Chapter 1. Introduction

Numerous epidemiological studies have shown that cyclic clusterings, such as monthly or seasonal variations of events, occur in several psychiatric conditions. For instance, the effect of seasons on the incidence of depression, on birth rates among patients with schizophrenia, or on hospital admissions due to a mental disorder have carefully been investigated (Fossey & Shapiro 1992, Takei et al. 1992, Modestin et al. 1995, Torrey et al. 1997). The purpose of this kind of studies has not only been to describe a seasonal pattern of events, but above all, to seek the explanations for the phenomenon. To study the seasonal variation of psychiatric disorders is, therefore, one way to clarify the aetiology and the course of a disorder.

Suicides are a major public health problem throughout the world. The suicide mortality in Finland is one of the highest in the world. In 1997 the suicide rate per 100 000 mean population was 25.7, while, for example, in Sweden it was 14.2 (year 1996) and in the USA 11.9 (year 1995) (Statistics Finland 1999a). Intensive research in Finland has considered the aetiology of suicides from various viewpoints. A number of Finnish studies have focused on suicide victims with different types of mental disorders or studied suicides in relation to social factors (Heikkinen 1994, Isometsä 1994a, Marttunen 1994, Henriksson 1996, Heilä 1999), concentrated on suicide methods (Öhberg 1998a) and suicides in specific subpopulations, like, for instance, among physicians (Lindeman 1997).

The seasonal distribution of suicides has been of continuous interest to researchers. In both the Northern and Southern Hemisphere a consistent finding has been that cases of suicide display a significant peak in spring or early summer and a trough in winter (see, for example, Kevan 1980, Näyhä 1982, Massing & Angermeyer 1985, Chew & McCleary 1995). The seasonal distribution of suicides has been found to vary in the different subgroups of a population, for example in males and females (Meares et al. 1981). Furthermore, the seasonal distribution of suicides has been shown to be related to the seasonal patterns of other phenomena, like climatic variables (see, for example Maes et al. 1994, Preti 1997). In Finland, the latest studies on suicide seasonality were published in the early 1980s (Näyhä 1980, Näyhä 1982, Näyhä 1983, Näyhä 1984).

The number of deaths due to homicides (murders, manslaughters) and injury inflicted by other persons varies largely between countries. In Finland this rate has been one of the highest in Europe being 2.9 per 100 000 mean population in the year 1995. The rate of homicides and injury inflicted by other persons was 1.2 (year 1996) in Sweden and 8.6 (year 1995) in the United States (Statistics Finland 1999a).

The seasonal distribution of homicides has only sparsely been studied in psychiatry. Most of these studies have failed to find any statistically significant seasonality in deaths due to a homicide (Michael & Zumpe 1983, Abel et al. 1985, Abel et al. 1987, Goodman et al. 1989, Maes et al. 1993a,b). Lester (1979) reported a significant excess of homicides in July and December in the USA. In Finland, Näyhä (1980) found a significant peak in July, August or October, when studying secular trends in the seasonality of homicides over several separate time periods between 1878-1972. Thereafter, the seasonal variation in homicides in Finland has not been investigated.

A variety of statistical techniques can be used to examine a seasonal pattern of events. These techniques vary from simple descriptive techniques, such as a chi-square test (Siegel & Castellan 1988), to specific time series analyses, like a spectral analysis (Chatfield 1996). The more sophisticated the statistical method, the more methodological knowledge and experience is required from a researcher, who wants to apply it correctly. In seasonality analyses it is important to use such a statistical method, which is sensitive to a specific type of seasonal or cyclic regularity in the data (Tennebaum & Fink 1994).

One way to raise the researchers’ awareness of statistical and other methodological issues in scientific research is to make surveys, whose purpose is to evaluate the quality of the data and the appropriateness of the statistical techniques used in the published articles. There are studies, which have dealt with statistical contents, statistical errors or the use of statistical techniques of published articles in psychiatric research (White 1979, DeGroot & Mezzich 1985, Hokanson et al. 1986, McGuigan 1995, Nieminen 1995, Everitt & Landau 1998). Although researchers have frequently raised concerns about poor statistical methods and inadequate samples found in suicide seasonality studies (Lester 1971, Lester 1979, Phillips & Wills 1987, Miccolo et al. 1989, Maes et al. 1993a,b, Tietjen & Kripke 1994, Ho et al. 1997), no one has systematically evaluated methodological and statistical features of suicide seasonality studies.

In this thesis all completed suicides in Finland during 1980-95 and all homicides that occurred in Finland during 1957-95 are investigated in terms of their seasonal variation. In addition, the use of statistical techniques and features of the data are evaluated from those studies, which have examined seasonal variations of suicides during 1970-97.