Comparison of heat-illness associations estimated with different temperature metrics in the Australian Capital Territory, 2006-2016.

While the associations of heat with health outcomes is well researched, there is less consensus on the measures used to define heat exposure and the short-term and delayed impacts of different temperature metrics on health outcomes. We investigate the nonlinear and short-term relationship of three temperature metrics and reported incidence of three gastrointestinal illnesses: salmonellosis, campylobacteriosis and cryptosporidiosis in the Australian Capital Territory (ACT). We also examine the nonlinear association of these illnesses with extreme heat (5th, 75th, 90th percentile of all heat measures). Generalized linear models with Poisson regression accounting for overdispersion, seasonal and long-term trend, weekly number of outbreaks and rainfall were developed for mean and maximum weekly temperature and the heat stress index (EHIaccl). Bacterial illnesses (salmonellosis and campylobacteriosis) showed an overall positive association with extreme heat (75th and 90th percentile of all three heat measures) and an inverse association with low temperature (5th percentile). The shape of the exposure-response curve across a range of temperatures and the lagged effects varied for each disease. Modelling the short-term and delayed effects of heat using different metrics across a range of illnesses can help identify the most appropriate measure to inform local public health intervention planning for heat-related emergencies.