Hurricane fatalities: the case of Helene

In April 2025, the National Hurricane Center (NHC) released its final report on Hurricane Helene’s impacts, confirming a total of 250 fatalities1. This number makes Helene the second deadliest hurricane to strike the continental United States in the past 50 years2.

The storm began forming on September 22 as a broad low-pressure system over the western Caribbean Sea. On September 25, the NHC officially named it Helene as it moved northward3. Favorable atmospheric conditions facilitated rapid intensification, and by September 26, Helene had reached hurricane status while crossing the warm waters of the Gulf of Mexico. Late that evening, it made landfall in Florida’s Big Bend region as a powerful Category 4 hurricane, with sustained winds of up to 140 mph. Helene’s destructive path brought heavy rainfall, storm surges, and tornadoes, causing severe flooding and widespread damage from the Gulf Coast to the North Carolina mountains4,5.

Hurricane Helene’s high death toll is particularly staggering, as U.S. fatalities from hurricanes have remained relatively low since Hurricane Katrina in 2005, despite the increasing frequency and intensity of such storms (see Supplementary Table S1). The Federal Emergency Management Agency (FEMA)6 designated 230 counties across six states as disaster areas eligible for individual assistance. North Carolina, with 107 confirmed fatalities, was the most severely affected, experiencing catastrophic flooding in its mountainous regions. South Carolina recorded 50 fatalities, primarily due to destructive tornadoes that caused significant damage. In Georgia, 37 deaths were attributed to record-breaking rainfall that led to severe urban flooding. Florida reported 34 fatalities, largely driven by the storm surge that inundated its coastal areas. Tennessee, with 18 fatalities, faced flash flooding and landslides as the storm stalled during its post-tropical phase. Virginia, which recorded three fatalities, was impacted by strong winds and flooding7,8.

This study investigates county-level fatalities across six states caused by Hurricane Helene in relation to key variables, including wind speed, rainfall, and the National Risk Index, a composite measure that integrates expected annual loss, social vulnerability, and community resilience. Mapping and analyzing these fatalities is crucial for identifying underlying drivers and highlighting the need for policymakers to address both physical hazards and community risks. The findings aim to provide valuable insights to guide disaster preparedness, strengthen risk reduction and communication efforts, and mitigate fatalities during future extreme weather events.

Mapping fatalities, physical hazards, and community risk

Data and methods

Guided by the official NHC report, county-level fatality data were compiled and cross-referenced using a combination of official sources, including the Tennessee Emergency Management Agency and the North Carolina Department of Health and Human Services9,10, along with news outlets11,12,13,14,15. For this analysis, a total of 244 fatalities were traced to specific counties, while five fatalities remained unspecified, and one fatality from the state of Indiana was excluded.

Figure 1 presents the spatial distribution of fatalities by county, with Hurricane Helene’s path included for context. Several counties experienced particularly high losses, including Buncombe County in North Carolina, which reported the highest toll with 43 fatalities. In South Carolina, Aiken and Spartanburg County recorded significant fatalities, with 11 and 9 deaths, respectively. Florida’s Pinellas County experienced a high fatality count of 12. Other areas with severe impacts include Unicoi County in Tennessee, with 8 fatalities, and Richmond County in Georgia, with 7 fatalities.

Fig. 1
figure 1

Hurricane Helene’s track, fatalities, and impacts from wind and rainfall.

Full size image

Figure 1 also illustrates the estimated maximum wind gust and rainfall accumulation, offering a spatial comparison of Helene’s physical impacts and fatality counts. Maximum wind gust data was obtained from the National Weather Service, which provided estimated peak wind gust from 1,669 monitoring sites across FEMA region 4. These site-level observations were spatially aggregated to generate county-level estimates. Rainfall data were obtained from the Quantitative Precipitation Forecast (QPF) provided by NOAA’s Weather Prediction Center. The rainfall variable was calculated by averaging the precipitation totals over the first two days following landfall and aggregating the results to the county level.

