Geographic Dimensions of Heat-Related Mortality in Seven United States Cities

Extreme heat is the leading cause of weather-related death and illness in the United States. Adaptation and mitigation strategies designed to protect the public would greatly benefit from knowing specifically where heat-related deaths are most likely. This dissertation represents a comprehensive examination of intra-city heat-related mortality.

Over two million records of mortality spanning multiple decades and cities comprise the primary data set. Time series models were employed to relate mortality to temperature after accounting for seasonality and long-term time trends. Threshold temperatures were established for each city associated with significant increases in mortality above typical summertime conditions. The mortality rate within each postal code on threshold-exceeding days was then calculated, quantifying spatial variability in heat-related risk. These intra-city mortality patterns were compared to demographic and environmental variables using multivariate regression. Temporal variability in the spatial mortality patterns was evaluated in an iterative model building and testing framework.

Mortality on hot summer days significantly varied within the study cities, and intra-city variability was far greater than inter-city variability. Characteristics of high-risk zones included more elderly residents, more isolated residents, and more intensely developed land, although the specific set risk factors varied from one city to another. There was evidence that the location of high-risk zones are predictable. A city-specific, health data-driven approach offers an improved strategy for municipalities to identify, understand, and reduce the risks associated with the most deadly natural hazard in the United States.