterogeneity across states. These patterns have clear public health consequences, including recurrent strain on healthcare delivery systems and disproportionate impacts on socially vulnerable populations.

The findings underscore the need for a shift from reactive response to proactive, data-informed preparedness that integrates hazard trends with health equity priorities. For the special issue focus, the results also highlight a workforce imperative: nursing education and continuing professional development should incorporate climate-informed disaster preparedness, trauma-responsive care, and equity-centered population health competencies, supported by stronger collaboration between academic nursing, public health agencies, and healthcare systems. Embedding longitudinal disaster trend analysis into preparedness planning and training can help build more resilient and equitable health systems in the face of climate change.