The predictability of high-risk zones for heat-related mortality in seven US cities

Heat-related mortality remains a public health challenge in the United States. The objective of this study was to determine the temporal consistency of high-risk zones for heat-related mortality using historical georeferenced mortality data from seven US cities. A generalized additive model was used to identify city-specific threshold temperatures associated with increased mortality, and then the mortality rate on threshold-exceeding days was calculated for each postal code comprising each study city. This process was iterated by withholding subsets of data from the model and assessing predictability via cross-validation. In all cities, the average mortality rate in postal codes targeted for intervention by the statistical model was higher than that in non-targeted areas. Targeted areas for interventions in the study data accounted for 50 % of excess heat-related deaths despite only accounting for 25 % of total mortality. Focusing intervention measures at certain geographical zones within urban areas could be an effective means of combating heat-related mortality because there is temporal consistency in places where the death rate is most sensitive to heat.