Beaten and poor? A study of the long-term economic situation of women victims of severe violence.

This 10-year follow-up study based on Swedish national registers compares the economic situation of women victims of violence leading to hospitalization (n = 6,085) to nonexposed women (n = 55,016) in 1992 to 2005. Women exposed to severe violence had a poorer financial situation prior to the assault. Violence seems to heavily reinforce this pattern, indicating a continued need of support from the social work profession. Assaulted women had a worse income development, lower odds for being in employment, and higher odds for having low incomes and means tested social assistance during the 10-year follow-up, independent of having children or not.

BACKGROUND

The prevalence of exposure to violence is increasingly being discussed as a public health issue. This is also recognized by the World Health Organization (Krug et al., 2002; World Health Organization [WHO], 2007, 2012). Women and men are exposed to violence, but there are important differences in where the violence takes place. To men, home is a place of safety. To women, home may be the most dangerous place of all. Men, especially those of young age, are often exposed to violence in public places by unknown perpetrators. Women, on the other hand, often suffer violence from a partner or someone else she knows (Brottsförebyggande rådet [BRÅ], 2009). The general difficulties of defining and studying violence against women as a public health problem are scrutinized by Ruiz-Pérez et al. (2007). In their focus on intimate partner violence these authors encounter problems in defining this type of violence, in finding data about the magnitude of the problem, related to lack of consensus with regard to measuring events, and more. Certain factors increase the risk of being exposed to violence. Research has shown a strong association between a reduced financial situation and several other negative circumstances, such as poor health, halting social networks, weak political resources and exposure to violence (Statens offentliga utredningar [SOU], 2000). Being young, unemployed, and divorced/single, especially those with children, increases the risk of becoming a victim of domestic violence (Nerøien & Schei, 2008). Other risk factors are substance/alcohol abuse (Lown et al., 2006; Tunving & Nilsson, 1985), homelessness, and physical or mental disabilities (BRÅ, 2009). These groups may also be more vulnerable to the effects of violence.

Physical violence against women affects their entire lives. Their autonomy and integrity is damaged, their relationships with family, friends, and working life become more difficult and their health is adversely affected. Women subjected to violence more frequently report poor health and physical complaints (Stenson et al., 2006; Sundaram et al., 2004; Wijma et al., 2007) than women not exposed to violence (Krantz & Östergren, 2000; Leserman et al., 1998). Women who are victims of domestic violence also have been shown to have a larger number of contacts with the health care services and an increased risk for depressive and posttraumatic stress disorder symptoms (Martinez-Torteya et al., 2009; Samelius, 2007).

An important issue is these women's financial situation. Assaulted women seem to have fewer resources in terms of economy and of education already to begin with. Low education goes together with low incomes that creates different kinds of dependencies, for example, on a partner and/or on economic assistance from the municipal social services. As is further discussed below, domestic violence may be presumed to have a further impact on this poor financial situation.

AIM OF STUDY

This study investigates the economic situation of women in Sweden who have experienced severe physical violence resulting in inpatient care. The research question is: What is the long-term financial situation after hospital discharge among women who have been exposed to severe violence compared to nonexposed women in the general population?

THE SWEDISH CONTEXT

The magnitude of violence against women in Sweden is uncertain. The number of persons claimed to have been victims of violence varies considerably between studies and is much dependent on how victim of violence is defined (see, e.g., SOU, 2004). If the definition also includes symbolic violence such as insults, threats, harassment, and so forth, the numbers affected increase (Lundgren, Heimer, Westerstrand, Kalliokoski, 2002). This study is limited to severe physical violence leading to at least one night of inpatient care. Each year in Sweden with its 9.4 million inhabitants, about 2,100 women are treated in hospital following undue violence inflicted by other persons and about 30 women die of their injuries (Leander et al., 2012).

As in many other national contexts, this is predominantly intimate partner violence, exercised by a person with whom the women have or have had a close relationship (Leander et al., 2012, p. 235). A number of circumstances are notable concerning violence in close relations. The majority of incidents take place indoors, often in the victim's own residence. Alcohol is often involved but is more common when the culprit is not the partner (BRÅ, 2009).

