Abstract
Background:
The health impact of violence against women by perpetrators other than intimate partners has received little attention. This study aims to analyze the effect of different forms of interpersonal violence on women's health.
Methods:
Adult women (10,815) randomly sampled from primary healthcare services around Spain were included. Women were grouped as follows: (1) no history of violence, (2) history of intimate partner violence only (IPV), (3) history of non-IPV only, and (4) history of both IPV and non-IPV. Lifetime prevalence of violence by IPV, non-IPV, and both was calculated. Adjusted multivariable regression analysis was performed to assess the effects of the different forms of violence on women's health status.
Results:
Of the women, 32.7% experienced lifetime violence. Poor self-perceived health, psychological distress, co-occurring somatic complaints, and use of antidepressant or tranquilizer medication were significantly higher for women with a history of violence than for women with no history of violence. Women who reported both types of violence, IPV and non-IPV, were almost five times more likely to suffer psychological distress and co-occurring somatic complaints and > six times more likely to use medication than women with no history of violence.
Conclusions:
The high prevalence of violence and its consistent association with a wide range of women's health problems suggest that violence seriously compromises women's health. Health providers should ask their female patients specifically about their history of violence, both IPV and non-IPV. Including this in patient's assessment would lead to more informed clinical decisions and more integrated care.
Introduction
Violence against women is now well recognized as a public health problem and human rights violation of worldwide significance. 1 The term “violence against women” encompasses multiple kinds of abuses directed at women and girls over their lifetimes. 2 Violence can be physical, sexual, or psychological. Perpetrators include intimate partners, family members, friends, coworkers, and strangers. 3 Violence occurs in women's homes, in educational institutions, at work, in other social settings, and in myriad public places. A previous study, using a population-based national sample, estimated that 60% of women have experienced at least one form of violence in their adult life. 4
Research has documented the impact of violence on women's health. Depression, anxiety, suicidality, posttraumatic stress disorder (PTSD), substance dependence, chronic pain, and somatic symptoms, among others, have been reported most frequently. 5 –11 Victimized women view themselves as being less healthy and having more physical complaints and symptoms of emotional distress than women who have not been victimized. 12
Violence also has long-term health consequences that persist beyond the period of abuse, 13 which may include physical symptoms, mental illness, and poor social functioning. 14 In addition to the impact on individual health and well-being, violence against women takes an enormous social toll in terms of medical costs, 15 decreased work productivity, 16 and an increased burden on the justice, social, and health systems. 17
Many studies have examined the health consequences of violence by an intimate partner (IPV), as it is probably the most prevalent type of violence against women on a worldwide scale. 18 Despite the significance of violence against women as a public health problem, few studies examine the health impact of violence perpetrated by persons other than intimate partners. Moreover, as most of the studies on violence and health come from English-speaking countries, the information from countries in Southern Europe is still limited.
We aim to (1) measure the prevalence of nonpartner violence, (2) identify the relationships involved in nonpartner violence (non-IPV), (3) compare the health of women with no history of violence with that of women with a history of IPV only, non-IPV only, and both types of violence. We hypothesize that women with any history of violence will have a prevalence of both physical and psychological health complaints that is 30% higher than that of women who have no history of violence.
Materials and Methods
Study sample and data collection
Data from a cross-sectional survey aimed at estimating the prevalence of violence and associated factors among adult Spanish women were used. Six research groups were involved in this broad research project, which was approved by the corresponding ethics committee in each group. 19
Independent and representative samples were taken in each of the Spanish regions based on a multistage cluster sampling scheme. Spain has 18 regions, and each region is made up of provinces (52 in total). Primary sampling units were the towns of each province. Secondary sampling units were primary healthcare services in the towns, and tertiary sampling units were the physicians in those services; 989 health professionals were asked for collaboration, and 605 (61.2%) in 547 primary healthcare centers accepted. Final sampling units were women aged 18–70 years randomly selected by physicians during 2006–2007 according to the scheduled time of the appointment. For a confidence interval (CI) of 95%, an expected lifetime IPV prevalence of 30%, a standard error (SE) of ± 4%, and a design effect of 1.30, we calculated that the sample size required was 11,808, 656 women for each region.
