Article Text


Suicide risk after spousal suicide or psychiatric admission: effects of assortative mating on heritable traits compared with environmental explanations
  1. Martin Voracek
  1. Correspondence to:
 Dr M Voracek
 School of Psychology, University of Vienna, Liebiggasse 5/3, A-1010 Vienna, Austria;

Statistics from

Agerbo reports a nested case-control study of all (n = 9011) Danish midlife suicides (25–60 years of age) 1982–97.1 He investigated sex specific midlife suicide risks associated with spousal psychiatric admission, spouse or child bereavement by suicide or other causes of death, and several sociodemographic factors. There is a wealth of original findings from this impressive population based register study. Particularly, increased suicide risks were seen after spousal psychiatric admission or spousal or child death, especially after spousal suicide. Several points of criticism regarding certain study premises and the interpretation of study findings are discussed here. In what follows, I will focus on five points: firstly, the age range restriction; secondly, the psychiatric admission criterion; thirdly, the environmental interpretations of certain findings; fourthly, the implications of assortative mating; and fifthly, the evidence for sex specific suicide risks. An attempt is made to place this study’s findings in the context of present information on the epidemiology of suicide. Furthermore, brief comments on the relevance of the study findings and on new research or methodology needed to overcome the present limitation of the knowledge in this area are given.

Firstly, the range restriction in age (25–60 years) clearly restricts the generalisability of this study: it does not capture total (lifetime) suicide mortality. Internationally, in most Western industrialised nations, there is a noticeable positive age relation of suicide risk that furthermore is more pronounced for men than for women. Consider the Danish suicide rates of the elderly population (aged 65 years or over), as available from the European Health for All database (http:/ during the study period (1982–97), yearly male suicide rates were up to 71.2 and never below 48.7 (median rate: 57.3), and female rates were up to 36.6 and never below 17.2 (median rate: 30.8). Although Agerbo’s study is population based, it misses many suicides that occurred in this population. General conclusions might thus be unwarranted. Across most Western industrialised nations, most suicides occur well beyond midlife and this pattern has become more pronounced over the past decades, especially so for men.2 Therefore, research is needed that takes into account the particular age pattern in the prevalence of suicide by analysing lifetime suicide mortality.

Secondly, a psychiatric admission criterion was used as a proxy variable for psychiatric disorders. This certainly constitutes a too restrictive criterion, as only the most severe cases are included. Agerbo emphasises that admission to all psychiatric inpatient facilities is registered and that there are no private psychiatric hospitals in Denmark, but it is well known and, in fact, a truism that most psychiatric illness episodes do not lead to admission to inpatient facilities, but rather are seen by general practitioners, practising psychiatrists, or outpatient facilities only. This further compromises the generalisability of this study’s findings. Research designs from field epidemiology, in the spirit of catchment area studies, would be needed to gain more clarity on this point. Crisis intervention centres and related facilities often maintain large consultation databases that considerably date back. Using more appropriate proxy variables or indicators for psychiatric disorders than the psychiatric admission criterion, such files could be screened for adverse mental health effects (including suicidality) of spousal psychiatric illness episodes.

Thirdly, although genetic evidence for suicide susceptibility is mentioned by Agerbo, an environmental stance in the interpretation of findings is obvious. Specifically, it is asserted that specific genes associated with suicide are yet to be identified. However, converging evidence from clinical, family, twin, and adoption studies suggest a non-trivial role for genetic factors in the predisposition to suicidal behaviour,3–6 which may further differ between populations and thus may also contribute to cross national differences in suicide rates.7,8 Most recently, several molecular genetic studies reported that polymorphisms of the tryptophan hydroxylase gene are related to suicidal behaviour, and these findings have been summarised and confirmed in meta-analyses.9,10 Further molecular genetic studies of suicide attempters or relatives of suicide victims, along with twin, family, and half sibling study designs, would be helpful to disentangle the relative contributions of genetic, non-shared, and shared environmental contributions of suicidality. The full relevance of the current evidence for heritability of suicidality becomes clear in connection with the next point.

Fourthly, assortative mating on human behavioural traits is the rule. Although the assortative mating principle is mentioned in Agerbo’s study, its far reaching implications, given widespread trait heritability, seem undervalued therein. Almost everything under the sun is heritable, and it has been stated that non-zero heritability is something like the “first law” of behaviour genetics and that zero heritability no longer is an interesting null hypothesis.11 Among other evidence, spousal correlations of up to 0.4012,13 on moderately heritable personality traits14 have been reported. Agerbo even cites a study reporting that patients with schizophrenia might find a partner during psychiatric admission. But possible assortative mating effects are too quickly downplayed on two grounds: first of all, because suicide risk was not further influenced when both partners had a psychiatric admission (the problematic nature of this criterion was emphasised above); and then, because of sex differences in specific suicide risks (the extent of these is discussed below). At least in a retrospective fashion, the extent of assortative mating effects on suicidal traits in the general population could be investigated in the course of psychological autopsy studies of suicide victims and their surviving partners. Available evidence regarding the extent of assortative mating among psychiatric patients is not extensive, so this topic definitely is worth examining further.

Fifthly, there is little evidence for sex differences in specific suicide risks in this study, thus invalidating Agerbo’s argument that these would suggest that assortative mating effects are not important here, and, by extension, that the results are more probably attributable to environmental than to genetic susceptibility factors. The sex specific analyses suffer from low statistical power because of sparse data and from possible misclassifications, and thus could be chance results. Agerbo found that a recent (past two years) spousal psychiatric admission increased the suicide risk more in women than in men (based on 48 male compared with 25 female cases only). Conversely, recent spousal suicide increased suicide risk more in men than in women (16 compared with 12 cases), as did recent spousal death by other causes (37 compared with 17 cases). This truly constitutes a complicated pattern of related findings that seems difficult to interpret in a parsimonious manner. I am not aware of any sound and replicated evidence from research on suicide survivors15,16 that would bolster these findings, which furthermore, importantly so, are all statistically not significant. Agerbo further concedes that the data might contain cases of dyadic deaths (homicide-suicide), with typically male perpetrators. While it is true that dyadic deaths are infrequent, such misclassifications alone could account for the third finding above. The above set of findings therefore is in need of clarification through replication in other populations. Additionally, it would be worthwhile to address the extent of misclassifications attributable to dyadic deaths directly with record linkage studies that use forensic registers. However, given the rarity of such cases, deepened knowledge on them would probably be of uncertain practical utility for general prevention and intervention strategies.

To summarise, this study leaves some important questions unanswered. Particularly, it may well be that assortative mating on heritable traits, such as personality variants, psychiatric disorders, and suicidality itself, contribute to the observed increased suicide risk after spousal suicide or psychiatric admission. This alternative hypothesis could be investigated and tested through convenient combinations of the research designs proposed above.

The final note here concerns the use of the term “gender” in this study. I think it would be more appropriately termed being a sex specific study rather than a gender specific study.17,18 The meaning of sex (male, female) is biological, referring to the biological distinction of males and females, whereas the meaning of gender is cultural, referring to men and women as social groups, and including more than biological maleness compared with femaleness—such as identity, norms, roles, personal and social status—which human characteristics undoubtedly are as important in suicidology than elsewhere.


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles