This study examined how socio-demographic and environmental factors shape population vulnerability to typhoon-related mortality in Japan. Using national mortality data from 2010–2019, the authors aimed to quantify cumulative excess deaths associated with typhoon wind exposure and to assess whether local characteristics—income, natural hazard vulnerability and access to healthcare—modify those effects.
The analysis linked 12.9 million registered deaths across Japan between 2010 and 2019 to measures of typhoon exposure and local context. Individual-level mortality records were provided under license by Japan’s Ministry of Health, Labour and Welfare and are not publicly available because of privacy protections under the Statistics Act. Meteorological data (temperature, humidity) were obtained from the Japan Meteorological Agency, and tropical cyclone wind speed data came from the International Best Track Archive for Climate Stewardship (IBTrACS). Measures of distance to medical facilities were derived from the Housing and Land Survey via official statistics, and estimates of flood- and landslide-prone areas were sourced from the Ministry of Land, Infrastructure, Transport and Tourism. The authors provide analysis code and a dummy dataset on GitHub to support reproducibility.
The investigators constructed a Bayesian spatio-temporal model to link population-weighted typhoon wind exposure to weekly mortality counts. The model incorporated temporal lags to capture short-term and delayed mortality responses following typhoon exposure and explicitly tested effect modification by local indicators: income level, measures of hazard vulnerability (flood- and landslide-prone areas) and access to healthcare (distance to facilities).
Across the study period, typhoon exposure was associated with cumulative excess deaths concentrated among older adults. For adults aged ≥70 years, the model estimated 2,426 cumulative excess deaths with a 95% credible interval of 139 to 4,632. For people younger than 70 years, the cumulative excess mortality estimate was more uncertain (estimated 781 deaths; 95% credible interval: −309 to 1,900), indicating weaker or inconsistent evidence for a national-level increase in this age group.
Temporal patterns differed by age group. Among adults aged ≥70 years, mortality increases were observed within 0–1 weeks after typhoon exposure, consistent with an acute effect of strong winds and immediate post-event hazards. By contrast, for individuals <70 years the observed increases in mortality were delayed and geographically concentrated rather than uniform across the country.
Geographic and socio-demographic context substantially modified the relationship between typhoon exposure and mortality. The association with excess deaths among older adults was stronger in areas with limited healthcare access and high natural hazard vulnerability. For younger individuals, delayed mortality increases were concentrated in lower-income and landslide-prone areas. These patterns indicate that typhoon mortality reflects interactions among the acute physical hazard, demographic ageing, and uneven local adaptive capacity.
Individual-level mortality data cannot be shared publicly due to legal privacy protections; qualified researchers may request access from the Ministry of Health, Labour and Welfare under license. Meteorological and hazard-layer sources cited in the paper are publicly accessible from Japan Meteorological Agency, IBTrACS and MLIT. The authors have made analysis code with a dummy dataset available on GitHub to facilitate reproducibility of the analytical workflow.
The article is a preprint and has not undergone peer review; the authors explicitly note this status and caution against using the findings to guide clinical practice. Specific limitations beyond the preprint status are not detailed in the provided abstract and metadata; any additional methodological caveats reported in the full manuscript or supplementary materials are not reproduced here.
The study’s findings highlight that typhoon-related mortality is not solely a function of hazard intensity but also of local vulnerability—including healthcare access, income and hazard-prone geography. The observed acute mortality among older adults and delayed, location-specific effects among younger people suggest different mechanisms of harm and windows for intervention. Incorporating geographically resolved vulnerability indicators into disaster preparedness and climate-resilient health system planning may reduce excess deaths following tropical cyclones.
The study lists multiple funders, including the Fulbright Foreign Student Program, JSPS KAKENHI, the Wellcome Trust, the US NIH (R00 ES033742, D43 TW012726), NFRF, and an Environment Research and Technology Development Fund administered by Japan’s Ministry of the Environment. The authors declared no competing interests and state that relevant ethical approvals and participant consent procedures were followed. The work is presented as a preprint under a CC-BY-ND 4.0 license.