Women were considered ineligible if they were illiterate, did not understand Spanish, or had severe cognitive disabilities that impaired the completion of a self-administered questionnaire. According to World Health Organization (WHO) recommendations for domestic violence research, 20 women who attended the practice with a male partner were also excluded. Women who met the eligibility criteria were invited by their physician to participate in the study once the consultation was over. If they gave informed consent for participation, a questionnaire was handed to them in an envelope. Inside the envelope was information on community resources for battered women in their areas. Women filled in the questionnaire and placed it inside the envelope in a closed box in the waiting room. This ensured the anonymity of the responses. The number of women approached was 16,419, and 73% accepted. The final sample for the present study was 10,815, after excluding 467 women who did not meet the sampling inclusion criteria and 660 (5.8%) who did not complete the violence- related questions. No differences were observed between women who responded to the violence-related questions and those who did not.
Survey instrument and measures
Data were collected by a self-administered questionnaire that contained 27 sections and could be completed in approximately 15 minutes. All women were made aware that their participation was voluntary and that the confidentiality and anonymity of their responses would be respected.
Violence-related variables
Lifetime IPV
A woman was considered to have ever experienced IPV if she answered that she had been exposed to violence from a partner in the previous year (current) or before the previous year (past).
Current IPV
This was measured using the Spanish version of the Index of Spouse Abuse (ISA), 21 a 30-items scale with two independent subscales with scores ranging from 0 to 100: physical (which includes two items of sexual violence) and psychological. A woman was considered to have experienced current IPV if she scored above the cutoff points in any of the two subscales (6 for ISA-Physical and 14 for ISA-Psychological).
Past IPV
The woman was asked if any intimate partner had abused her physically, sexually, or psychologically before the previous year. These three questions had shown high comprehensibility and acceptability when used in a previous study. 22 It was considered that a woman had been victim of abuse in the past if she answered many times or sometimes and unexposed if she answered never to any of these questions.
Lifetime violence by perpetrators other than a partner
It was considered that a woman had been a victim of abuse by other perpetrators when she reported having been abused physically, sexually, or psychologically in her lifetime by a nonpartner and nonabused when she answered never to all related questions. The list of nonpartners provided in the questionnaire included family members, coworkers, boss/superiors, college/university colleagues, friends, neighbors, and unknown persons. Given that questions were not mutually exclusive, we grouped violence perpetrated by nonpartners according to the similarity of their settings, using the Two Step Cluster analysis. Four settings were identified: family, social, workplace, and community.
Violence by both (partner and other perpetrators)
It was considered that a woman had been a victim of abuse by different perpetrators when she reported having been abused physically, sexually, or psychologically by a partner and by a nonpartner.
Health indicators
Four health indicators were selected based on literature that examines the long-term negative consequences of violence: self-perceived health, psychological distress, co-occurring somatic complaints, and use of medication (tranquilizers or antidepressants).
Self-perceived health
This was based on the dimension of health perception of the SF-36, Spanish version. 23 Women were asked to describe their general health as very good, good, neither good nor bad, bad, or very bad. The variable was dichotomized by combining the categories neither good nor bad, bad, and very bad to indicate perceived poor health. Very good and good were also combined.
Psychological distress
This was measured using the General Health Questionnaire (GHQ-12), which has been internationally validated 24 and has been used extensively with women and in studies on the impact of violence. 25 The scale considers anxiety, depression, and self-esteem as experienced in the last month and contains 12 items with four response options. It provides a final continuous score between 0 and 12 and scores ≥3 indicate the presence of psychological distress. 26
Co-occurring somatic complaints
This was defined as the presence of coexisting or additional somatic health problems in women with psychological distress. 27 The following physical health problems over the past year were explored: headaches, migraines, kidney or urinary complaints, gastrointestinal disorders, neck, shoulder, or back pain, and menstrual and gynecological problems. In psychologically distressed women with co-occurring somatic complaints, the response was coded as 1, and the response was coded as 0 when no co-occurring somatic complaints were reported.
Use of medication
Self-reported use of tranquilizers or antidepressants over the past 12 months was recorded.
Social support
Social support was measured using a single question that explores the availability of specific tangible help for a particular situation 28 : How many people can you really turn to when you have a problem or difficulty? Answers were coded as a dichotomous variable of value: 0, no support (no-one to turn to for support), and 1, support (one or more). Finally, the following sociodemographic variables were collected: age, marital status, children, occupation, education, monthly family income, and country of origin (coded as Spain or other).
Statistical analysis
Sociodemographic characteristics among women with a history of violence (by partner, nonpartner, and both) were compared with characteristics of women with no history of violence. Prevalence of lifetime violence by different perpetrators from different relationships settings was calculated. To assess the impact of each violence relationship setting on each outcome variable, logistic regression models were fitted. Odds ratios (ORs) were adjusted for age, education, marital status, monthly household income, and native country of origin. The extent to which the precision of the results might have been affected by cluster sampling (design effect) was also explored with the statistical software STATA (Stata Corp., College Station, TX)
Given that the global sample used in this study is not proportional to the population size in each region, we used the inverse of inclusion probability multiplied by the sampling fraction as the weighting factor for all analyses.
Results
Of the 10,815 women who met inclusion criteria and completed the questionnaire, 3,538 (32.7%) reported having been exposed to an abusive situation during their lifetime. Of these, 52.8% indicated abuse perpetrated by a partner only, 30.5% by a nonpartner only, and 16.7% reported abuse by both partners and nonpartners. Nonpartner perpetrators were predominantly family and friends/colleagues (Table 1).
The categories of abuse are mutually exclusive.
Unclassified: n (%) = 16 (0.4).
Prevalence corrected by weight sampling.
CI, confidence interval.
Table 2 shows the characteristics of nonabused women and women abused by a partner only, nonpartner only, and both. Women who experienced violence by a partner and by both were more likely to be divorced at the time of the study, to have lower education and lower monthly income, and to have been born outside Spain than those who never experienced violence. Women with any history of violence reported significantly more often than women without a history of violence not having anyone to turn to support.
Reference group for all bivariate analyses.
The results of the multivariate analysis are shown in Table 3. After adjusting models by age, education, monthly household income, marital status, country of origin, and social support, the ORs indicated that women with a history of violence reported significantly poorer physical and mental health than women with no such history. Women who reported IPV only were three times more likely to suffer from psychological distress and co-occurring somatic symptoms and twice as likely to use medication (antidepressants or tranquilizers) and to consider their health as poor as nonabused women. The odds of poor health were higher for women reporting each type of non-IPV than for women with no violence but lower than for women reporting IPV.
Totals differ due to missing data.
The categories of abuse are mutually exclusive.
AOR, adjusted adds ratio, adjusted for age, education, monthly income, current marital status, social support, and country of origin.
Odds ratio corrected by random effect and weight sampling.
Compared to women who reported no history of violence, women who reported both IPV and non-IPV were almost five times more likely to suffer from psychological distress and co-occurring somatic symptoms, more than six times more likely to use tranquilizers or antidepressants, and more likely to perceive their health as poor.
Discussion
One in three women who attend primary healthcare for any reason in Spain has been a victim of violence at some point during the course of her lifetime. Current or former partner incidents and most of the nonpartner assaults involved perpetrators known to the victim. Assault by strangers comprised <4.0% of all violence against women in this sample. Overall, these findings point in the same direction as the few previous studies focused on interpersonal violence against women, although the reporting of interpersonal violence varies among studies, ranging from 20% to 60%. 4,25,29 This variation may be explained by the definition of violence and how violence has been assessed, 30 culture, age distribution of the study population, length of the observation period, and sampling extraction method. 31
Although health data refer to the time of the survey and violence may have occurred earlier, women with any history of abuse reported significantly worse current health status than nonabused women regardless of the perpetrator. Abused women are more likely to suffer from psychological distress, co-occurring somatic symptoms, and chronic pain; to consider their health as poor; and to use antidepressants or tranquilizer medication. Similar results were found in studies that compared the association between lifetime IPV exposure and current health. 17,22,29,32 Although the associations appear to be stronger when the perpetrator is a partner than when the perpetrator is a nonpartner, abuse by others also negatively impacts women's mental and physical health. Similar findings were reported by previous studies that analyzed mental health among women recruited in primary care services in European countries. 25,31 Even so, the range of forms of violence and the time frames considered limited direct comparison between studies.