Sweden has a broad concept of income redistribution, pensions being the most costly among them and welfare benefits the most debated. For the individual of working age in Sweden there are basically three sources of income:

Income from work

Income from social insurance (level depending on previous earnings)

Social assistance (means-tested support)

1. Most people work and support themselves through paid work or self-employment. According to original data from Statistics Sweden (the Arbetskraftsundersökningarna [AKU] database), in 2010 the work participation rate among women varied between age groups from about 39% (age 15–24) to 84% (age 35–44 and age 45–54), declining again to about 67% in the age group 55 to 64, but these figures vary somewhat from year to year.

Generally, women earn less than men. Many women, particularly those with small children, work part-time. Women and men also to a considerable extent work in different sectors of the labor market. Together these factors largely explain the wage differences (for more detailed information on Sweden, see e.g., National Mediation Office, 2009, 2010; for Sweden in a comparative perspective, see e.g., Korpi et al., 2010). Government benefits are granted to individuals regardless of their family arrangements; taxes are progressive and related to income, but because all income tax is individual and not family based, there is an incitement for both spouses to work; and as marriage/cohabitation therefore is less based on economic necessity, this may be a factor in the greater instability of partnerships.

2. Out of work persons are mostly entitled to individual, income-related insurance such as sick leave or unemployment benefits. However, this presupposes a previous income for a certain amount of time.

3. Persons with no or very low income are generally entitled to social assistance. Social assistance is different from other benefits in that it is needs based and related to household (not individual) income, presupposes no property or possessions of notable value, is distributed through the municipality after means testing by a social worker, and is considered a last resort, when no other options are available. Several studies point to need of social assistance as an important marker of situations characterized by different forms of extreme vulnerability (see, e.g., Gustafsson, Zaidi, & Franzén, 2007; Ringbäck Weitoft, 2003).

Other complementary support systems are general child allowances and income-tested housing allowance. The social insurance office may also pay a child care allowance to a child minder if that person has a low income and if the other parent cannot afford or refuses to pay the required child support until the child is age 18. There are, besides, also subsidized services, childcare being one of the most important among them.

METHOD

Inclusion Criteria and Sampling Frame

This is a study based on national register data. The study population consists of all cases in the National Patient Register (NRP) of Swedish women age 18 to 64 who had at least one night of inpatient care following brutal assault during the period 1992 to 2005. The comparison group consists of a random sample from the Statistics Sweden (SCB) national population register (RTB) of 55,016 age- and year-matched women. Altogether, 6,085 women victims of violence are compared to 55,016 nonexposed women.

In Sweden, every citizen has a personal identification number on which all registers containing personal information are based. Using the personal identification number, the exposed and the unexposed group was matched with annual data from Statistics Sweden (the longitudinal integration database for health insurance and labor market studies [LISA] database; http://www.scb.se/Pages/List_257743.aspx) covering income (all sources), education, marital status, cohabitation, and number of children. These data include information from the year prior to the assault that resulted in inpatient care and up to 10 years after. The cases were followed up at most until 2006 or until the year the women under study passed age 64, meaning that for women younger than age 55 with incident assault in the years 1992 to 1996, the follow-up period is up to 10 years. For those with incident assault in 1997, the follow-up is up to 9 years, and so on until those with incident assault in 2005 have a follow-up time of only one year. This also means that women with incident assault at age 63 were followed up for one year, whereas those who were age 54 or younger were followed up for up to 10 years. The average follow-up time was 4 years. About 35% of cases—2,121 exposed and 20,864 nonexposed women—were followed during the maximum time of 10 years (Table 1).

TABLE 1

Number of Observations in Different Follow-Up Years

Year	Exposed	Population
−1	6,085	55,016
0	6,085	55,016
1	5,953	54,777
2	5,499	50,844
3	4,996	46,685
4	4,520	42,627
5	4,090	38,821
6	3,640	35,108
7	3,246	31,572
8	2,834	27,622
9	2,472	24,312
10	2,121	20,864

Note. Year − 1 means the year prior to the violence that lead to inpatient care/beginning of the follow-up period, year 0 = the year of the violence/year the follow-up starts, year 1 = the first year after the violence/the first whole follow-up year, and so on. Average follow-up time was 4.1 years.

Figure 1 shows the incidence of inpatient care after assault (%) for 1992 to 2005.

FIGURE 1

Incidence of inpatient care (%) due to violence.