The National Risk Index is a comprehensive tool developed by FEMA to evaluate the relative risk of natural hazards across the U.S. It includes 18 types of hazards and assesses risk using three key components: expected annual loss, social vulnerability, and community resilience. Expected annual loss quantifies the estimated human and economic impacts from natural hazards, drawing on data related to exposure, historical loss rates, and annualized frequencies. Social vulnerability measures a population’s susceptibility to hazard impacts, incorporating socioeconomic and demographic indicators such as income, age, minority status, and housing quality. Community resilience evaluates a community’s capacity to prepare for, respond to, and recover from disasters, considering factors such as infrastructure robustness, access to healthcare, and governance strength. Risk values are determined by multiplying expected annual loss by a scaling factor that varies across counties based on their social vulnerability and resilience scores16. All scores are normalized and reported as percentiles; for example, a risk score of 90 indicates that a county’s relative risk is higher than 90% of U.S. counties. This study utilized the most recent version (2023) of the county-level index to explore how pre-existing risks are associated with fatalities caused by Hurricane Helene.

The detailed description of each variable used in the analysis and the mean statistics across all states are provided in Supplementary Table S2.

Descriptive statistics

Table 1 examines the statistical differences in variables between counties with and without hurricane fatalities. As shown in Panel A, counties with fatalities experienced significantly higher wind speeds and rainfall amounts. These counties also had significantly higher risk scores (RISK) and expected annual loss (EAL). However, no significant differences were observed in social vulnerability (SOVI) or community resilience (RESL) between the two groups. Given that not all counties were directly impacted by Hurricane Helene, Panel B reports the statistics to FEMA-designated disaster counties to provide a more equitable comparison. Within these counties, the difference in wind speeds between counties with and without fatalities decreased from 13.42 mph for all counties to 6.22 mph, while the difference in rainfall decreased from 2.20 inches to 1.41 inches. In contrast, the difference in RISK increased from 9.13 to 16.95, and in EAL from 8.53 to 17.20, highlighting the increasing significance of pre-existing risk and loss factors in affecting fatality outcomes. Surprisingly, RESL were significantly higher in counties with fatalities, while SOVI displayed no statistically significant difference. This could be explained by the fact that community resilience, which measures recovery capacity and resource availability, often emphasizes long-term recovery rather than immediate life-saving measures17. Overall, with the differences in wind and rainfall now minimal in disaster-designated counties, the data more clearly reveal the role of underlying risk in shaping fatality outcomes.

Table 1 Mean statistics between counties with and without fatalities
Full size table

The analysis of FEMA-designated disaster counties was further disaggregated by state, and the results are presented in Fig. 2. In North and South Carolina, counties with fatalities had higher RISK and EAL compared to those without fatalities. RESL were also significantly higher in counties with fatalities. In Georgia, counties with fatalities showed higher RISK and EAL, though differences in SOVI and RESL were not statistically significant. Florida exhibited the largest differences in RISK and EAL between counties with and without fatalities, while SOVI and RESL remained statistically insignificant. Finally, both Tennessee and Virginia showed no significant differences across any variables, possibly due to lower exposure or smaller sample sizes for counties with fatalities.

Fig. 2
figure 2

Risk index among FEMA-designated counties with and without fatalities, by states.

Full size image

Empirical findings

The final analysis, presented in Table 2, employed multivariate Poisson regression to test the relationship between risk factors and fatalities while controlling for wind and rainfall impacts. Poisson regression is commonly used for modeling count data and estimating the influence of explanatory variables on rare events such as fatalities18,19. Across all states combined, both wind speed and rainfall emerged as significant predictors of fatalities, underscoring the destructive impact of intense precipitation and flooding. At the state level, North Carolina and Georgia showed significant effects from both wind and rainfall, highlighting the their combined influence on the high fatality counts. In South Carolina and Virginia, rainfall played a more dominant role in affecting fatalities. Conversely, in Florida and Tennessee, wind was a significant predictor, suggesting that storm surge or other factors likely played a larger role.

Table 2 Multivariate poisson regression results for the number of fatalities
Full size table

The National Risk Index (RISK) emerged as a significant predictor of fatalities across all states (see Table 2 Model (1)). Higher fatalities were positively associated with higher pre-existing risk factors. At the state level, RISK was significant in North Carolina, South Carolina, Georgia, and Florida, with Florida showing the strongest relationship between RISK and fatalities. EAL was also a significant predictor at the aggregate level and in states such as North Carolina, Florida, and Virginia. SOVI showed no significant impacts in the combined model. In Georgia, however, it was positively associated with fatalities, indicating that socially vulnerable communities experienced higher fatality counts. In contrast, Virginia exhibited a negative relationship, possibly reflecting differences in exposure patterns. RESL also displayed mixed results between these two states. In Georgia, RESL was positively associated with fatalities, suggesting that even counties with higher resilience experienced severe impacts. However, in Virginia, resilience was negatively associated with fatalities, with less resilient counties facing higher fatality counts.

