Temperature-mortality relationship in North Carolina, USA: Regional and urban-rural differences

Health disparities exist between urban and rural populations, yet research on rural-urban disparities in temperature-mortality relationships is limited. As inequality in the United States increases, understanding urban-rural and regional differences in the temperature-mortality association is crucial. We examined regional and urban-rural differences of the temperature-mortality association in North Carolina (NC), USA, and investigated potential effect modifiers. We applied time-series models allowing nonlinear temperature-mortality associations for 17 years (2000–2016) to generate heat and cold county-specific estimates. We used second-stage analysis to quantify the overall effects. We also explored potential effect modifiers (e.g. social associations, greenness) using stratified analysis. The analysis considered relative effects (comparing risks at 99th to 90th temperature percentiles based on county-specific temperature distributions for heat, and 1st to 10th percentiles for cold) and absolute effects (comparing risks at specific temperatures). We found null effects for heat-related mortality (relative effect: 1.001 (95% CI: 0.995–1.007)). Overall cold-mortality risk for relative effects was 1.019 (1.015–1.023). All three regions had statistically significant cold-related mortality risks for relative and absolute effects (relative effect: 1.019 (1.010–1.027) for Coastal Plains, 1.021 (1.015–1.027) for Piedmont, 1.014 (1.006–1.023) for Mountains). The heat mortality risk was not statistically significant, whereas the cold mortality risk was statistically significant, showing higher cold-mortality risks in urban areas than rural areas (relative effect for heat: 1.006 (0.997–1.016) for urban, 1.002 (0.988–1.017) for rural areas; relative effect for cold: 1.023 (1.017–1.030) for urban, 1.012 (1.001–1.023) for rural areas). Findings are suggestive of higher relative cold risks in counties with the less social association, higher population density, less green-space, higher PM 2.5, lower education level, higher residential segregation, higher income inequality, and higher income (e.g., Ratio of Relative Risks 1.72 (0.68, 4.35) comparing low to high education). Results indicate cold-mortality risks in NC, with potential differences by regional, urban-rural areas, and community characteristics. • Results suggest potential regional and urban-rural disparities for high and low temperature-mortality risks in North Carolina, USA. • The Piedmont region had the highest cold-mortality risks, whereas the Mountain region had the lowest cold mortality risks. • Although not statistically different from rural temperature-mortality risks, urban areas had higher estimated mortality risks associated with both heat and cold. • Findings suggest that high social association and high NDVI value were associated with lower heat relative effects and that low population density, low PM 2.5 , higher education level, lower residential segregation, lower income inequality, and high income were associated with a low cold relative effect.