Seasonality of mortality with temperature adjustment in Japan over 44 years

Abstract

TPS 664: Climate change: temperature effects 2, Exhibition Hall, Ground floor, August 27, 2019, 3:00 PM - 4:30 PM Background: The seasonality of mortality can be a major contributor to a consistent pattern of health-care demands throughout the year. Temperature- one of the key season defining factors, has been consistently linked with increasing mortality. However, little is known about its impact on seasonality of mortality. Investigating seasonality and the effect of temperature on it will provide important information for health risk management in different seasons. Aims: To assess seasonality of mortality with temperature adjustment in Japan and its temporal changes Method: Daily mean temperature and daily death cases for all-cause, circulatory and respiratory mortality from 1972 to 2015 were collected for 47 prefectures in Japan. A generalized linear model with quasi-Poisson distribution was used, with a one-dimensional cyclic spline function with 4 degrees of freedom for the calendar day. The long term trend and day of week were controlled. A cross-basis function, with natural cubic B-spline functions and extended lag period to 21 days, was used for temperature. The ratio of estimated mortality counts of peak versus trough day (RR) was estimated with and without temperature adjustment, by treating trough day as a reference day. RR was estimated for whole country over 44 years first and then for each single year to investigate the temporal changes. All estimates were conducted for all-cause, circulatory and respiratory mortality, respectively. Results: U-shaped relationships were observed for both adjusted and unadjusted seasonality, with peaks in cold seasons and troughs in warm seasons. Seasonality became less obvious after adjusting for temperature. Both adjusted and unadjusted RR showed a decreasing trend for all-cause and respiratory mortality. No significant trend was observed for circulatory mortality. Conclusions: Seasonality risk was higher in cold seasons. Temperature may be an important driver. Additionally, seasonality of mortality decreased over 44 years. Its underlying reasons should be investigated in the future.