Temporal variability in mortality response to hot weather across US metropolitan areas, 1975-2004

Daily all-cause mortality totals across twenty-eight metropolitan areas across the United States, representing a variety of climate zones and city types, were extracted for the study period of 1975-2004. These data were age-standardized to account for demographic changes over the period of study. Mean mortality response (zero-day lag) was then calculated for those Spatial Synoptic Classification (SSC) weather types associated with elevated mortality. To assess temporal trends, mean response over the 30 years was evaluated in running three-year windows. The entire calendar year was also assessed, not just the summer, with each semi-month analyzed separately as well.

Results show significant variability across the US. While heat-related mortality on the hottest summer days has declined since the 1970s, the rate of heat-related mortality is still statistically significant in many cities at the end of the period, and mean anomalous mortality on oppressive days has remained relatively constant since the mid-1990s. Over the course of the year, mean mortality response is greatest in early- to mid-summer across most of the larger northeastern and Midwestern cities, but is shifted to as early as February and March across Phoenix and Dallas.