Decadal shifts in summer weather/mortality relationships in the United States by region, demography, and cause of death

It is commonly accepted that human mortality increases during extreme weather conditions. In the United States, even though winter mortality rates are substantially higher than those in summer, excess deaths in the warm season are more closely coupled to weather variables than in the cold season. Furthermore, death rates from some broad disease classes have considerable seasonal components.

Global warming concerns have led to the development, using historical data, of weather/mortality models that are subsequently used to predict future mortality rates given certain climatic parameters. These future mortality projections often serve as the basis for policy decisions. However, in most cases, the models do not explicitly account for temporal changes in death rates for given weather conditions (the equivalent of the technological trend term in agri-climate models). The goals of this research are to: 1) examine decadal changes in summer death rates for comparable weather conditions; and 2) determine if particular disease categories are responsible for any temporal death rate trends.

Daily mortality data for every person who has died in the United States over 25 non-consecutive years from 1964–1994 are organized by 20 Metropolitan Statistical Areas and coupled with hourly weather observations. These data are standardized to account for demographic changes and to simplify comparisons both between cities and within a given city over time. Data are further subdivided by age, race, and cause of death categories as determined from coroner’s reports on the death certificates and categorized according to the International Classification of Diseases, Ninth Revision.

Examination of daily mortality versus the previous day’s apparent temperature shows substantial regional variations. Excess summer mortality has occurred in most northern U.S. cities, but there is little or no mortality response to high apparent temperatures in southern cities regardless of the severity of the conditions.

Decadal plots demonstrate that standardized mortality has declined over time in all cities. But most importantly, the incremental mortality response to high apparent temperatures was greater in the 1960s than in the 1990s. Most of the mortality declines can be accounted for by changes in a few cause-of-death categories, particularly circulatory and respiratory diseases. In some cities, the response of minorities to high apparent temperatures differs from that of the general population.

These results suggest that human and infrastructural adaptations (acclimatization, increased access to air conditioning, public health initiatives, etc.) have reduced the sensitivity of the population to high temperatures and have resulted in cities in some regions that no longer exhibit a mortality response. For future mortality predictions to have value, particularly forecasts of climate change impacts, these temporal trends must be factored into weather-mortality models.