Some Die Young: A Follow-Up Study of Young Men Exposed to Violence in Sweden

ABSTRACT

This register-based study from Sweden examines mortality risks among young male victims of violence. In total, 8722 men aged 18–24 years who spent at least one night in inpatient care after exposure to violence in the years 1992–2005 were compared with 78,702 age-matched men from the general population. Cox regressions showed that men exposed to violence had a >3 times higher risk of all-cause mortality, a >4 times higher risk of suicide, and about a 20 times higher risk of death from later violence. The risk of suicide remained stable during the follow-up, and the risk of all-cause mortality and death from later violence decreased.

Abbreviations: NBHW: The Swedish National Board of Health and Welfare (Socialstyrelsen); Brå: The Swedish National Council for Crime Prevention (Brottsförebyggande rådet)

KEYWORDS:

• Assault
• violence
• male victims
• mortality risk
• suicide

Introduction

Violence is a global phenomenon that can have negative social and individual consequences for the individual, as well as economic costs for society. Much research on the victims of violence has focused on female victims of male violence, and only limited research specifically examines men as victims of violence. Worldwide, more men than women die of homicide and men are the group most exposed to non-fatal violence (WHO, 2014).

Being exposed to violence is a traumatic experience that can have various serious and long-lasting negative consequences. The National Board of Health and Welfare [NBHW] (2014) has shown that women exposed to violence have an increased risk of long-term sick leave, and it is also known that women exposed to major violence are at higher risk of death, regardless of cause, up to 10 years after the violent episode (Trygged et al., 2014a). These women also have an increased risk of a deteriorating economic situation for a long period after the violence has occurred (Trygged et al., 2013, 2014b).

In Sweden, men are overrepresented both as perpetrators and victims of violence. According to the Swedish Crime Survey (Molin & Lifvin, 2019), 3.5% of respondents (aged 16–84 years) stated that they had been victims of assault in 2018. Men more commonly than women reported being victims of assault (4.4% for men vs. 2.8% for women), and the youngest age group (aged 16–19 years) more frequently reported being victims of assault than did other age groups (9.2%). In the Swedish Crime Survey, more men than women stated that they had been victims of serious assault (0.9% of men vs. 0.5% of women). For both men and women, serious assault is more common in the youngest age bracket (aged 16–19 years). In 2018, 55% of all reported assault offenses were against boys/men, and 45% were against women/girls (Molin & Lifvin, 2019).

Despite men being overrepresented as victims of violence in Sweden, there is limited research on this matter, though some qualitative studies have examined masculinity in male victims’ narratives (Burcar Alm, 2013) and identities (Burcar & Åkerström, 2009). Other studies have focused on social workers’ interpretations of male victimization (Kullberg & Skillmark, 2017), victimization in homosexual cohabiting relationships (e.g., Holmberg & Stjernqvist, 2008), and male victims of violence and the role of alcohol (Tryggvesson, 2008). However, there is limited quantitative research into male victims of violence. Given that many men are subjected to violence, there is clearly a knowledge gap here.

To the best of our knowledge, there are no similar studies of male victims of violence in Sweden. The overall objective is to study mortality risks among men discharged from hospital after inpatient care due to violence. We will study all-cause mortality, risk of suicide, and risk of death from later violence among victimized versus non-victimized men.

Literature review

The consequences of victimization

Some studies have scrutinized the consequences of violence among women. In Sweden, women victims of violence, both with and without children, were much more likely than women not exposed to violence to be dependent on means-tested social assistance and to be exposed to premature death, this being the case in all socioeconomic groups. (Trygged et al., 2013, 2014a, 2014b). To our knowledge, similar figures cannot be found for men. Some studies recognize that intimate partner violence can contribute to suicide. Brown and Seals (2019) studied intimate partner violence and suicide, finding that intimate partner problems were present in 26% (N = 1327) of suicide cases for which circumstances were known and that intimate partner violence was present in 43% (N = 575) of cases in which intimate partner problems were identified.

