2.3. Seasonal variation of homicides

2.3.1. Definition of homicides

The term “homicide” means either the act of killing of one human being by another or a person who kills another (Webster 1996). According to the International Classification of Diseases (ICD-9), the homicides are defined by the codes E960 through E968 presented in Table 4 (World Health Organisation 1977). Sometimes the codes E970-E978 (legal interventions), which include injuries inflicted by the police or other law enforcement agents in the line of duty, are also considered under the category of homicide (Fingerhut & Kleinman 1990).

When a researcher is interested in homicides, a clear definition must be presented so that no ambiguity remains as to whether she/he investigates homicide offenders or victims of a homicide. Official homicide rates usually measure the number of people killed rather than the number of people who have killed others (Holinger 1979).

2.3.2. Homicide research

The deaths due to a homicide and other injuries inflicted by another person are a major global health problem. The rates per 100 000 mean population vary largely between countries. In Finland, the rate of homicides and injury inflicted by other persons has been one of the highest in Europe being 2.9 in the year 1995. The figure was about twice that of Sweden (1.2, 1996) or Norway (1.0, 1995) and about three times that of Great Britain (1.0, 1994) and France (1.1, 1994). In the USA, the homicide rate was high, being 8.6 in 1995. In the Southern Hemisphere country Australia, the homicide rate was 1.6 in 1995 (Statistics Finland 1999a).

In Finland, statistics kept on homicides are reliable. During 1995-98 the clearance rate of homicides has varied between 86-100% (Statistics Finland 1999b). A thorough psychiatric examination ordered by the Finnish courts of law has to be conducted on most of the subjects accused of homicides (Tiihonen et al. 1993). All the forensic psychiatric reports are filed by the National Board of Health and Welfare. The latter has offered a good opportunity to study reliably possible risk factors linked with homicidal behaviour in Finland.

During the last decade, men have committed about 90% of all deaths due to homicide in Finland (Eronen et al. 1996a, Statistics Finland 1999b). On the other hand, about 70% of deaths due to homicides and injury inflicted by other persons were suffered by male victims (Statistics Finland 1997).

Several studies from Finland, which have used methodologically valid data of homicide offenders, have shown that homicide violence is strongly associated with mental disorders - such as antisocial personality, alcoholism and schizophrenia alone or with interactions between each other (Tiihonen et al 1993, Tiihonen & Hakola 1994, Eronen 1995, Eronen et al.1996a, Eronen et al.1996b, Tiihonen et al.1996, Putkonen et al. 1998). A similar relationship between patients with a mental disorder and violent offences has also been suggested in studies from other countries than Finland like, for instance, in Sweden (Lindqvist & Allebeck 1990, Hodgins 1992, Modestin 1998).

2.3.3. Seasonal variation in aggressive behaviour

According to Tennenbaum and Fink (1994), it has long been assumed that there exists a relationship between crimes, particularly violent crimes and seasons. Increased violent offending has been shown to be correlated with increased temperature (Anderson & Anderson 1984, Rotton & Frey 1985, Anderson 1987). In the United States, Michael & Zumpe (1983) reported a significant summer peak in the monthly numbers of rape in 12 out of 13 separate regions and of aggravated assaults in 14 out of 16 separate regions. Recently, Schreiber et al. (1997) reported that individual violent crimes (i.e. sexual offences, aggravated assaults) and collective acts of hostility (i.e. wars) were characterised by a circannual rhythmicity with a summer peak and a winter trough. These rhythms were shown to correlate significantly with the duration of the daily photoperiod.

The effect of the seasons on the incidence of aggressive and impulsive behaviour has been demonstrated also with biological data. Low levels of CSF-5 HIAA have been reported to be associated with increased impulsiveness (Lidberg et al. 1985, Linnoila & Virkkunen 1992). It has been suggested that certain types of violence may display annual patterns that are influenced by seasonally regulated mechanisms in a human’s biological processes (Fossey & Shapiro 1992).

