Studies, public reports, news reports, and Web sites cite a wide range of values associated with deaths resulting from excessive heat and excessive cold. For example, in the United States, the National Climatic Data Center's Storm Data statistics of temperature-related deaths are skewed heavily toward heat-related deaths, while the National Center for Health Statistics Compressed Mortality Database indicates the reverse—4 times more people die of “excessive cold” conditions in a given year than of “excessive heat.” In this study, we address the fundamental differences in the various temperature-related mortality databases, assess their benefits and limitations, and offer suggestions as to their use. These datasets suffer from potential incompleteness of source information, long compilation times, limited quality control, and the subjective determination of a direct versus indirect cause of death. In general, these separate mortality datasets should not be combined or compared, particularly with regard to policy determination. The use of gross mortality numbers appears to be one of the best means of determining temperature-related mortality, but those data must be detrended into order to remove a persistent winter-dominant death maximum and are difficult to obtain on a regional daily basis.
Office of Climatology, Department of Geography, Arizona State University, Tempe, Arizona