Being a victim of violence can also have negative consequences for men. One study comparing the physical and mental health effects of intimate partner violence on men and women (Coker et al., 2002) found that, for both men and women, intimate partner physical violence was associated with increased risk of poor health, depressive symptoms, substance use, developing chronic disease, chronic mental illness, and injury. Fröberg et al. (2012) showed that young male victims of violence were more addicted to gambling than were non-victimized young men. Another study found that traumatized men expressed stronger symptoms of anger, aggression, and violence than did non-traumatized men (Jakupcak & Tull, 2005).

Because the context in which violence occurs often differs significantly between women and men, it is difficult to generalize results concerning violence against women to men exposed to violence. In Sweden, it is primarily young men aged 20–24 years who are at greatest risk of being exposed to assault (Brå, 2016a). It can be difficult for these men to admit to being victims, this being reinforced by norms that masculinity represents power and strength while the crime victim role instead expresses weakness (Burcar & Åkerström, 2009; Heber, 2017; Pettersson, 2020). Another problem is that the healthcare system is unprepared to provide care for male victims of violence. A survey of emergency departments in Sweden found that most were prepared to care for women and children who were victims of violence, but that they were less prepared to care for other groups, such as victimized men (Rahmqvist Linnarsson et al., 2012).

Men as victims of violence

Violence against women and violence against men often have radically different contexts. When women are victims of violence, the perpetrator is usually a man, but male violence is also directed against other men (Sarnecki, 2008). Violence against women occurs largely within close relationships, and although some studies show that men are also exposed to violence in close relationships (Stolt, 2010; Tsui, 2014), male victims of partner abuse seem less likely than female victims to seek help or report such incidents (Tsui, 2014). The Social Services Act (2001:453) specifically states that women victims of violence should receive support, and there are documents from the authorities to assist in this (NBHW, 2015a). Violence in close relationships is often more severe against women and has farther-reaching consequences (Lövestad & Krantz, 2012; Nybergh, 2013). On the other hand, men are often exposed to violence in places other than the home, and such violence is often committed by unknown perpetrators (Brå, 2016a). The typical case of such violence is assault occurring in a public place at night by one or several unknown people. It is also common for both the perpetrator and victim to have been under the influence of alcohol when the crime was committed (Brå, 2015).

Some studies show that perpetrators of violence also have an increased risk of being victims of violence. Several studies have found that perpetrators of street violence have an increased risk of being victims of such violence. For example, DuRant et al. (2000) found that exposure to community violence significantly predicted adolescents’ own use of violence. Whelan (2013) argued that many young men who become perpetrators do so because they themselves have experienced violence as victims. Wood et al. (2017) showed that, compared with violent men, gang members had experienced more severe violence and sexual assaults and suffered more serious/life-threatening injuries. They also found that, compared with violent men, gang members reported more suicide attempts and self-harming. However, in general, there is limited previous research on the link between gang membership and victimization (Barnes et al., 2012).

Methods

The study is based on register data covering all Swedish cases of men 18–24 years old being discharged from hospital after at least one night of hospital care due to injury caused by another person. All data are hospital data and based on the coding from the International Classification of Diseases (ICD) developed by the World Health Organization (WHO). The current version is ICD 10; the Swedish healthcare system used ICD 9 until 1996 and ICD 10 starting in 1997 (NBHW, 2020). The codes used are supposed to cover all diagnoses related to death (ICD 9: E960–E969; ICD 10: X85–Y09, Y87.I) in the years 1992–2005. These patient register data are linked to the Longitudinal Integration Database for Health Insurance and Labor Market Studies (LISA) and the Cause of Death Register. This means that even those who did not make a police report were included in the study. For each hospitalized patient, nine men of the same age in the same year were randomly selected from the Swedish population register, which contains data on all people living in Sweden. Data were available for the year before the injury or inclusion in the data for the men randomly selected from the population register up to December 31, 2008. In total, 8722 exposed men 18–24 years old were compared with 78,702 age-matched men from the general population. Only cases with information on all variables were included in the analyses. Due to missing information on some control variables (≈0.3% in total), the number of controls is not exactly nine times the number of exposed.

