Exploring the influence of short-term temperature patterns on temperature-related mortality: a case-study of Melbourne, Australia

John L. Pearce, Madison Hyer, Rob J. Hyndman, Margaret Loughnan, Martine Dennekamp, Neville Nicholls

Several studies have identified the association between ambient temperature and mortality; however, several features of temperature behavior and their impacts on health remain unresolved. We obtain daily counts of nonaccidental all-cause mortality data in the elderly (65 + years) and corresponding meteorological data for Melbourne, Australia during 1999 to 2006. We then characterize the temporal behavior of ambient temperature development by quantifying the rates of temperature change during periods designated by pre-specified windows ranging from 1 to 30 days. Finally, we evaluate if the association between same day temperature and mortality in the framework of a Poisson regression and include our temperature trajectory variables in order to assess if associations were modified by the nature of how the given daily temperature had evolved. We found a positive significant association between short-term mortality risk and daily average temperature as mortality risk increased 6 % on days when temperatures were above the 90th percentile as compared to days in the referent 25-75th. In addition, we found that mortality risk associated with daily temperature varied by the nature of the temperature trajectory over the preceding twelve days and that peaks in mortality occurred during periods of high temperatures and stable trajectories and during periods of increasing higher temperatures and increasing trajectories. Our method presents a promising tool for improving understanding of complex temperature health associations. These findings suggest that the nature of sub-monthly temperature variability plays a role in the acute impacts of temperature on mortality; however, further studies are suggested.