The mortality burden of hourly temperature variability in five capital cities, Australia: Time-series and meta-regression analysis

Background: Unstable weather, such as intra- and inter-day temperature variability, can impair the health and shorten the survival time of population around the world. Climate change will cause Earth's surface temperature rise, but has unclear effects on temperature variability, making it urgent to understand the characteristics of the burden of temperature variability on mortality, regionally and nationally. Objectives: This paper aims to quantify the mortality risk of exposure to short-term temperature variability, estimate the resulting death toll and explore how the strength of temperature variability effects will vary as a function of city-level characteristics. Methods: Ten-year (2000–2009) time-series data on temperature and mortality were collected for five largest Australia's cities (Sydney, Melbourne, Brisbane, Perth and Adelaide), collectively registering 708,751 deaths in different climates. Short-term temperature variability was captured and represented as the hourly temperature standard deviation within two days. Three-stage analyses were used to assess the burden of temperature variability on mortality. First, we modelled temperature variability-mortality relation and estimated the relative risk of death for each city, using a time-series quasi-Poisson regression model. Second, we used meta-analysis to pool the city-specific estimates, and meta-regression to explore if some city-level factors will modify the population vulnerability to temperature variability. Finally, we calculated the city-specific deaths attributable to temperature variability, and applied such estimates to the whole of Australia as a reflection of the nation-wide death burden associated with temperature variability. Results: We found evidence of significant associations between temperature variability and mortality in all cities assessed. Deaths associated with each 1°C rise in temperature variability elevated by 0.28% (95% confidence interval (CI): 0.05%, 0.52%) in Melbourne to 1.00% (95%CI: 0.52%, 1.48%) in Brisbane, with a pooled estimate of 0.51% (95%CI: 0.33%, 0.69%) for Australia. Subtropical and temperate regions showed no apparent difference in temperature variability impacts. Meta-regression analyses indicated that the mortality risk could be influenced by city-specific factors: latitude, mean temperature, population density and the prevalence of several chronic diseases. Taking account of contributions from the entire time-series, temperature variability was estimated to account for 0.99% to 3.24% of deaths across cities, with a nation-wide attributable fraction of 1.67% (9.59 deaths per 100, 000 population per year). Conclusions: Hourly temperature variability may be an important risk factor of weather-related deaths and led to a sizeable mortality burden. This study underscores the need for developing specific and effective interventions in Australia to lessen the health consequences of temperature variability. Keywords: Temperature variability, Temperature change, Mortality, Death burden, Australia