We use Cox proportional hazard regressions and Cox regressions with continuous time-varying covariates to analyze differences in mortality risk between the exposed and non-exposed. Results are presented as relative risks, which indicate the differences between the two groups. Cox regressions take account of both the presence and timing of an event, i.e., whether or not the event occurs and, if it occurs, how long time after baseline. A censored observation is when the event (e.g., death) does not occur during the follow-up period. For censored observations, the time since baseline is considered in the analyses. We analyze three separate outcomes: all-cause mortality, mortality from suicide, and mortality from a new incidence of violence: Suicide, ICD 9: E950–E959, ICD 10: X60–X84; and Death from new violence ICD 9: E960–E969, ICD 10 X85–Y09. When mortality from suicide and violence is analyzed, observations are also censored when a man dies from another cause. Time is measured in days: day 1 is the day after discharge from hospital for injured men and January 1 for men randomly chosen from the population. Data cover all cases of hospitalization for violence in Sweden. Since 2.6% of the hospitalized men covered by the data were hospitalized more than once after violence, they occur more than once in the dataset. As this can lead to erroneously low standard errors, we used the cluster-correlated robust estimate of variance (Huber, 1967; White, 1980; Williams, 2000) to correct for this.

We used Cox proportional hazard regressions for the first table of regression results and Cox regressions with continuous time-varying covariates for the last table. The time variable used is number of days since discharge from hospital. With Cox proportional hazard regressions, we estimate the average association over the follow-up period. To analyze the temporal pattern of mortality, i.e., how the mortality risk varies over the follow-up, we implement continuous time-varying covariates. This means that we relax the assumption about proportional hazards and let the association vary over time, though we do not let the independent variables vary over time. In our case, this entails introducing an interaction term between time and receiving or not receiving inpatient care after violence (0/1). The time variable is centered on zero to obtain the difference in mortality at the beginning of the follow-up and on 3652.5 days, to obtain the mortality difference after 10 years. The time variable (i.e., days) is divided by 365.25 to obtain the change per year.

All background/control variables are from the year before the injury/inclusion in the data, except for the number of days of inpatient care from violence. We chose to not use time-varying independent variables since other aspects of life also might change because of the violence.

Analyses were performed using three different models to determine whether a specific group of control variables affected the results. Model 1 is controlled for what we consider basic variables related to mortality risk differences: age, age-squared, days of inpatient care from violence (the non-exposed are set to the average number of days), days-squared, degree of urbanization (the three major cities/medium-sized and small towns/countryside), country of birth (Sweden/the Nordic countries/Western countries [Western Europe, Australia, Canada, New Zealand, and the United States]/Eastern Europe/the former Yugoslavia/the Middle East/Asia/Sub-Saharan Africa/Latin America), and days of hospital care for other reasons. In addition, Model 2 is controlled for variables related to the social situation: presence of preschool children, gainful employment/not, and financial assistance/not. In addition, Model 3 is also controlled for socioeconomic position: highest education level (compulsory/upper secondary/university) and income quartiles (four groups). Since the non-exposed do not have any days of inpatient care, they are set to have the average number of days of inpatient care from violence in the regression models. This means that in terms of controlling, they are compared to those with 2.5 days of inpatient care. Age and days of inpatient care from violence were represented both linearly and quadratically in the analyses to capture any possible curvilinearity in the association with mortality risk.

Results

Tables 1 and 2 present descriptive statistics for the people included in the study. Men who received hospital care after violence had a 4.5 times (6.42/1.43) higher mortality risk than did men from the general population. Among the men exposed to violence, 2.35% died (any cause of mortality) during the follow-up period compared with 0.50% of the general male population (Table 1). Of the men exposed to violence, 0.65% died of suicide, 0.38% of another incident of violence, and 1.32% of other causes (Table 2). Corresponding proportions in the general male population were 0.14%, 0.02%, and 0.34%.

