Introduction
Suicide rates are elevated after stressful life events such as the loss of a partner, or separation. Bereavement is a typical example of such stressful events and is responsible for the high suicide rates of the widowed (Kreitman, Reference Kreitman1988). This work aimed to examine the effect of time on suicide rates immediately after bereavement, that is within 1 week, 1 month and 1 year. The results were expected to suggest conclusions regarding necessities and opportunities in the prevention of suicide.
Method
The suicide data were extracted from the Swiss cause of death statistics (Minder & Zingg, Reference Minder and Zingg1989), courtesy of the Federal Statistical Office in Neuchatel. The data entered in the analyses cover the period 1987–2005. ICD-8 coding was in use until 1994, and ICD-10 coding after that. Since 1998, an extra code (X618) reveals assisted suicides, thus enabling this cause of death to be differentiated in the subsequent analyses. Assisted suicides that were recorded since 1998 were excluded from the analysis. Like other violent deaths, suicides figured regularly as the main cause of death. Population data were derived from the 1990 and 2000 censuses.
The time between bereavement and subsequent death, specifically suicide, was determined by linkage of individual records. This linkage was enabled by merging two data streams from the death records: the records of deceased married persons comprising the date of birth of the partner, and the records of deceased widowed persons comprising the date of death of the partner. This information has been available since 1987. Additional matching criteria were gender and location of habitation (canton).
The analyses were restricted to deaths of married persons for the period 1987–2004 in order to retain the 12-month follow-up period in all widowed persons. We included only couples in the analyses that we could match regarding all criteria. Probable suicide-pact couples (identified by marital status, same day of suicide, and same method of suicide) were excluded. Probable homicide-suicides were identified in a similar way (identified by marital status, suicide within 10 days after homicide).
The suicide mortality of the widowed was calculated specifically for the first 12 months and, applying the person-years approach (Rothman & Greenland, Reference Rothman and Greenland1998), for the first month and the first week after bereavement. Apart from sex we differentiated the data by age (<60 years versus ⩾60 years).
Results
During the period 1987–2005 there were 26 841 suicides registered in the Swiss mortality statistics. Thirty-nine suicides were identified as probable homicide-suicides, all but one being committed by men; 35 (90%) of these were carried out using firearms. Eight pairs of suicides that were probably due to suicide pacts were reclassified. Thirteen assisted suicides that occurred within 12 months of bereavement were also excluded from the analyses. Three hundred and sixty-six suicides of widowed persons could be related unequivocally to a partner who had died within the previous 12 months, whereas 44 suicides of widowed persons that seemingly fullfilled the 12-month criterion could not be matched unequivocally.
About one out of 600 widowers (2500 widows) committed suicide in the first year of widowhood. The annualized suicide rate was highest in the first week after bereavement: 941 men and 207 women per 100 000. The corresponding standardized mortality ratios (SMRs) were approximately 34 and 19. The overall SMR in the first week after bereavement (22, see ‘all’ in Table 1) is only slightly higher than the SMR of the women; this is attributed to the fact that most widowed persons are women, and women generally have distinctly lower suicide rates than men. After the first week, the rates and the ratios decreased much more rapidly than after the first month after bereavement.
Table 1. Suicide rates and standardized mortality ratios (SMRs) 1 year, 1 month and 1 week after bereavement; Switzerland, 1987–2005, by gender
Baseline suicide rates within the same period: all, 18.9; all 20–59 years, 21.8; all ⩾60 years, 31.7; men, 27.4; men 20–59 years, 31.5; men ⩾60 years, 49.8; women, 10.7; women 20–59 years, 11.7; women ⩾60 years, 39.7.
The separation of data by sex and age (<60 years versus ⩾60 years) showed interesting differences even though some of the figures are fairly small. The SMRs of young and middle-aged women are at least as high as the SMRs of men. By contrast, the SMRs of older women are distinctly lower than in men and they smooth down immediately after the first week after bereavement.