Statistics

Analyses were based on logistic regression with the exception of income change, which was based on linear regression. All analyses build on comparisons between women who were assaulted leading to inpatient care and women not exposed to such violence. Women with children who were assaulted were compared to non-exposed women with children and women without children who were assaulted were compared to non-exposed women without children. The only analyses where all women were included are the ones regarding income change. When the odds for being gainfully employed were analyzed, only women that were gainfully employed in the year prior to the assault/beginning of the follow-up period were included. When low income was analyzed, only women that did not have low income prior to the assault were included. When means tested social assistance was analyzed, only women without means tested social assistance prior to the assault were included.

Income change is analyzed with 10 linear regressions, one for each follow-up year. Gainful employment, low income and means tested social assistance are analyzed with 10 logistic regressions each, one for each follow-up year. Year −1 is the year prior to the violence that lead to inpatient care/beginning of the follow-up period, year 0 the year of the violence/year the follow-up starts, year 1 the first year after the violence/the first whole follow-up year, and so on. The population is set to 1.0 in all analyses.

All analyses are controlled for age, age squared, year (from 1992 to 2005) given dummy representation, country of birth (Sweden/the Nordic countries/Western countries/Eastern Europe/the former Yugoslavia/the Middle East/Asia/Africa south of the Sahara/Latin America), degree of urbanization (the three major cities/big and medium-size cities/small cities and countryside/not specified), education, number of days of inpatient care for other health problems, number of days squared, and cohabitation or not.

Dependent Variables

Gainfully employed is defined as having an income from work of at least 60,000 krona (SEK) per year (yes/no) expressed in 2004 prices (approximately US$8,400). Income change is defined as relative change in disposable income in relation to the year before incidence or sampling and is also expressed in 2004 prices. Low income is defined as having an individual income of less than 60% of the Swedish median income each year (yes/no). Receiving social assistance is a dichotomous variable (yes/no) and is defined as having been granted social assistance at some time during the year.

Procedures and Ethics

Information taken from the (Swedish) National Board of Health and Welfare register of inpatient care for all women exposed to violence leading to hospitalization during the period 1992 to 2005 was added to other population data such as income and education at Statistics Sweden. The new joint database was then deidentified before being delivered to us. The study deepens results presented in Social rapport 2010 (National Board of Health and Welfare, 2010), a national report on social conditions in Sweden (approval to use data Dnr 11-1464/2009).

RESULTS

Descriptive statistics show that the women in the study exposed to violence differ substantially in some regards from the general population already in the year before the assault that led to hospitalization (Table 2). These women who were assaulted were younger than the nonexposed women and were more often foreign born (>20% vs. 10%). Also, among the exposed women 56% were cohabitating compared to 74% among the nonexposed (not shown). They less often had higher education (9% vs. 28%), were less often gainfully employed (30% vs. 65%) and—particularly the assaulted women without children—more often earned less than 60% of the Swedish median income. Forty-eight percent of the assaulted women with children received means tested social assistance, compared to 8% in the comparison group.

TABLE 2

Descriptive Statistics for Women that Have Received Inpatient Care Due to Violence and Women in the Population, with and without Children