Table 1. Descriptive statistics: exposed group and general population sample

	Received hospital care (n = 8722)		General population (n = 78,702)
Proportion of deceased during the follow-up period (from any cause)	2.35%		0.50%
	Mean	Range	Mean	Range
Total survival time (time observed), days	3662	1–6210	3522	17–5844
Deaths/1,000,000 days	6.42		1.43
Age at inclusion in the study	21	18–24	21	18–24
Days of inpatient care from violence	2.5	1–389	–	–
Inclusion year	1998	1992–2005	1998	1992–2005

Table 2. Descriptive statistics for the year before inclusion in the study

		Received hospital care (n = 8722)	General population (n = 78,702)	P-value for difference (2-tests)
Variable		%	%
Country of birth	Sweden	88.68	93.19	<0.001
	Nordic countries	1.24	0.68
	Western countries^a	0.41	0.38
	Eastern Europe	1.05	0.71
	Former Yugoslavia	0.97	0.79
	Middle East	3.45	1.69
	Asia	1.26	1.23
	Sub-Saharan Africa	1.04	0.42
	Latin America	1.88	0.90
Degree of urbanization	Major cities	34.06	30.22	<0.001
	Medium-sized and small towns	47.87	50.90
	Countryside	17.98	18.86
	Not specified	0.09	0.02
Children, 0–6 years	Yes	8.27	7.10	<0.001
Gainfully employed	Yes	33.02	34.96	<0.001
Means-tested social assistance	Yes	30.45	11.75	<0.001
Hospital care for other reasons	Yes	0.55	0.18	<0.001
Education, highest level	Compulsory	54.13	38.38	<0.001
	Upper secondary	42.99	51.02
	University	2.88	10.59
Income, quartiles	Q1 (lowest)	25.48	24.88	0.034
	Q2	25.88	24.78
	Q3	24.46	25.14
	Q4 (highest)	24.19	25.20
Cause of death	Suicide	0.65	0.14	<0.001
	Violence	0.38	0.02	<0.001
	Other causes	1.32	0.34	<0.001

^aWestern Europe, Australia, Canada, New Zealand, and the USA.

Men who received inpatient care for violence differed significantly from the general male population regarding degree of urbanization, being parents of children aged 0–6 years, gainful employment, and income; they were more often born outside of Sweden (11.3% vs. 6.8%), had more often received means-tested social assistance the year before inclusion in the study (30.5% vs. 11.8%), had more often received hospital care for other reasons (0.6% vs. 0.2%), and had lower education (54.1% vs. 38.4%) than did men from the general population. However, there are only small differences in income between men who received and did not receive inpatient care after violence (Table 2).

Relative mortality risks are presented in Table 3. Controlling for demographics and hospital care (Model 1), the relative mortality risk due to suicide among men exposed to violence was more than five times higher (RR = 5.52) than in the general male population, while the relative mortality risk due to another incident of violence was about 20 times higher. The relative mortality risk from all causes of death in men exposed to violence was more than four times higher than in the general male population (Table 3). In the fully adjusted model, which also controls for employment, presence of preschool children, and socioeconomic factors (Model 3), the relative risks were slightly lower, but the general pattern remained the same.

Table 3. Relative mortality risk (all-cause mortality/mortality risk from suicide/violence) for males (aged 18–24 years) in hospital care after violence compared with men among the general population (reference category)

	Model 1	2	3
All-cause mortality
RR	4.55	3.64	3.34
P-value	<0.001	<0.001	<0.001
Mortality risk from suicide
RR	5.52	4.41	4.16
P-value	<0.001	<0.001	<0.001
Mortality risk from violence
RR	22.32	19.87	19.63
P-value	<0.001	<0.001	<0.001

Results based on Cox regressions.

The bold values indicate the p-values <0.05.

Model 1 controlled for age, age-squared, days of inpatient care from violence (non-exposed set to the average number of days), days-squared, degree of urbanization (the three major cities/medium-sized and small towns/countryside), country of birth (Sweden/the Nordic countries/Western countries [Western Europe, Australia, Canada, New Zealand, and the United States]/Eastern Europe/the former Yugoslavia/the Middle East/Asia/Sub-Saharan Africa/Latin America), and days of hospital care for other reasons; Model 2 also controlled for presence of preschool children, gainful employment/not, and financial assistance/not; Model 3 also controlled for highest education level (compulsory/upper secondary/university) and income quartiles (four groups).

