The objective of this study was to assess the relation between temperature and suicide mortality in Japan using time series analysis with a semiparametric approach.

We analyzed the relation between daily fluctuations in suicide mortality and maximum temperatures for all regions in Japan over the period of time from 1972 to 1995 using a generalized additive model. The model controls for the time trend, season, selected meteorological parameters, day of the week, and holiday. Adjustment was based using penalized splines and the decision on the amount of smoothness was based on minimizing the unbiased risk estimation criterion.

The results show that suicide mortality in Japan has a seasonal character and it varies from year to year, with the highest occurrence in April, as well as in the first part of the week, especially on Mondays and Tuesdays. As for the day of the week, there were only few suicide cases on Saturdays and holidays. We found that for all regions in Japan when temperature increased the suicide mortality increased on the same day (lag = 0). Analysis by method of suicide showed that when temperature increased mortality significantly increased only for suicide by a violent method. The pattern of the relation for other methods remained unclear.

This study suggests that an increase in temperature has a short-term effect on suicide mortality in Japan.