Attributing mortality from temperature extremes: A time series analysis in Varanasi, India

Climate extremes are often associated with increased human mortality and such association varies considerably with space and time. We therefore, aimed to systematically investigate the effects of temperature extremes, daily means and diurnal temperature variations (DTV) on mortality in the city of Varanasi, India during 2009-2016. Time series data on daily mortality, air quality (SO2, NO2, O-3 and PM10) and weather variables were obtained from the routinely collected secondary sources. A semiparametric quasi-Poisson regression model estimated the effects of temperature extremes on daily all-cause mortality adjusting nonlinear confounding effects of time trend, relative humidity and air pollution; stratified by seasons. An effect modification by age, gender and place of death as semi-economic indicator were also explored. Daily mean temperature was strongly associated with excess mortality, both during summer (5.61% with 95% CI: 4.69-6.53% per unit increase in mean temperature) and winter (1.53% with 95% CI: 0.88-2.18% per unit decrease in mean temperature). Daily mortality was found to be increased by 12.02% (with 95% CI: 4.21-19.84%) due to heat wave. The DTV has exhibited downward trend over the years and showed a negative association with all-cause mortality. Significant association of mortality and different metric of temperature extreme along with decreasing trend in DTV clearly indicate the potential impact of climate change on human health in the city of Varanasi. The finding may well be useful to prioritize the government policies to curb the factors that causes the climate change and for developing early warning system. (c) 2019 Elsevier B.V. All rights reserved.