Table 4 shows the temporal pattern of the mortality difference between those receiving hospital care and men from the general population. At the beginning of the follow-up period, the relative risk of all-cause mortality was 4.5 times higher among men who received hospital care due to violence than among men from the general population; the difference decreased significantly during the follow-up, becoming 2.7 after 10 years. The relative mortality from suicide remained fairly stable over time and did not change significantly: the RR was 3.9 at the beginning of the follow-up and 4.6 after 10 years. The relative mortality risk from later episodes of violence was extremely high (RR > 45) at the beginning of the follow-up period, but decreased significantly during the follow-up. After 10 years there was no statistically significant difference between the exposed group and men from the general population (Table 4).

Table 4. Relative mortality risk (all-cause/suicide/violence) for males (aged 18–24 years) receiving hospital care after violence compared with men among the general population (reference category)

	All-cause mortality		Suicide		Violence
Temporal pattern	RR	P-value	RR	P-value	RR	P-value
Average over follow-up^a	3.34	<0.001	4.16	<0.001	19.63	<0.001
At beginning of follow-up	4.48	<0.001	3.78	<0.001	45.23	<0.001
Change per year	0.95	0.028	1.02	0.709	0.73	0.007
After 10 years	2.66	<0.001	4.62	<0.001	2.06	0.468

The bold values indicate the p-values <0.05.

Results based on Cox regressions. Controlled for the same factors as Model 3 in Table 3: age, age-squared, days of inpatient care from violence (non-exposed set to average number of days), days-squared, degree of urbanization (the three major cities/medium-sized and small towns/countryside), country of birth (Sweden/the Nordic countries/Western countries [Western Europe, Australia, Canada, New Zealand, and the United States]/Eastern Europe/the former Yugoslavia/the Middle East/Asia/Sub-Saharan Africa/Latin America), days of hospital care for other reasons, presence of preschool children, gainfully employed/not, financial assistance/not, highest education (compulsory/upper secondary/university), and income quartiles (four groups).

^aThe same results presented as for Model 3 in Table 3.

Discussion and conclusions

Young males are the population group at greatest risk of being exposed to violence. This study examines the mortality risks of young men aged 18–24 in comparison with age-matched men among the general population. We study all-cause mortality, risk of suicide, and risk of death from later episodes of violence.

Men exposed to violence in the year before the episode of violence leading to hospitalization constitute a selected group with lower education, which indicates a difference in socioeconomic position. At the same time, however, there is no substantial income difference between those exposed and not exposed to violence. Young men in this age group are still in an establishment phase, when many start or continue with higher education, making comparisons somewhat unstable.

The main finding is that the risk of all-cause mortality and violence among men receiving inpatient care after violence is highest close to the period of inpatient care. In particular, the risk of dying from violence is extremely high at the beginning of the follow-up period, though it declines to insignificance after 10 years. In contrast, the risk of dying from suicide does not decrease over the follow-up.

In a previous study, we found that female victims of violence had an increased risk of premature death (Trygged et al., 2014a). This study finds that the same pattern a ppears among young men, even though the contextual conditions for violence toward women and toward men often differ. It is uncertain whether the process leading toward suicide starts with the violent incident, as violence may be combined with other factors. The increased risk of suicide indicates that exposure to physical violence may be a marker of poor mental health and/or lifestyle factors that trigger the risk. Among young men such factors could be excessive alcohol use, belonging to criminal gangs, or living under extremely insecure conditions, such as homelessness.

There is a strong relationship between crime and alcohol use, for both the perpetrator and victim. Besides, young men are often both perpetrators and victims of violence. According to the Swedish Crime Victim Survey (Brottsförebyggande rådet, 2017), in more than 6 out of 10 assaults, the perpetrator was influenced by alcohol and in 35% of assault crimes both the perpetrator and victim were under the influence of alcohol (Olseryd, 2015). Altogether, young men are overrepresented in all kinds of “negative statistics,” for example, concerning alcohol and drug use, unemployment, criminality, and suicide. We also note that young men born outside Sweden are more often victims of violence than are native-born Swedes, but we do not know whether this is related to their living environment, socioeconomic position, ethnicity, xenophobia, or other causes, as we do not know who the perpetrators are.

