Seasonal variation in specific methods of suicide: a national register study of 20 234 Finnish people

Abstract

Background: Previous studies have suggested differences in seasonal variation of violent and non-violent suicides, but seasonal distribution of the specific methods of suicide is less clear. Method: We investigated the seasonal distribution of the specific suicide methods for both genders during a 16-year period. The seasonal effect was explored by monthly ratio statistics from the large population-based data including all suicides in Finland during years 1980–1995 (n=20 234). Results: Hanging peaked in spring among both genders. In summer, the significant peaks were occurred in suicides by drowning, jumping and gassing among males. The bimodality in female suicides with autumn peaks was explained by poisoning and drowning. Traffic suicides showed substantial winter-troughs for both genders. Conclusions: The results suggested that specific violent and nonviolent methods of suicides grouped into their traditional clusters on the basis of their seasonality, except suicides by gassing and shooting. The seasonal distribution in specific suicide methods should take into account in suicide prevention at clinical work and also it should be considered while determining which methods of suicides are regarded as violent or non-violent methods in research.

Introduction

A seasonal variation in suicides has been reported in several studies (Kevan, 1980, Skutsch, 1981, Maes et al., 1993, Chew and McCleary, 1995). It has also shown to be a gender-specific phenomenon. For males, the major peaks occur in spring and early summer with troughs in winter (Chew and McCleary, 1995, Hakko et al., 1998a, Preti and Miotto, 1998). Correspondingly, there exists a bimodal distribution among females with peaks in spring and autumn (Meares et al., 1981, Näyhä, 1982, Lester and Frank, 1988, Micciolo et al., 1989, Hakko et al., 1998b).

In addition to a gender difference, previous studies have shown that seasonal pattern differ between methods of suicide. In violent suicides (i.e. ICD-9 suicide codes E 953–E 958), for example, significant peaks have been observed in spring and summer (Hakko et al., 1998b, Preti and Miotto, 1998). Examples of both a bimodal cycle with peaks in spring and autumn (Lester, 1985, Hakko et al., 1998b), or no seasonal trend at all (Maes et al., 1993, Preti and Miotto, 1998) have been reported in suicides committed by non-violent methods (i.e. ICD-9 suicide codes E 950–E 952). Furthermore, the troughs in violent suicides have been reported to exist in winter, while no significant troughs in a subgroup of non-violent suicides were found (Hakko et al., 1998b).

The effect of seasons on suicides have been explained by the theories that changing climatic factors affect seasonal patterns of serotonergic function in the brain, and that psychosocial behavior habits vary among human during different seasons (Fossey and Shapiro, 1992, Maes et al., 1995, Pine et al., 1995, Preti and Miotto, 1998). However, an explanation for both the seasonal distribution and differences in methods between violent and non-violent suicides remains elusive. The first study to analyze suicide seasonality by a single method while simultaneously controlling for gender was reported by Lester and Frank (1988). They compared the seasonal variation between three different methods of suicide (poisoning, hanging, firearms) and found a spring peak among males for all three methods, as well as an autumn peak for both hangings and shootings. Females had a bimodal distribution for all methods, but the peaks occurred later for suicides by firearms. More recently, a study by Yip et al. (1998) reported a method seasonality for suicides by hanging in Australia, but not in any methods in New Zealand. In the latest study of method seasonality, based on registered suicides in Italy, Preti and Miotto (1998) found that the more violent methods, i.e. jumping from a high place, hanging and submersion, but not weapons, showed seasonality among males, whereas among females, a seasonality was found for only hanging and submersion.

Recently, we reported an unimodal distribution in violent suicides in Finland with a peak during spring (May–July) and a bimodal distribution of nonviolent suicides with peaks during spring (May) and autumn (October) during 1980–95 (Hakko et al., 1998b). However, it is entirely possible that these two broad categories of suicides, in which two or more suicide methods are combined, might not sufficiently reveal important elements of seasonality phenomenon. For example, if the seasonal peak in one method and the trough in another method occur during the same season, combining these methods into the same category will cancel each other out and, thus, lead to a smoothing of the seasonal patterns. On the other hand, there might be some dominating suicide methods which are alone sufficient to explain the seasonality between violent or nonviolent suicides.

Therefore, we examined our 16-year time series of suicides according to the original classification of suicide methods. The seasonal distribution of suicides was analyzed separately according to each method — namely poisoning, gas, traffic, shooting, hanging, jumping and drowning. Separate analyses were performed for both males and females, because several studies have shown that suicide seasonality is a gender-specific phenomenon (Meares et al., 1981, Preti and Miotto, 1998). We tried to overcome the methodological discrepancies found in previous studies by using a long time series of suicides, monthly data and appropriate statistical methods for seasonality.