What drives cold-related excess mortality in a south Asian tropical monsoon climate-season vs. temperatures and diurnal temperature changes.

Despite the tropical climate which is characterized by generally high temperatures and persistent mild temperatures during the winter season, Bangladesh, along with many other tropical countries, experiences strong winter and cold-related excess mortality. The objective of this paper was to analyse the nature of these cold effects and understand the role of season vs. temperature and diurnal changes in temperature. For approaching these questions, we applied different Poisson regression models. Temperature as well as diurnal temperature range (DTR) were considered as predictor variables. Different approaches to seasonality adjustment were evaluated and special consideration was given to seasonal differences in atmospheric effects. Our findings show that while seasonality adjustment affected the magnitude of cold effects, cold-related mortality persisted regardless the adjustment approach. Strongest effects of low temperatures were observed at the same day (lag 1) with an increase of 1.7% (95% CI = 0.86-2.54%) per 1 °C decrease in temperature during the winter season. Diurnal temperature affected mortality with increasing levels at higher ranges. Mortality increased with 0.97% (95% CI = 0.17-1.75%) when looking at the entire season, but effects of DTR were not significant during winter when running a seasonal model. Different from effects observed in the mid-latitudes, cold effects in Bangladesh occurred on a very short time scale highlighting the role of temperature versus season. Insufficient adaptation with regard to housing and clothing might lead to such cold-related increases in mortality despite rather moderate temperature values. Although the study did not demonstrate an effect of DTR during the cold season, the strong correlation with (minimum) temperature might cause a multicollinearity problem and effects are difficult to attribute to one driver.