A large retrospective study of Veterans Health Administration data has found that apparent increases in cardiovascular and respiratory events following Hurricanes Sandy and Harvey were largely explained by preexisting regional health differences — not hurricane exposure itself.
The study analyzed 960,178 veterans in regions affected by Hurricane Sandy and 654,178 in regions affected by Hurricane Harvey, evaluating acute cardiovascular and respiratory events during the year before and the year after each storm. Researchers used Andersen-Gill Cox regression models designed to account for recurrent events and baseline regional differences.
That methodological distinction proved central to the findings. When analyses were restricted to the posthurricane period alone, significant associations between hurricane exposure and cardiopulmonary outcomes appeared. But those associations disappeared once prehurricane event rates were incorporated into the models — indicating that many apparent poststorm differences reflected underlying health disparities that predated the storms entirely. Among healthier veterans in areas later flooded by Hurricane Harvey, for instance, cardiovascular event hazards were already 39% higher before the storm compared with veterans in unexposed regions.
Neighborhood Disadvantage as the Dominant Factor
Across both hurricane cohorts, neighborhood disadvantage emerged as the strongest and most consistent predictor of adverse outcomes. Among healthier veterans in the highest Area Deprivation Index (ADI) quartile, cardiovascular event hazards were 75% higher in the Sandy cohort and 63% higher in the Harvey cohort compared with veterans in the least disadvantaged neighborhoods. Respiratory event hazards rose substantially with worsening neighborhood disadvantage as well.
The two cohorts differed markedly in baseline disadvantage. The mean ADI score was 63 in the Harvey cohort versus 51 in the Sandy cohort, and 35% of Harvey veterans lived in the most disadvantaged neighborhoods compared with 20% of Sandy veterans.
Age, Sex, and Geography
Older age was consistently associated with higher cardiovascular event hazards. In the Harvey cohort, veterans aged 56 to 69 and those 70 or older faced approximately fourfold higher cardiovascular event hazards compared with younger veterans.
Female veterans had lower cardiovascular event hazards than male veterans but higher respiratory event hazards across several analyses. The researchers cautioned that lower cardiovascular risk in women should be interpreted carefully, given prior evidence that cardiovascular disease in women is often underdiagnosed and undertreated. They also noted that women may be more susceptible to asthma and COPD exacerbations tied to environmental exposures and postdisaster mold.
Geography introduced additional complexity. In the Sandy cohort, elevated cardiopulmonary event rates in some public-assistance regions overlapped with rural Appalachian areas known for higher baseline comorbidity burdens. In the Harvey cohort, elevated respiratory risks in some nonflooded regions coincided with the Permian Basin and Eagle Ford Shale areas, where oil and gas extraction has been linked to increased respiratory morbidity and air pollution.
Study Design and Limitations
Researchers linked geocoded residential address data with FEMA exposure maps and ADI scores. Hurricane exposure was categorized as no exposure, public assistance, individual assistance, or flooded with individual assistance. Primary outcomes included acute cardiovascular and respiratory events requiring urgent care, emergency department visits, or hospitalization — among them myocardial infarction, heart failure, stroke, COPD exacerbations, asthma, pneumonia, and acute respiratory failure.
Several limitations may have biased the findings toward underestimating hurricane-related risk. The analysis captured only VHA-delivered care, potentially missing veterans who sought treatment elsewhere during storm-related disruptions — and flooded regions, where VHA access was most likely disrupted, may have had the most missed events. Exposure misclassification was also possible, as some flooded residences may not have met FEMA housing-damage thresholds used to define exposure intensity. All subgroup analyses were exploratory and unadjusted for multiple comparisons.
Implications
"Although changes in hazard ratios specifically attributable to hurricane exposure were not significant, neighborhood disadvantage, advancing age, and sex remained robust factors independently associated with acute health events," wrote lead author Caryn S. Yip, PhD, of the University of Iowa, and colleagues.
The findings point toward a reframing of disaster preparedness — away from reactive, storm-by-storm responses and toward longitudinal strategies that address the social determinants of health and regional environmental hazards shaping vulnerability long before any storm arrives.
Disclosures can be found in the study.
Source: JAMA Network Open
