Search

Abstract Many previous literatures suggested that high temperature and diurnal temperature range (DTR) are prominent risk factors to increase risk of mortality. However, the role of effect modification of temperature on the DTR-related mortality is unclear. We examined whether temperature was an effect modifier on the DTR-mortality association and how the modification patterns differed by cause of deaths and different regional climates using a nationwide 47 prefecture data in Japan (1972–2015). We used a two-stage analysis. For the first stage, we used a time-series regression with a distributed lag model to estimate the DTR-mortality association according to five levels of temperature (extreme cold, cold, moderate, hot, and extreme hot days) for each prefecture stratified by twelve cause-specific deaths. Then, we applied a meta-analysis to pool the estimates across the 47 prefectures in Japan and separately by cooler vs. warmer regions. Our findings showed that the risk of mortality associated with DTR was strongly modified by temperature for all causes and cardiovascular deaths (p < 0.001) in the total population, suggesting that the influence of DTR on mortality increases at higher levels of temperature. These findings were consistent across warmer and cooler regions. Similar patterns were observed for respiratory and renal disease deaths which demonstrated the associations with DTR were the highest during extreme hot days, although it was statistically not significant and varied depending on the climate regions. Our findings suggest that the DTR-related mortality may be modified by daily mean temperature and the most elevated during extremely hot temperatures. Highlights • We aimed to identify the DTR-related mortality with cause-specific deaths. • Weather and 12-specific deaths data of Japan (47 prefectures; 1972–2015) were used. • We found the DTR-related risks were highest in respiratory and renal deaths. • The modification of temperature was significant in total and cardiovascular deaths. • The DTR-related mortality changed during the study period. [ABSTRACT FROM AUTHOR]