Tania Busch Isaksen

Abstract:

Increased mortality and morbidity have been associated with extreme-heat events, particularly in temperate climates. Climate change is predicted to increase the intensity and duration of these extreme heat events. Understanding the local heat-health relationship and its spatial distribution is important to predict future health-related burdens from climate change, and to assist local public health officials in emergency heat-preparedness efforts.

This study used a non-parametric Poisson regression model with a piece-wise linear function to estimate heat’s effect on mortality and morbidity, beyond a model-derived threshold. Relative risk and time series analyses were conducted to explore extreme heat effects on mortality and morbidity outcomes for three different time frames: calibration (1980-2006 mortality/1990-2006 hospital admissions), validation (2007-2010), and complete (1980/1990-2010). The results from these analyses were used to test the validity of our piece-wise linear approach to interpreting heat’s effect on health outcomes. This study also provided translational materials to local public health practitioners, detailing heat-risk estimates, vulnerable populations, and heat-risk distribution.

The results demonstrate that heat, expressed as humidex, is associated with increased non-traumatic mortality and hospital admissions on heat days, and that the risk increases with heat intensity - especially for the older populations. The all-ages relative risk of mortality on a heat day (above the 99th percentile) was 10% greater than on a non-heat day, with risk increasing 2.12% for each degree increase in humidex above 36.0 °C. The all-ages relative risk of hospitalization on a heat day (above the 99th percentile) was 2% (non-significant) greater than on a non-heat day, with risk increasing 1.59% for each degree increase in humidex above 37.4 °C. This study found that with the available individual-level characteristics data, only age modified heat’s effect on health outcomes. While the 65+ age groups were found to be at greater risk of poor health outcomes on an extreme heat day, younger age groups were also found to be at risk for specific causes of death and hospitalization.

We found that, overall, our method of using a piece-wise linear term to estimate heat’s effect on mortality and hospital admission rates held up well, even when using a validation time frame that contained the most extreme heat event on record for King County, Washington. Our model’s prediction estimates are conservative, and under predict mortality at the lower bound of the expected confidence interval. Future estimates using this method will likely predict at least as many deaths and hospitalizations as observed.

Our findings warrant additional research into the role heat exposure plays in several specific causes of death and hospitalization, such as diabetes and renal syndromes in the 45-64 age group and circulatory and cardiovascular conditions in the 85+ age group. Future areas ofstudy include improving exposure assessment and vulnerable population data, as well as assessing the next tier of heat-health effects: emergency response system/emergency room data. Finally, what we learn from additional research should improve of our understating of spatial heat-risk distribution and provide our public health partners with additional health related data to share with their constituents.