2.3.4. Homicide seasonality

Only a few studies in psychiatric research have dealt with the seasonal variation of homicides. Most of these studies have failed to detect any statistically significant homicide seasonality (Michael & Zumpe 1983, Abel et al. 1985, Abel et al. 1987, Ferreira de Castro et al. 1989, Maes et al. 1993a). Although Abel (1986) was unable to detect any significant seasonality, he noted that most homicides of children occurred during spring and fall. Lester (1979) reported that the significant peak months for homicides in the USA were July and December. Goodman et al. (1989) found a significant summer homicide peak in Oklahoma. Tennenbaum and Fink (1994) noticed significantly more homicides in July and August than in January in the USA.

It is suggested that the lack of statistical significance in homicide seasonality studies is a reflection of the methodological or statistical limitations of these studies. Many studies have used too short a time series of homicides leading to a small sample size and thus, a lack of power in statistical analyses (Lester 1979). In a long time series it is easier to distinguish the random fluctuations on the series from those that are systematic (Tennenbaum & Fink 1994). In addition, the use of inadequate and less powerful statistical methods has been a focus of criticism (Maes et al. 1993a).

2.3.5. Meteorological factors and homicide seasonality

Over 100 years ago, Quetelet presented the “thermic law of crime”, which stated that “crimes against the person increase, and those against property decrease, with seasonal and geographic increases in temperature” (cited by Michael & Zumpe 1983). Basically, the temperature-aggression hypothesis states that high temperatures increase aggression through several (possibly related) psychological and biological processes (Anderson 1989).

Michael and Zumpe (1983), however, were unable to demonstrate any clear association between temperature or latitude (photoperiod) and the monthly number of murders in locations differing geographically. Anderson and Anderson (1984) showed that increased temperature was significantly related to both increased number of criminal assaults (study 1: homicide, rape, battery, and armed robbery) and increased number of aggressive crimes (study 2: murder, rape). However, they did not find any time-relationship between temperature and non-aggressive crimes (study 2: robbery, arson). Maes et al. (1994) failed to detect any significant time-relationships between homicides and any of the weather variables studied.

2.3.6. Biological factors and homicide seasonality

Impulsive behaviour of humans has been shown to be closely related to a low concentration of the major serotonin metabolite 5-HIAA in the CSF (Brown et al. 1979, Linnoila et al. 1983, Virkkunen et al. 1994). Linnoila and Virkkunen (1992) suggested that there exist a “low serotonin syndrome, a disorder of impulse control afflicting a subgroup of violent offenders” and this syndrome is characterised by the early onset of impulsive violent behaviour and alcohol abuse, an increased risk of suicide and a family history of type II alcoholism. Low levels of CSF 5-HIAA have proposed to reflect a disorder of serotonin turnover that may make subjects more prone to acts of violence in states of emotional turmoil: the men, who had killed a sexual partner, were shown to have lower levels of CSF 5-HIAA than the healthy control subjects (Lidberg et al. 1985). Testosterone concentration in CSF has also proposed to associate with outward directed aggressiveness: mean CSF free testosterone was found to be higher in alcoholic, impulsive offenders with antisocial personality than in healthy volunteers (Virkkunen et al. 1994).

2.3.7. Subgroup analyses in homicide seasonality studies

Subgroup analyses of homicide seasonality are lacking, although the incidence of homicide has been shown to differ in the subgroups of a population when compared to the population as a whole. Rates of homicides have found to be higher among males and in younger age groups (Ferreira de Castro et al. 1991). Furthermore, the rates of homicide have found to differ in relation to the method of the homicide. For example, among young males the proportion of homicides caused by firearms was 75% in the USA, while it was 18% in Finland (Fingerhut & Kleinman 1990).

2.3.8. Time trends in homicide seasonality

Time trends in the seasonality of homicides have only been sparsely studied. Lester (1979) compared the homicides of 1973 with those of 1972 and found similar seasonal patterns with peaks in July and December. Näyhä (1980) found a significant seasonal variation of homicides in Finland over the period of 1878-1972. He observed that the month of peak incidence was October in the years 1878-89 and 1890-99, August during the years 1900-16 and 1946-53, and July in the period of 1954-72.