Conclusions
This study attempts to provide, for the first time, an insight into the time-related dynamics of suicide immediately after bereavement. It is well recognized that bereavement is a stressful life event that increases the risk of various causes of death (Kaprio et al. Reference Kaprio, Koskenvuo and Rita1987), and in particular of suicide (Erlangsen et al. Reference Erlangsen, Jeune, Bille-Brahe and Vaupel2004; Agerbo, Reference Agerbo2005). It has also been shown that the time effect is crucial; the risk of suicide in widows and widowers decreases after the first year of widowhood (Erlangsen et al. Reference Erlangsen, Jeune, Bille-Brahe and Vaupel2004) until it is similar to the overall rate (MacMahon & Pugh, Reference MacMahon and Pugh1965). Thus, analyses on a more general level, such as on suicides by marital status, are often flawed with regard to the suicide risk of widowed persons. Hence, it is not surprising that analyses that take no account of the time after bereavement in longitudinal modelling (Kposowa, Reference Kposowa2000) may fail to find noteworthy associations between widowhood and suicide.
The person-year calculations in this study indicate that the suicide rate is dramatically high in the initial period after bereavement. The SMR for suicide was about 20–30 in the first week after bereavement. This dramatic increase in the SMR clearly surpasses the odds ratio for the onset of major depression immediately after bereavement, which is of the order of 15 (Zisook & Kendler, Reference Zisook and Kendler2007). The analyses have also confirmed findings from earlier studies that the suicide rates in young widowed persons are distinctly higher than in older widowed persons (Kreitman, Reference Kreitman1988; Smith et al. Reference Smith, Mercy and Conn1988; Luoma & Pearson, Reference Luoma and Pearson2002; Yip & Thorburn, Reference Yip and Thorburn2004). The finding that suicide after bereavement is more frequent in men than in women (Li, Reference Li1995) requires some clarification. It applies only to older widows and widowers, whereas younger widows have a suicide risk that is similar to, or even higher than, widowers within the same age group.
We assume that many suicides immediately after bereavement, that is within the first week of the death of the partner, represent what we might call ‘rash act’ suicidesFootnote †. Furthermore, we assume that such suicides are similar to many other suicides occurring immediately after stressful life events, for example after the loss of a partner or a close friend, or after significant quarrels, major disappointments, or bad news. There have been different approaches to categorize such suicides. Some authors differentiate between planned and unplanned suicides (Conner, Reference Conner2004), whereas a different perspective arises in connection with impulsive suicides. It is noteworthy that one out of four nearly lethal suicide attempts is an impulsive attempt (Simon et al. Reference Simon, Swann, Powell, Potter, Kresnow and O'Caroll2001). ‘Rash act’ suicides deserve closer investigation regarding prevention opportunities. These might differ from measures set up to prevent suicides after several weeks or months of grief.
It has been suggested that widows and widowers are a group suitable for preventive efforts (Luoma & Pearson, Reference Luoma and Pearson2002), especially as they have contact with undertakers, priests and often also general practitioners after bereavement. The first days after bereavement provide a preliminary time ‘window’ to intervene and establish support for persons at risk of suicide. For many of them, such an intervention might have beneficial effects in assisting them in coping with any subsequent stressful life events and crises. Prevention projects are needed that address the occupational groups involved and help them to identify and support bereaved persons at risk for suicide.
Limitations
This study is subject to several limitations. First, suicide counts underestimate the actual number of suicide deaths to some extent. This is probably true even for Switzerland, despite its well-developed health system, which includes comprehensive forensic medicine. Moreover, the records of the cause of death include only incomplete information about psychiatric morbidity, but no information on previous suicide attempts. The second limitation relates to incomplete information regarding widowhood and divorce: for example, the length of marriage before widowhood, first-time widowhood or repeated, the number and ages of children, and whether the suicidee was cohabiting with a new partner. In addition, the analysis of linked data of married partners showed that the overall data overestimate the rates of widowed men due to specific suicide types (homicide suicides, suicide pacts). Analysis of suicide records in Basel (Frei et al. Reference Frei, Han, Weiss, Dittmann and Ajdacic-Gross2006) has suggested that the number of homicide suicides and suicide pacts is probably greater than that detected by the procedure used in this analysis. Perhaps this is associated with another limitation: there is doubtless a sizeable proportion of cases that could not be linked due to errors in input data.
Acknowledgements
Suicide data were extracted from Swiss mortality records with kind permission of the Swiss Federal Statistical Office in Neuchatel.
Declaration of Interest
None.