As exposure to violence has such negative effects, paying attention to the victims of violence and the consequences of violence could be important for suicide prevention. During hospital care, injuries are treated and documented, which may be important for juridical reasons, for example, in bringing cases to court or making insurance claims, but to our knowledge, there are no national guidelines in Sweden (besides regular medical ones) for acute medical services to male victims of violence. Instances of specific documentation and screening for violence usually refer to female cases.

The NBHW (2014) has developed instruments, such as FREDA, with which social services can screen for violence in intimate partner relationships. FREDA, which focuses particularly on women and children, consists of three standardized assessment methods to be used in social work and other activities concerned with people exposed to violence in close relationships. There is also a regular survey sent to all municipalities called Open Comparisons Related to Support of Victims of Violence, but violence committed by an unknown perpetrator (e.g., in public spaces, mostly related to men) is excluded from the questionnaire (NBHW, 2015b).

In Sweden, there is ongoing discussion of different forms of violence, such as “honor”-related violence (Brå, 2012), LGBT-related violence (Brå, 2016b), work-related violence (Edwards & Buckley, 2016; Vaez et al., 2014), and gang violence (Rostami, 2017), but the dominant perspective is that of intimate partner violence related to women and children. Comparatively, little research addresses men’s violence toward other men and male victimization, even though this is the most common expression of physical violence. There is also a lack of research into perpetrators as victims of violence, which might be related to a focus on the innocent “ideal victims” (see Christie, 1986; Strobl, 2004). The startling results of this study regarding outcomes for male victims of violence support the need for more attention to this type of violence and to suitable interventions for victims of violent crime as well.

Although many young men live in vulnerable circumstances, they may be hard to assist as they tend not to ask for help and to avoid contact with the authorities. However, when they are exposed to severe violence leading to hospital care it may be possible to intervene, if hospitals, the social services, and perhaps other authorities establish routines for approaching violated patients regarding their mental health and life situations. On one hand, the Social Services Act (2001:453, Chapter 5, 11§) states the responsibility of social services to support all victims of violence. On the other hand, the law explicitly says that social services should consider the need to support women exposed to violence, saying nothing about men, which may be viewed as a form of gender discrimination. A study based on focus group interviews with social workers suggested that the violence to which young men are subjected in cases of street violence and interpersonal violence almost seems unavoidable or even “natural” (Skillmark & Kullberg, 2019). The authors concluded that social workers’ attitudes toward young male crime victims may contribute to making these men appear to be less legitimate victims.

There seems to be no linkage between hospital care and the social services for adult men aged 18 years and older. These young men are unlikely to contact social services unless they do so for other reasons, such as for financial assistance. However, such intervention might be initiated by a hospital counselor as a means of suicide prevention, by offering psychosocial support, discussion of the violent event, and connections to the Swedish Crime Victim and Compensation Authority, a possibility that merits scrutiny in additional research. Although this is a study from a single country, its results may be relevant to other Western countries with similar population compositions. Further speculation is beyond the scope of this article, but the findings do identify severe consequences of violence against males that indicate a need for intervention.

Limitations

The register data used only cover individuals spending at least one night in hospital care and surviving the hospital stay. This criterion excludes both the small group of people dying before reaching hospital care or at the hospital, for example, due to shootings, and those with less severe injuries.

The present analyses are based on the comparison of all men in Sweden receiving inpatient care after violence with men randomly chosen from the population register of all people living in Sweden. Men randomly chosen from the population might have been hospitalized after violence before or after the year they were randomly chosen to be included in the study; however, this is unlikely to have affected the results since under 1/1000 of young men become hospitalized due to violence.

Ethics

The research application was approved by the Regional Ethics Review Board in Uppsala (Dnr 2017/065).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.