Modeling Past And Future Mortality Effects Of Temperatures Accounting For Adaptation

Abstract

While climate change will inevitably lead to warmer temperatures, the impact of these changes on future mortality risks remains unclear, due in part to large uncertainties in the magnitude of future exposure-response functions. Studies have shown that the impact of temperature on daily deaths varies considerably by location and time, including time of year, and it is reasonable to assume that further changes will occur as cities develop and adapt in coming decades. We modeled changes in exposure-response over time, time of year, and space in relation to a range of adaptation-related factors (mean temperature and air conditioner usage) then projected future mortality impacts taking into account both climate change and projected trends in those factors. We used Poisson regression to estimate monthly functions relating daily average temperature to excess mortality in 209 U.S. cities in five-year intervals from 1973 to 2006. In a second stage, we meta regressed the effect estimates on average monthly temperature and percent air conditioner use, controlling also for time, race, cluster and month. We developed daily average temperature projections at 1 degree spatial resolution using two climate models, each run with the RCP 6.0 emission scenario. Results showed diminished heat effects and enhanced cold effects in locations and at time periods with higher mean temperatures. A/C use was associated with a reduction in mortality effects of very high temperatures. Projections into the future using exposure-response functions modified by mean temperature showed reduced mortality effects of both hot and cold temperatures as compared with “naïve” analyses that simply held exposure-response constant in the future, with, for example, between 10 and 33 additional temperature-related deaths per million population in 2050 depending on the climate model. Results illustrate the potential influence of adaptation in modifying future climate impacts on premature mortality.