To address potential limitations of the Poisson regression, we employed two alternative regression techniques: Zero-Inflated Negative Binomial regression to model count outcomes with overdispersion and excess zeros, and logistic regression on fatality as a binary outcome, focusing on modeling the probability of fatality occurrence rather than the number of fatalities. Regression results are provided in Supplementary Information S3–S5, and a high degree of consistency was observed, demonstrating the robustness of the findings.

Toward risk-informed preparedness and disaster risk reduction

Hurricane Helene’s devastating toll underscores the profound impact of extreme wind and rainfall in driving hurricane fatalities. Its intense winds caused widespread destruction, while record-breaking rainfall triggered catastrophic flooding, particularly in mountainous regions where rising waters overwhelmed communities and directly led to fatalities. These physical forces not only caused immediate harm but also disrupted evacuation efforts and emergency responses20,21. FEMA’s National Risk Index and Expected Annual Loss proved to be valuable for understanding hurricane fatalities, offering policymakers practical tools for prioritizing resource allocation and preparedness efforts. Although community resilience and social vulnerability showed mixed and sometimes counterintuitive effects, they remain important scaling factors influencing overall risk outcomes. Policymakers should interpret each metrics carefully, integrating them with physical exposure data and local context to design targeted, effective disaster preparedness and interventions.

Future research should consider additional factors to further understand the drivers of hurricane fatalities, including the effectiveness of evacuation policies, early warning systems, and response efforts. Investigating the role of government communication and compliance with evacuation orders could provide crucial insights into human behavior during disasters22. Data on resource access, transportation infrastructure, and shelter availability can help identify obstacles that hinder vulnerable populations from reaching safety23,24. Given that this study focuses on a single hurricane event, future research is needed to test the generalizability of these findings across different hurricanes, geographic regions, and disaster types. Comparative analyses using more recent hurricane events could offer deeper insights into fatality patterns and strengthen the evidence base on the role of pre-existing risk in disaster outcomes.

Importantly, Hurricane Helene’s impacts cannot be separated from the broader context of climate change25,26. Scientific studies revealed that climate change significantly amplified Hurricane Helene’s impacts. Warmer global temperatures have increased the intensity and likelihood of extreme rainfall events, with Helene’s rainfall estimated to be approximately 10% heavier due to climate change27. Most of Helene’s fatalities occurred inland, where the steep terrain, aging infrastructure, and limited experience with hurricanes left communities unprepared, emphasizing the urgent need for climate adaptation measures that extend beyond coastal regions28,29.

Adding to the urgency, Hurricane Milton struck Florida’s west coast less than two weeks after Helene, underscoring the emerging reality of back-to-back catastrophic storms fueled by a warming climate. This pattern reflects a new norm of systemic and compounding disaster risks, demanding a rethinking of how disaster preparedness and response are approached. The evidence presented in this study clearly demonstrates that disaster fatalities are not merely the result of physical hazard exposure but are deeply rooted in underlying community risk conditions. Given this, we conclude with two urgent perspectives to guide future action:

First, on disaster risk communication

The strong link between risk and fatalities highlights the need for significant improvements in how risk is communicated. Policymakers should promote localized, transparent disclosure of county-level and neighborhood-level risk, transforming vague warnings into personalized, actionable messaging. It is essential that residents not only understand that hurricanes are dangerous but also recognize how dangerous their specific community is and what concrete steps they can take to protect themselves.

Second, on disaster risk reduction investment

The concentration of fatalities in high-risk counties underscores the urgent need to allocate mitigation and resilience funding based on pre-existing risk, rather than solely on recent disaster exposure. Federal and state programs should prioritize investments—such as strengthening infrastructure, expanding community shelters, and improving housing resilience—in counties with the highest risk index scores. Supporting risk-informed pre-disaster intervention is critical to reducing future hurricane fatalities at scale.

Reducing disaster fatalities demands more than forecasting the next storm; it requires dismantling the deep-rooted vulnerabilities that continue to put lives at risk. By confronting these risks head-on and investing in stronger, more equitable preparedness efforts, we can build safer, more resilient communities for the future.