	Exposed to Violence		Population
	Having Children[a]		Having Children
	No	Yes	No	Yes
Age
18–24	20.7	18.6	17.1	9.6
25–44	38.4	69.3	25.3	72.0
45–64	41.0	12.1	57.5	18.4
Country of birth
Sweden	80.4	72.3	89.3	87.6
Nordic countries	10.7	7.5	4.6	3.6
Western countries[b]	1.0	0.8	1.0	0.9
Eastern Europe	2.0	1.7	1.2	1.3
Ex-Yugoslavia	1.1	2.3	1.3	1.5
Middle East	1.8	8.4	0.9	2.6
Asia	1.3	2.4	0.8	1.1
Africa south of the Sahara	0.5	2.4	0.2	0.5
Latin America	1.1	2.3	0.5	0.9
Degree of urbanization
Major cities	37.2	35.7	33.4	31.0
Medium-sized and small towns	46.5	46.7	47.9	48.1
Countryside	16.1	17.5	18.7	20.9
Not specified	0.2	0.1	0.03	0.03
Inpatient care for other health problems
Yes	2.8	1.3	0.3	0.3
No	97.2	98.7	99.7	99.7
Income, quartiles[c]
1 (lowest)	38.2	21.6	26.0	17.1
2	33.4	16.4	25.7	19.6
3	18.2	26.4	25.8	30.1
4 (highest)	10.2	35.7	22.5	33.2
Low income[d]
Yes	24.8	16.4	17.5	12.8
No	75.2	83.6	82.5	87.2
Education
Compulsory	45.5	46.4	27.2	19.3
Upper secondary	45.3	44.6	46.4	51.0
University	9.2	9.0	26.4	29.7
Gainfully employed[e]
Yes	30.4	33.9	65.7	67.3
No	69.6	66.1	34.3	32.7
Means tested social assistance
Yes	43.6	48.1	5.6	7.9
No	56.4	51.9	94.4	92.1
Cohabiting[f]
Yes	34.8	95.8	56.8	97.1
No	65.2	4.2	43.2	2.9
Children 0–6 years
Yes	—	52.9	—	49.8
No	—	47.1	—	50.2
Total	100	100	100	100
Number of persons	4,007	2.078	29,129	25,887

Note. Information from the year prior to the period of inpatient care for violence/beginning of the follow-up period.

With/without children (age 0–17) living at home in the year of the violence that led to inpatient care.

Western Europe, Australia, Canada, New Zealand, and United States.

The income quartiles are based on the incomes of all Swedish women age 18–64 in the year of inpatient care.

Defined as having an income below 60% of the median income.

Defined as having an income from work of at least 60,000 SEK per year. Expressed in 2004 prices (approximately US$8,400).

Shows whether cohabiting or not at the end of the year. Only cohabiting with mutual children is registered in Swedish registers. The children do not need to be living with the parents.

Financial Situation during Follow-Up

During the follow-up, several differences were noted between the women exposed to violence and the nonexposed women regarding labor market participation, low income, proportion receiving (means-tested) social assistance, and income change.

The percentage of women exposed to violence were also found to have a negative income change compared to nonexposed women and the relative difference becomes larger during the whole follow-up period. Whether there are children in the picture has a further impact. Women with children had a poorer income development than those without children (see Figure 2).

FIGURE 2

Relative difference in income change. Results from 10 linear regressions, one for each follow-up year. Women with children who were exposed to violence leading to inpatient care are compared to nonexposed women with children. Analyses are controlled for age, baseline year, country of birth, degree of urbanization, education, number of days of inpatient care for other health problems, and cohabitation or not.

The odds of being gainfully employed, among women that where employed before the assault, were about 4 times lower among women exposed to violence compared to nonexposed women. The risk of not being gainfully employed remained low during the whole 10-year follow-up For example, in year 5 of follow-up, 55.5% of the women who were assaulted had paid work compared to 77.7% among the nonexposed group (not shown).

The odds of having low income, among women not having low income before the assault, were doubled during the whole follow-up for women without children, whereas women with children initially did not have the same increase in risk. Later during follow-up, women without children also had doubled odds (see Figure 3). This means for instance that in year 5 of the follow-up, 14.5% of the assaulted women had a risk of low income compared to 10.9% among the nonexposed (not shown).

FIGURE 3

Odds for being gainfully employed and having low income (less than 60% of the median income). Results from two times 10 logistic regressions, 10 for each outcome—one for each follow-up year.

Exposure to violence seems to reinforce the need for social assistance. Among women not receiving means-tested social assistance before the assault, the likelihood of receiving social assistance during follow-up was several times higher among women exposed to violence and especially high among those without children (see Figure 4). For instance, in year 5 of follow-up, 44% of the women who were assaulted compared to 4% among nonexposed women received social assistance (not shown). This heavily increased risk (odds ratio = 8–9) remained up to 10 years after the abuse (see Figure 4).

FIGURE 4

Odds for means tested social assistance.

DISCUSSION

This study is a register-based investigation of the long-term economic consequences for women of being exposed to severe violence that only includes cases ending in hospitalization. The use of such other sources as police reports would have involved a higher risk of selection bias because women in the higher social strata are presumably more able to dodge being recorded by the police. We circumvented such selection bias by using population data with all women that received inpatient care. The women exposed to severe violence are compared to age matched women from the population—exposed women with children are compared to nonexposed women with children and exposed women without children are compared to nonexposed women without children.

