Is precipitation a predictor of mortality in Bangladesh? A multi-stratified analysis in a South Asian monsoon climate.

While numerous studies have assessed the association between temperature and mortality in various locations, few have addressed the relationship between precipitation and mortality. Given the high amounts of rainfall in many tropical monsoon areas and the often seasonally pronounced differences, there might be a potentially strong impact on health outcomes and death. In this study, we investigated the association between precipitation and daily death counts in Bangladesh from 2003 to 2007 using regression models with a quasipoisson distribution adjusting for long-term time and seasonal trends, day of the month, age and perceived temperature. Effects were assessed for all ages, the elderly and by gender. During the dry season a sharp increase in death risk was found at very high precipitation amounts which are most likely to be cyclone-related. This cyclone effect was most pronounced for females at the immediate day with an increase of 18.7% (3.8-35.6%) in non-external cause mortality per mm precipitation above 5mm. At longer lags we found a negative association between precipitation and mortality indicating some kind of dry effect which was more pronounced for the elderly with a mortality increase of 4.4% (2.6-6.2%) per mm decrease in precipitation. During the rainy season, we observed a protective effect of rainfall which was strongest during periods of seasonally high equivalent temperatures with a decrease in mortality of 4.0% (2.3-5.6%) per mm increase in precipitation on the immediate day. The observed associations between precipitation and mortality differed by season, age and gender. Generally, a strong short-term increase in mortality was associated with cyclonic activity during the dry season, while ongoing low rainfall seemed to have an adverse impact at higher lags. During the rainy season, precipitation seemed to mitigate heat effects.