Regional Disparities in Stroke Hospitalization and Mortality Associated with Long-Term Fine Particulate Matter Exposure: A Nationwide Ecological Study in Spain.

BACKGROUND/

OBJECTIVES

Stroke remains a leading cause of death and disability globally, with considerable regional heterogeneity in incidence and mortality reflecting differences in classical vascular risk factors and environmental determinants. Fine particulate matter (PM2.5) is an established modifiable environmental risk factor, yet its association with regional stroke disparities remains understudied in Mediterranean European contexts. We examined regional differences in stroke hospitalization and mortality and their ecological associations with long-term air-pollution exposure across Spain's 17 autonomous communities from 2013 to 2021.

METHODS

We conducted a nationwide longitudinal ecological study using aggregated stroke indicators from INCLASNS and ambient air-pollution data from MITECO. The dataset comprised 306 sex-specific region-year observations. The primary outcome was the age-standardized hospitalization rate (ASHR); secondary outcomes were CHR, CMR, ASMR, CFR, and IMI. Associations were assessed using correlation analyses, temporal trends, regional comparisons, and multivariable regression. Sex-stratified analyses were considered exploratory because pollutant concentrations were measured at the regional-year level and were not sex-specific.

RESULTS

Mean ASHR was 20.1 ± 5.4 per 10,000 population, with marked regional variation. PM2.5 showed the largest positive ecological correlation with ASMR among all pollutants examined (r = 0.264, p < 0.001), although the correlation remained weak. Overall pollutant-outcome correlations were weak and heterogeneous in direction. Regional variation in stroke burden was not adequately explained by the available air-pollution indicators.

CONCLUSIONS

PM2.5 was weakly associated with ASMR, whereas associations with hospitalization and other pollutants were small and inconsistent. These ecological findings do not support causal or individual-level inference and warrant confirmation using higher-resolution exposure and individual-level data.