Assessing global factors associated with tropical cyclone-related mortality: A population-based longitudinal study

The underlying factors associated with the substantial global tropical cyclone (TC)-related mortality burden, characterized by its highly variable spatiotemporal patterns, remain unclear. We aimed to identify and assess the key factors associated with TC-related mortality on a global scale.

We collected mortality records from 2034 locations in 68 countries/territories across five continents (2000-2019) to identify and assess the key factors associated with TC-related mortality on a global scale. Bayesian ensemble models were applied to estimate the associated mortality for each TC event in each location. A random forest regression (RFR) model was employed to assess the relative statistical importance of TC and location characteristics, as well as their interactions, regarding the TC-related mortality.

For TC physical characteristics, the TC-associated cumulative rainfall consistently exhibited stronger influence on mortality than TC-induced surge-driven flood depth or maximum sustained windspeed. However, sociodemographic factors, including population traits (e.g., population density, proportion of the population aged ≤ 9 years, percent of the population aged ≥ 65 years) and socioeconomic and infrastructure development (e.g., built-up ratio), showed greater relative importance than the TC physical attributes and accounted for the majority of the explained variations in TC-related mortality. Furthermore, cumulative rainfall interacted strongly with local sociodemographic conditions and was potentially the most important associated factor for disparities in mortality arising from factor interactions.

The findings highlight the critical role of sociodemographic factors in explaining the global spatiotemporal variability of TC-related mortality, surpassing the relative importance of TC intensity. TC-related rainfall could be a key associated physical factor of the global mortality burden.