The results show that women who were assaulted have a much worse development of their financial situation than other women. Women who are victims of severe violence less often have paid work, more often low income, lower income growth, and much more often receive social assistance up to 10 years after the assault. In sum, all indicators of their financial situation show a very negative development compared to the comparison group.

However, the assaulted women differ from the comparison group already before their inpatient care. They had lower education, lower income, were less often gainfully employed, received social assistance far more often, and more often were single (not cohabiting). We have controlled for the effect of these factors in the analyses, but it may be so that this control did not capture the entire effect of poor resources or a weak position on the labor market. If the victimization is combined with addiction and/or exposure to other such risky environments it may well be compounded.

We are able to describe the situation preceding the incident of violence that led to hospitalization but do not know whether this is the first incidence of exposure to violence in each woman's life. However, there are good reasons to assume that the incident leading to hospital care has been preceded by several occasions of violence with less severe physical injuries. For example, one study focusing on intimate partner violence found that violence tends to become a normalized, integrated part in the relationship (Hydén, 1994). Therefore, the differences we found may be the consequence of repeated incidents of violence. Such a life situation is characterized by severe stress and often includes psychological as well as physical ill health. The women in this study live in a complex and sometimes chaotic situation where labor market participation may be further compromised. In the worst case scenario, the woman may have to hide from a violent partner, perhaps in a shelter. This could hinder her from going to work where she would run the risk of being identified and threatened or further assaulted.

The role of the financial contribution of a husband or partner represents a particular difficulty in the evaluation of different factors affecting the economic situation of assaulted women. Generally, living with a husband or partner means living on two incomes and consequently being in a better financial position. On the other hand, economic control can be a very sophisticated means of oppression, leading to an increased dependency on an abusive partner. The woman may be prevented from having a job or be excluded from any command over the household economy, for example, the husband/partner may take out joint loans or loans in her name alone and then default on payment, leaving her with “inherited” debts. This is also a recurrent theme in the power and control wheel model for understanding domestic violence. The so-called Duluth model (Domestic Abuse Intervention Project, n.d.) shows different control strategies used by the (male) abuser, of which economic abuse is an important part.

Further, in the event of separation, unmarried couples do not automatically share belongings. In injurious relationships, the partner may even destroy goods and items of material worth along with items of affective value to the other; so that a person may all of a sudden find herself with neither belongings nor any place to live. The legal system has gradually adapted to the factual situation of many cohabitants, particularly if there are mutual children, but there are still profound differences in equality of treatment compared to married couples.

A woman who finally separates from the man who has abused her improves her general situation in many ways by the separation, but separation is also likely to have negative economic consequences, at least as described by the register data. Although economic incitements to stay in a dangerous relationship may be weak, a clearly deteriorating economic situation after divorce is evident from earlier research (Avellar & Smock, 2005; Finnie, 1993; Jarvis & Jenkins, 1997).

With regard to low-income risk, women with children who were assaulted are more similar to mothers who were not assaulted than to women without children who were assaulted during the first 5 years of the follow-up. In spite of this, the likelihood of receiving means-tested social assistance was much higher among assaulted women, independent of if they had children or not, with the odds consistently being about 8 times higher during the follow-up. For women without children, the selection might be of women already living in even more difficult life circumstances because they work less than other women and tend anyhow to have low incomes. Women with children may not have extremely low income but may yet not have enough for adequate child support. In this study we focus on the women, but obviously the children of mothers who were assaulted need special attention in a number of ways. In these vulnerable families economic hardship may also be a sign of child poverty.

Given these results, we propose that being battered by a spouse or an ex-spouse, most probably during an extended period of time with all the attendant ramifications, in combination with background factors impeding a stable position on the labor market, offers some explanation for the differences found in this study.

Generalizability of Results

This study only scrutinizes violence to women resulting in severe physical injuries. To what extent can the results be generalized to all women who are assaulted? To asses this, we compared information in various studies on the women reporting assault with different Swedish data sources. Among women reporting exposure to partner violence in a national safety survey (BRÅ, 2011), 27% had higher education, whereas among women reporting exposure to any form of violence the percentage with higher education was 19%. In the same study, women reporting severe violence, a lower proportion, 16% had higher education. Among women seeking help at a shelter for battered women in the Swedish university town of Uppsala, 28% had higher education (Stenson et al., 2006)—these statistics to be compared to those of this study where only 9% of the women who were assaulted had higher education. The low proportion of persons with high education (16%) in the National Safety Survey possibly indicates an even stronger selection in this study. This suggests that women subjected to severe violence resulting in hospital care differ from all assaulted women and may be described as a less resourceful group.

