The association between temperature variability and cause-specific mortality: Evidence from 47 Japanese prefectures during 1972-2015

Background: In the context of climate change, extreme temperature events are known to be associated with increased mortality risk. However, data about the mortality risk related to temperature variability (TV) accounting for both intra-and inter-day variations in temperature are limited. Objectives: The present study aims to quantify the associations between TV and cause-specific mortality in Japan, evaluate whether the effects of TV are modified by prefecture-level characteristics and examine the temporal trend in mortality risk of TV. Methods: Data on daily all-cause and 11 cause-specific mortality and meteorological variables in 47 Japanese prefectures from 1972 to 2015 were collected. TV was defined as the standard deviation of daily minimum and maximum temperatures during exposure days. A quasi-Poisson regression model combined with a distributed lag non-linear model was firstly applied to assess the prefecture-specific mortality effects of TV, adjusting for potential confounders. The pooled effects of TV at the national level were then obtained via a meta-analysis through the restricted maximum-likelihood estimation. Potential effect modification by prefecture characteristics was firstly examined using a meta-regression analysis, and the joint modification of season and humidity was then evaluated after including product terms in two-stage analyses. Finally, the temporal trend in TV effects was evaluated by a random-effect meta regression model after obtaining the prefecture-year-specific effects. Results: TV had significant adverse effects on all-cause and cause-specific mortality. The effects of TV were more detrimental to those with asthma and senility. In general, the estimates of mortality risk increased with longer exposure days. A 1 degrees C increase in TV at 0-7 days of exposure was associated with a 0.9% (95% confidence intervals: 0.82%-0.98%) increase in all-cause mortality. All-cause mortality risk of TV showed a decreasing trend during our study period. TV effects were larger in densely populated prefectures and on warm and humid days. Conclusions: TV-related death is a significant issue in Japan that requires effective interventions.