Environment/Climate Change
Floods and county-level mortality in the U.S.: A triply robust approach Lingzhi Chu* Lingzhi Chu Joshua L. Warren Erica Spatz Sarah Lowe Yuan Lu Xiaomei Ma Joseph S. Ross Harlan M. Krumholz Kai Chen
Background: With more frequent floods due to climate change and population expansion into high flood hazard areas, it is expected that flood-related human health burdens are increasing. However, current evidence on flood-related mortality is sparse. This study evaluated the medium- to long-term associations between floods and cause-specific mortalities over the entire contiguous U.S.
Methods: We stratified county-level monthly observations into treatment/control groups based on presence/absence of floods in a 12-month timespan. We used a triply robust approach: (1) regressing outcomes on covariates in the control group with inverse probability weighting; (2) estimating counterfactual outcomes in the treatment group had floods not occurred, using models fitted in (1); and (3) in the treatment group, comparing actual and counterfactual observations adjusting for confounders. Monthly death counts came from the CDC National Center for Health Statistics (2000-2020). We assessed average monthly counts of flood days over a lagged period (up to lag 0-11 months; e.g., the exposure at lag 0-7 months was the average monthly count of flood days over the current month and the past 7 months).
Results: Overall, there were 41,983,819 deaths. It was estimated that one flood day was associated with 9.3 [95% CI: 3.7, 15.0] excess all-cause deaths at lag 0-10 months (per 10,000,000 individuals), 3.4 [95% CI: 0.5, 6.3] excess deaths due to diseases of the circulatory system at lag 0-7 months, 2.9 [95% CI: 1.0, 4.7] excess deaths due to diseases of the respiratory system at lag 0-10 months, and 5.9 [95% CI: 3.6, 8.2] excess deaths due to external causes at lag 0-11 months.
Discussion: These findings suggest that floods are associated with excess county-level deaths. Further research is warranted, and a better understanding of the flood-health relationships will facilitate the allocation of healthcare resources in post-flood management.