The aim of this study was to investigate the age-, gender- and suicide
method-related seasonality of suicide occurrence by using the largest
database examined so far (n=21 279). The Chi-square test for multinom
ials was used as the overall measure of deviation. The monthly observe
d and expected numbers of suicides were calculated and classified by y
ear, month, gender, age groups and suicide methods. To identify the st
atistically significant peak and trough months, the ratio of observed
numbers of suicides to expected numbers with 95% confidence intervals
was calculated. For males, there was a suicide peak from April to July
, while for females the distribution was bimodal (with peaks in May an
d October). In elderly people there was a significant excess in the nu
mber of suicides in autumn, and the troughs were deeper in winter. For
violent suicides there was a unimodal spring peak, but for non-violen
t suicides the distribution was bimodal. The results indicate that sui
cides among elderly subjects, as well as non-violent suicides, occur s
ignificantly more often during autumn than would be expected.