We read with interest Yip et al's (Reference Yip, Chao and Chiu2000) report on the decrease or disappearance in seasonal variation of suicides in the 1980s and 1990s in England and Wales. The authors have suggested a similar trend in other countries (Reference Ho, Chao and YipHo et al, 1997; Reference Yip, Chao and HoYip et al, 1998). This paper inspired us to explore seasonality of suicide in Japan, where national data on monthly suicides are collected and published every year by the Ministry of Welfare. We examined monthly data by gender for the 15-year period 1982-1996. A daily mean suicide incidence was calculated for each month. Harmonic analysis (Reference PocockPocock, 1974) was applied to identify seasonal components in the variation of monthly suicides.

A total of 332 651 suicides (215 686 male, 116 965 female) were identified for the period (mean suicide rate 25.8 and 13.2 per 100 000 for men and women, respectively). During the entire period, the age distribution of suicide rates, which generally increases with age, and popular methods of suicides (approximately 60% violent, nearly 35% non-violent, remainder unclassifiable or method not known) were generally unchanged. Regardless of gender, mean daily suicide incidence by month demonstrated a clear bimodal distribution, with a largest peak in April and a smaller peak in September. Harmonic analyses revealed that over 65% (66.7% for men and 66.2% for women) of the total variance in suicides was explained by the seasonal component, where one-cycle and two-cycle components illustrate the majority of total seasonal harmonics (one-cycle: 64.3% for men and 69.1% for women; two-cycle: 32.9% for men and 28.7% for women). No gender difference was found in the results from the harmonic analyses.

Unfortunately, no data on monthly suicides by age, by method or by place are available from the Japanese Government, which complicates further analyses regarding the backgrounds of such a marked seasonality in suicides. It may be noted that the school and business year begins in April in Japan, which may affect the seasonality of suicides, at least the largest peak of suicides in April. Overall, the results of our analyses indicate limitations in generalising from Yip et al's conclusions and suggest a wide variability in seasonality of suicide in different areas.