Although psychological violence is far more frequent than other types of violence, 33 most previous research inquires only about severe physical and sexual IPV. Recent studies documented that long-term enduring psychological abuse, even in the absence of a history of physical or sexual abuse, was associated with poorer physical and mental health, 34,35 as well as limitation in social functioning. 36 Our inability to establish the different types of violence (physical, sexual, and psychological) experienced in each relationship setting requires cautious interpretation and should be considered in further studies.
Prior studies have shown that the magnitude of the relationship between IPV and adverse health increases as the number of violence types increases, 37,38 but few studies have assessed the health impact when violence was inflicted by very different types of perpetrators. We found that women who experienced abuse by different perpetrators reported more psychological distress, co-occurring somatic complaints, and use of antidepressants or tranquilizers than women who experienced abuse by one perpetrator. Therefore, the negative health consequences could be more pronounced in women who accumulated abuse by different perpetrators than for women who experienced IPV only. Given that medication may be taken to relieve the many mental and physical symptoms experienced by women with a history of violence, the relatively high rate of medication use among abuse victims may be understandable. 32
The design of our study does not allow for inferences on the causal relation or the mechanism of association between violence and mental and physical health. However, our data are consistent with the hypothesis that violence has a direct and negative impact on women's health. Rates and impact of interpersonal violence have been suggested to be among the most important indicators of women's future health. 39 Several studies, including this one, have shown that women who experienced abuse were more likely to be divorced and to report low income and lack of social support, 25,40 conditions that are themselves associated with poor physical and mental health. These social conditions could be consequences of violence and could explain, at least partially, the increased risk of poor health outcomes for women victims of violence.
As in other countries, there is a significant and controversial debate in Spain about whether healthcare providers should universally screen for violence and abuse among women. No studies to date, however, have evaluated the effectiveness of screening to improve women's health. 41 In addition, the health sector has been criticized for providing less than optimal care to women exposed to violence. This is in part due to perceptions that abuse is rare in their practices as well as specific barriers to addressing violence. 42 Injuries, which have been viewed as the most obvious indicator of abuse in clinical settings, may not identify women who have long-term health problems related to abuse. If clinicians are expected to appropriately identify and respond to abused women, they must be provided with relevant education. 43 This is becoming increasingly available in some areas in Spain, but it is not yet available to all health providers. Recognition of the underlying connection between lifetime violence and current health is important so that it can be discussed explicitly with the female patient and appropriate treatment and referral options can be agreed on.
The generalizability of our results is limited by the fact that our sample includes only patients accessing primary care services. Women who do not visit primary care clinics may be systematically different from women in our sample. It is not clear, however, how such differences would influence violence and its association with poor physical and mental health. Another limitation is our exclusion of women accompanied by a male partner. This criterion may have excluded women whose current partners are more coercive or abusive than those who attend alone. Finally, we did not specifically ask about violence during childhood or adolescence, which may have led to a conservative estimate of the prevalence of lifetime violence among women in our sample.
Despite these limitations, this study addresses violence against women by two very different types of perpetrators, presents new epidemiological data from Spain on the prevalence of violence by different perpetrators, and describes the health consequences of violence perpetrated by someone other than a partner. This topic is of interest to researchers and practitioners alike.
Conclusions
Non-IPV is almost as prevalent as IPV, yet it receives very little attention as a public health issue. The high prevalence of violence and its consistent association with a wide range of women's health problems suggest that violence seriously compromises women's health. Further studies should consider long-term exposure to any form of violence, including stalking, for a better understanding of the long-term impact of violence.
Footnotes
Acknowledgments
This study was partially supported with funding from the Health Institute Carlos III (1/06-36) (Ministry of Health, Spain) and Ciberesp (AE08018).We thank all the physicians who voluntarily participated in this study and all the women who gave their time to complete the questionnaire and shared with us their violence experiences. Without them, this study would not have been possible. We gratefully acknowledge the research assistants who collaborated in this study, who all worked with immense dedication and commitment to ensure the successful completion of the study.
Disclosure Statement
No competing financial interests exist.