The US EPA responses to challenges to its Endangerment Finding for increasing CO2 concentrations blows hot and cold.
Several commenters (e.g., 3347.3, 11453.1, 3187.3) note that the April 2009 TSD indicated that cold-related deaths presently exceed heat-related deaths in the United States and that this provides evidence that a warming climate will have beneficial effects on temperature-related mortality. Commenters note that on page 70 of the April 2009 TSD, EPA states that 5,983 heat-related deaths were reported in the United States between 1979 and 2002. In the same timeframe, 16,555 people died of extreme cold. A commenter (3187.3) provides a paper by Goklany (2007), which indicates that death from extreme cold exceed death from extreme heat.
We have revised the TSD’s estimates of heat-related deaths based on the latest findings of the assessment literature (Karl et al., 2009). Based on these results, other supporting evidence presented in the TSD, and additional evidence cited below, we have determined that the available literature strongly supports the conclusion that extreme heat is, on an average annual base, the leading cause of weather-related death in the United States. We agree that the April 2009 TSD contained statistics that could be interpreted as suggesting that cold-related mortality has recently been higher in the United States than heat-related mortality. The cold-related mortality statistics in the TSD from Ebi et al. (2008) are similar to those cited by Goklany (2007). However, the methods and data for estimating heat-related mortality were recently updated and these revised values are presented in Karl et al. (2009).
The more recent heat-related mortality numbers from Karl et al. (2009) reflect results from the Centers for Disease Control and Prevention (CDC). CDC (2006) reports more than 3,400 deaths from 1999 to 2003 for which exposure to extreme heat was listed as either a contributing factor or the underlying cause of death. This result of roughly 680 heat-related deaths per year is almost identical to the 689 deaths per year from cold exposure reported by Ebi et al. (2008) and summarized in the TSD. CDC (2006) suggests that even the revised heat-related mortality numbers may underestimate total heat-related mortality, noting: “Because heat-related illnesses can exacerbate existing medical conditions and death from heat exposure can be preceded by various symptoms, heat-related deaths can be difficult to identify when illness onset or death is not witnessed by a clinician. In addition, the criteria used to determine heat-related causes of death vary among states. This can lead to underreporting heat-related deaths or to reporting heat as a factor contributing to death rather than the underlying cause.” This issue has long been recognized in attempting to estimate the mortality impact of extreme heat using information from death certificates (American Medical Association Council on Scientific Affairs, 1997). As noted in a subsequent response (5-29), cold-related deaths are likely also underestimated. One complication with these death certificate–based estimates of extreme cold and heat is they are not limited to periods that would be considered heat waves or cold snaps in the location where the death occurs. Therefore, while these results are based in a consistent methodology and data source, they have an uncertain overlap with the occurrence of the weather events of primary interest to the TSD, cold snaps and heat waves. As a result, these data alone do not provide strong evidence the heat-related mortality is presently greater than cold-related mortality.
However, we note that alternative and much higher estimates of heat-related mortality come from analyses of daily urban summertime mortality patterns in Kalkstein and Greene (1997) and Davis et al. (2003a), which use a different methodology to compute heat-related deaths compared to CDC (2006). These studies first define extreme heat events by identifying threshold conditions for an event in a location and then calculate the number of extreme heat–attributable deaths based on differences in daily deaths on extreme heat days compared to longer-term averages. In these studies, heat’s mortality impact is quantified in terms of the excess deaths that result during the extreme heat conditions. By evaluating changes in daily deaths attributable to all causes, this approach also effectively eliminates differences or restriction in using certain causes of death as potential sources of bias in estimating the extreme heat’s mortality impact. This method is also more consistent with the view that heat waves are effectively identified through exceptional weather conditions that result in increases in daily mortality (e.g., Confalonieri et al., 2007; U.S. EPA, 2006a). Although differences in the time series, definitions of urban populations, and other analytical methods prevent an exact comparison of the results in these two studies, both studies (Kalkstein and Greene, 1997; Davis et al., 2003a) estimate that there are approximately 1,700–1,800 excess deaths per year during extreme heat events based on an evaluation of a subset of approximately 40 U.S. metropolitan areas (see U.S. EPA, 2006a). These estimates of extreme heat’s mortality impact are much higher than the corresponding death certificate–based estimates for heat as well as the Ebi et al. (2008) estimate for cold-related mortality summarized in the TSD.
We also note that Davis et al. (2004) find that the net impact of the observed temperature increase from 1964 to 1998 (considering both reduced temperature mortality in winter and increased temperature mortality in summer) was an extra 2.9 deaths (per standard million) per city per year in 28 major U.S. cities. This indicates that extreme heat has been the larger cause of mortality in the recently observed record when temperatures have warmed.
Furthermore, we note that the USGCRP assessment (Karl et al., 2009) specifically refers to a recent study by Borden and Cutter (2008), which concludes heat is the most deadly natural hazard in the United States. It also cites Medina-Ramon and Schwartz (2007), which found that in 50 U.S. cities between 1989 and 2000, extreme heat increased death rates 5.7% while extreme cold increased death rates by only 1.6%. These results are summarized in the TSD.
Though we are aware of a recent study by Andersen and Bell (2009) that finds a similar mortality risk for extremely hot and cold days based on the synthesis of results from 107 U.S. communities (contrasting with Medina-Ramon and Schwartz), Andersen and Bell are clear that cold temperatures more indirectly affect mortality than heat. In addition to the longer lag times for exposure incorporated for the effects of extreme cold (up to 25 days, compared a one-day lag for heat), they note that infectious diseases, which are more common in industrialized countries during colder weather (when people spend more time indoors and in proximity) could account for a substantial portion of the cold-related effect.
Summarizing, both recent studies and the assessment literature provide strong evidence that heat-related mortality presently exceeds cold-related mortality in the United States