An Important Field in Social Work

Violence against women in intimate partner relationships is a public health problem globally, also in industrialized countries such as Sweden, the United States, and others (see, e.g., Balvig & Kyvsgaard, 2006; Canadian Centre for Justice Statistics, 2005; Finney, 2006; Haaland et al., 2005; Heiskanen & Piispa, 1998; Helweg-Larsen & Fredriksen, 2007; Heuni, n.d.; Mouzos & Makkai, 2004; Tjaden & Thoennes, 2000). Further, this is a problem that involves community members and should therefore be treated as a community responsibility; not only as an individualized crime.

There are many things that could be done within in the health care sector to avoid repeat hospitalization and fatalities; for example, better routines in emergency rooms for identifying women who were violated, work with treatment and documentation, better information about women's aid, better enforcement of legal protection, and better sheltered housing (Lindblom et al., 2010).

There are also many things that could be done within the social services to improve preinpatient care (e.g., National Board of Health and Welfare, 2009a, 2009b). The study shows that most posthospitalization financial situation of women who are battered is very weak. Even though it is not possible to generalize from these severe cases of violence to all types of violence and assault against women, the economic effect of violence is a significant problem with important implications for the social work profession. Being a long-term recipient of social assistance is common in this group and may be an indicator of exposure to violence. The National Board of Health and Welfare (2006, p. 35) has shown that women with long-term dependence on social assistance run an 18 times higher risk of becoming hospitalized due to assault than the general population. This is surely a hint for the social worker to ask the long-term client whether she is faced with violence and have preparedness to support the woman if the answer is yes. The kinds of assistance that may alleviate a poor financial situation could be such services as finding alternative housing, assistance with getting employment, and giving money and debt-handling advice (Trygged, 2012).

The study raises two issues with implications for the development of policy in this area:

From a research perspective it would be useful with more documentation on the part of the health services concerning perpetrators. From whom does the violence emanate? A partner/ex-partner or from someone else? This would be relevant for further understanding of whether it is particularly the intimate partner violence or violence in itself that reinforces the difficulties.

Because the results are very important from a social work perspective, they raise the question of whether social workers should carry out a general screening of all long-term female clients to find out whether they are victims of violence. Such knowledge could help in understanding the women's situation and give the chance of offering adequate support.

LIMITATIONS

This is a study on women survivors of severe violence leading to hospital care, not a mapping of the incidence of violence against women in general, so there is no information about cases that have not resulted in hospitalization.

We used a strong database. All treatment in hospital care is registered, and the system of obligatory personal ID numbers facilitates research. This makes it possible also to link to other population registers. Much research in Sweden is based on these registers. For many diagnoses there are validated quality registers (stroke, cardiovascular diseases, cancer subtypes and many others), but we have not found the same systematic work done on the diagnosis in this study. We have no reason to believe that there is a bias in the registration, but because we do not know for sure, there is some remaining insecurity in the validation of our results.

A limitation to the generalizability of results is the lack of knowledge about the perpetrators and thereby different types of violence. According to investigations by Swedish authorities, severe violence toward women is mostly in the context of close relationships (BRÅ, 2009), but we do not know the full extent of partner violence versus other forms of violence, for example, related to the pursuit of entertainment or in working life. We do know that women generally run a greater risk of being exposed to partner or family violence than to violence in other contexts, but we have no adequate information on the abusers themselves (Krantz & Garcia-Moreno, 2005).

CONCLUSIONS

Women victims of severe violence have a very weak financial situation up to 10 years after an assault leading to hospitalization. This may be directly related to the assault or may be an effect of a wider set of harmful living conditions, including repeated maltreatment. It may also be related to more general characteristics of the assaulted group, for example, low education and low income. From a social work perspective it is important to recognize a likely need for more extensive social assistance than mere income support even if many years have elapsed since the assault.