Role of temperature, influenza and other local characteristics in seasonality of mortality: a population-based time-series study in Japan

Objectives To investigate the extent to which temperature and influenza explained seasonality of mortality in Japan and to examine the association of the seasonality with prefecture-specific characteristics. Design We conducted time-series analysis to estimate the seasonal amplitude before and after adjusting for temperature and/or influenza-like illness (ILI). Next, we applied linear mixed effect models to investigate the association of seasonal amplitudes with each indicator on prefecture-specific characteristics on climate, demographic and socioeconomic factors and adaptations. Setting 47 prefectures in Japan Participants Deaths for all-cause, circulatory, and respiratory disease between 1999 and 2015. Outcome measures Peak-to-trough ratio (PTR, a measure of seasonal amplitude). Results The nationwide unadjusted-PTRs for all-cause, circulatory and respiratory mortality were 1.29 (95% CIs: 1.28 to 1.31), 1.55 (95% CI: 1.52 to 1.57) and 1.45 (95% CI: 1.43 to 1.48), respectively. These PTRs reduced substantially after adjusting for temperature but very little after a separate adjustment for ILI. Furthermore, seasonal amplitudes varied between prefectures. However, there was no strong evidence for the associations of PTR with the indicators on prefecture-specific characteristics. Conclusions Seasonality of mortality is primarily driven by temperature in Japan. The spatial variation in seasonal amplitudes was not associated with prefecture-specific characteristics. Although further investigations are required to confirm our findings, this study can help us gain a better understanding of the mechanisms underlying seasonality of mortality.