Simple temperature-mortality models all showed a clear and strong relationship with historical daily heat-related deaths. Mean temperature was more strongly associated with heat-related deaths over the period 2006-2015 than minimum or maximum temperature, although differences between the models were very small. However, there was a much weaker association of the relationship between modeled and observed heat-related deaths when the model predictions and observations were aggregated by year. Using a simple mean temperature-mortality model, we would have anticipated 80 heat-related deaths in 2016 based on the observed meteorology. Regardless of the exposure variable or model parameterization chosen, the prediction for 2016 using a simple temperature-mortality model fell well below the historical average number of heat-related deaths.
In summary, we were unable to find strong evidence that the unusually high number of heat-related deaths observed in Maricopa County, Arizona in 2016 was related to observed meteorological conditions. Most models, in fact, suggested that 2016 should have been a year with normal or below-normal number of heat-related deaths. If the conventional methods for estimating the temperature-mortality association are reasonably approximating a causal relationship, our results indicate that factors other than the weather were entirely responsible for the surge in heat related deaths in 2016. These findings highlight the importance of non-meteorological factors as drivers of the health burden associated with extreme temperatures, which have generally not been included in quantitative retrospective or prospective studies. Further, they highlight a shortcoming in preparedness and response efforts for heat in the community that should be diagnosed more specifically and addressed as soon as possible. We are challenged to look more broadly and deeply to identify and address root causes of heat-related deaths.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner