Relationship violence occurs at every socioeconomic level and across all age-groups, but women in their early 20s living in economically disadvantaged urban communities are especially vulnerable. Nonfatal intimate partner violence peaks during late adolescence and young adulthood,^1,2 and women living in poverty are more likely to be victimized by intimate partners than are more affluent women.^3,4 Intimate partner violence has frequently been examined in terms of its impact on reproductive health. This is understandable, given the potential seriousness of violence, especially when women are pregnant and their own health and the health of their fetus may be in jeopardy. In this article, we flip the equation to consider the relationship between women's risk behaviors and pregnancy during adolescence and their exposure to intimate partner violence during their early 20s.

A constellation of factors, including family histories of abuse and high levels of community violence, and such aspects of structural violence as discrimination and racism, are known to contribute to increased risks of intimate partner violence in settings where relationships and families are challenged by poverty.⁵ Low-income, predominantly minority, young women may also be at risk because of their reproductive histories, which differ in a number of ways from those of their more affluent peers.⁶ Socioeconomically disadvantaged women are more likely than others to initiate sex before entering high school, get pregnant unintentionally as teenagers, become mothers before age 21 and be single parents. If they become pregnant, they face hurdles to high school graduation as well as postsecondary education, which limit their employment prospects and potentially constrain their ability to leave abusive relationships.⁷ Some evidence suggests that young women in these circumstances may be more likely to bear a child than to get an abortion, even if they are not in a stable relationship or married.⁸ Further, despite the growing acceptance of single parenthood and declines in the overall U.S. abortion rate from 1994 to 2000, the rate of abortions rose sharply for women with incomes of less than twice the federal poverty line.⁹ A disproportionate number of abortions in the United States are obtained by never-married women, residents of metropolitan areas, and blacks and Hispanics—in sum, a large proportion of women of reproductive age residing in our nation's inner cities.¹⁰

Research has also drawn links between behaviors in adolescence and intimate partner violence during the teenage and young adult years.^11–13 This is especially a concern for urban young women, who report relatively high levels of peer-directed violence, especially during middle school, a time when such behaviors as fighting often peak among both boys and girls.¹⁴ Once initiated, these aggressive patterns of interacting and resolving interpersonal conflicts may be difficult to break.¹⁵ Indeed, it appears that those who engage in aggressive behavior during the middle school years are particularly vulnerable to dating violence during adolescence, as well as adult partnership violence, and that those who experience intimate partner violence once are at risk of repeat occurrences.^16–18 Whether such violence occurs, however, may be influenced by other life events and relationship stressors, including women's experiences with pregnancy.

Because far more women than men are killed or seriously harmed by a partner every year,^19–21 most research on the connections between reproductive health and intimate partner violence understandably focuses on women as victims. However, in contrast to other forms of interpersonal violence, intimate partner violence is as often perpetrated by women as by men, perhaps in part because women underestimate the potential harm of partners' retaliation: Statistics clearly indicate that men strike with more deadly force.²² Here, we focus on women as both victims and perpetrators, to obtain a better understanding of overall intimate partner violence involvement and its links to prior risk behaviors and pregnancy experiences.

Intimate partner violence has been correlated with negative maternal and infant health outcomes that may be a consequence of HIV and other STDs, as well as with stress and exacerbation of chronic health problems.²³ In a 2007 review, Coker identified 51 research reports, mostly cross-sectional, that address the link between intimate partner violence and sexual and reproductive health. Intimate partner violence was associated with sexual risk-taking, inconsistent condom use, partner nonmonogamy, having an unplanned pregnancy or induced abortion, having an STD and sexual dysfunction.²⁴ In a review of research on the consequences of unintended pregnancies—which account for an estimated half of U.S. pregnancies²⁵—Pallitto et al. also draw a connection to intimate partner violence, while noting that research on associations between the two is limited.²⁶ In a study that compared women visiting family planning and prenatal clinics, those seeking an abortion were more likely than those continuing their pregnancies to have been victims of intimate partner violence in the past year.²⁷ In a multistate survey of women who had recently given birth, those whose pregnancies had been unwanted or mistimed reported rates of physical violence that were about five times the rate among women who had intended their pregnancies; in analyses controlling for income and education levels, women who had had mistimed or unwanted pregnancies accounted for 70% of those who reported physical violence.²⁸

Women who experience physical violence and emotional abuse during pregnancy may have an elevated risk for miscarriage, although evidence is mixed.^29,30 Data from women giving birth in 26 states and participating in the 2000–2003 Pregnancy Risk Assessment Monitoring System indicate that those who experience intimate partner violence either prior to or during pregnancy are at risk for multiple poor maternal and infant health outcomes, including having low-birth-weight infants and infants who require neonatal intensive care. The associations were strongest among those experiencing violence during but not prior to pregnancy.³¹ As Blake and coauthors point out in a study of minority young women, the father's disinclination to have the child may be a factor.³² However, in an abusive relationship, a woman may have an unintended pregnancy if her partner's desire to have a child leads him to sabotage the couple's birth control or condom use, or to make explicit demands for unprotected sex.³³

This study builds upon the foundation of research that has drawn connections between intimate partner violence and reproductive health. We use data from the Reach for Health Longitudinal Study, which has followed a large sample of females from middle school into their mid-20s. We examine two questions: First, are women who reported sexual initiation and aggressive behaviors during middle school more likely than others to be involved in intimate partner violence victimization and perpetration during young adulthood? Second, are women's reproductive histories—and, especially, their reports of unintended pregnancies and problems with pregnancy—associated with current intimate partner violence involvement?

METHODS

The Reach for Health Study

Our analyses are based on data from 581 female participants in the Reach for Health Longitudinal Study, who had been followed for approximately 10 years, from middle school into young adulthood. Described in more detail elsewhere,³⁴ the larger study was initiated in Brooklyn, New York, as part of a multisite research agreement supported by the Office of Minority Health of the U.S. Department of Health and Human Services and the National Institute of Child Health and Human Development to explore strategies for promoting health and reducing risk in economically disadvantaged communities.

Youth attending seventh or eighth grade at three public middle schools were eligible for participation. Each school was located in an economically disadvantaged area with high rates of teenage pregnancy, STD infection, violence-related injuries and other sources of morbidity; about 80% of youth were eligible for free-lunch programs. At each survey administration, all students at each school were invited to participate. Written parental permission and youth assent were obtained. Parental consent was provided for 89% of eligible students; 95% of students with parental permission completed baseline surveys.

Students who completed an eighth-grade survey in the 1995–1996 or 1996–1997 academic year, and who remained in the city, were eligible for follow-up. Through high school, information was collected on risks associated with sexual activity, substance use and interpersonal youth violence. With subsequent support from the Centers for Disease Control and Prevention and new written consent from participants themselves, two young adult survey waves have been completed, the first when participants were about 19–20 years of age (in 2002–2003), and the second when respondents were 22–25 years old (in 2005–2007). All surveys have been administered as paper-and-pencil questionnaires. As participants have gotten older, surveys have been administered either in small groups or individually at locations where privacy could be assured, including participants' schools and the study's office.

Included in these analyses are young women who completed the eighth-grade survey and the second adult survey. Of the 768 middle school females who completed eighth-grade surveys and lived in New York City during the initial high school survey, 76% have been successfully followed into adulthood. Young women were followed whether or not they had remained in school; more than a third did not graduate from high school on time. Attrition analyses indicate few differences between those who completed a young adult survey and the original, larger pool of eighth graders, including about 90 girls without New York City addresses prior to the high school survey. Those who were followed up did not differ from those who were not by race or ethnicity, or by eighth-grade measures of aggression, sexual experience or substance use. However, those who were successfully followed were less likely than those who were not to be old for their grade level in middle school.

Measures

•Partner violence. Information on victimization and perpetration of partner violence was collected during the last survey wave. As defined here, intimate partner violence incorporates physical, sexual and emotional abuse. These types of intimate partner violence are typically interrelated; those who experience one type have a substantially elevated likelihood of experiencing another.³⁵

Assessment of partner violence referred to respondents' experiences in the last year. To measure whether a respondent had been a victim of physical abuse, the questionnaire asked if anyone she had dated or had had sex with in the past year had done the following: hit, punched or slapped her; thrown something at her or hit her with an object; pushed, grabbed or shoved her; pulled her hair; scratched or bitten her; kicked or choked her; threatened her with a knife; threatened her with a gun; used a knife against her; and used a gun against her. One item addressed sexual violence: "Has a partner forced or threatened to hurt you to have oral, anal, or vaginal sex?" Four items assessed emotional intimate partner violence. Items were based on the Conflict Tactics Scale (Version 1) and focused on content recommended by Straus for assessment of more serious violence.³⁶ One item asked the respondent if a partner had made her family or friends worry about her. Another asked if a partner had threatened to hurt her or someone she cared about (examples provided were a child, friend, family member and pet). The third asked if a partner had said he would hurt himself or the respondent if she tried to leave him or break up with him. The last asked if a partner had been jealous or possessive of the respondent, had checked up on her or had refused to let her go out with her friends. These items were reworded to assess perpetration of abuse; respondents were asked whether they had done any of these things to someone they had dated or had sex with.

Respondents indicated whether each victimization or perpetration act had happened once, more than once or not at all. Individual victimization items are positively correlated; the Cronbach's alpha exceeds 0.8. The same is the case for items measuring perpetration. For these analyses, participants who reported any occurrence of a given type of violence received a score of 1; all others received a score of 0.

•Risk behaviors. Information on early sexual initiation and middle school aggression was collected in the eighth-grade survey. One question was used to assess early sexual initiation: "Have you ever had sexual intercourse? This is sometimes called 'going all the way." This item was coded 0 for no and 1 for yes. Middle school aggressive behaviors were assessed by five items that asked about behavior in the past 90 days: Respondents were asked whether they had threatened to "beat [someone] up, not including your brothers and sisters or other children you live with"; had been "in a physical fight (a fight with hitting, kicking, or pushing)"; had carried "a knife or razor (including a box cutter)"; had carried a gun; and had threatened "to cut, stab, or shoot" someone. Items on guns were prefaced by the statement "We want to know about real guns—the ones that shoot bullets, not BB guns, water guns, or any toy guns." Items on knives were prefaced with "We want to know about using a knife or razor to hurt somebody, not about accidents where someone gets cut." Items were coded 0 for no and 1 for yes; responses were summed and then dichotomized because the distribution was skewed. At the last follow-up, respondents reported on their number of lifetime sex partners.

•Pregnancies. Respondents to the young adult survey were asked whether they had ever been pregnant, were currently pregnant, were currently trying to get pregnant or had ever given birth. Those who had been pregnant were asked their age at first pregnancy and whether each pregnancy had been intended (e.g., "The first time you got pregnant, were you trying to get pregnant?"). Responses were collapsed into any unintended pregnancy versus none. Women were asked if they had ever experienced a miscarriage, had an abortion (and, if so, if they had had more than one), or tried for six or more months to get pregnant but been unable to. For the multivariate analyses, we created a summative scale of unintended pregnancy and pregnancy problems (abortion, miscarriage, difficulty getting pregnant); this permitted us to assess levels of risk associated with increasing numbers of potentially adverse pregnancy-related experiences.

•Relationship characteristics. Three relationship characteristics were assessed: whether respondents were currently living with a partner, whether they were married and whether they were raising children (biological or other). These characteristics are not mutually exclusive. For example, 73% of women who were married lived with a partner, compared with 25% of unmarried women. The majority (75%) of married women were raising children, as were 58% of unmarried women; 70% of those living with a partner were raising children, compared with 56% of those who were not.

•Social and demographic characteristics. The following social and demographic measures were assessed: age at the time of the young adult survey, race and ethnicity (dichotomized as black vs. other), education (completed high school or more vs. non–high school graduate) and employment (employed vs. not).

Analysis

First, we examined bivariate associations between any intimate partner violence victimization or perpetration and risk behaviors, pregnancy-related experiences and current relationship characteristics. Second, we performed separate logistic regressions exploring associations between intimate partner violence involvement and risk behaviors, pregnancy-related measures and current relationship characteristics; these analyses controlled for age, race or ethnicity, education and employment status. We present adjusted and unadjusted odds ratios and 95% confidence intervals. Analyses were conducted using SPSS, version 13.

RESULTS

Descriptive Findings

Respondents' mean age at the time of the young adult survey was 23 years; they had been 13–14 years of age when they provided information during middle school. Approximately three-quarters of women described themselves as black; most others were Latina (Table 1). About a quarter (23%) had not graduated from high school; 33% were not currently employed, and 46% had full-time jobs. About six out of 10 women were raising children, although only 11% were married and 30% were living with a partner.

One in five women had initiated sexual intercourse by eighth grade, and three in five had reported at least one type of aggressive behavior during middle school. Six percent of respondents reported no sexual partners by the last survey, 22% reported one or two, and the rest said they had had three or more.

The majority of women—72%—reported that they had ever been pregnant; 42% of this group had had a first pregnancy before age 18, including 14% who reported a pregnancy at age 15 or younger. Eight percent were currently pregnant, and an additional 11% were trying to get pregnant; half had given birth.

Unintended pregnancies were common in this sample: Sixty-three percent of respondents reported at least one unintended pregnancy, most of which were first pregnancies. More than four in 10 women had had an abortion, and nearly a quarter had had multiple abortions. Ten percent of women reported a fertility problem (that is, they had tried to get pregnant but not been able to), and 17% reported a miscarriage.

Twenty-nine percent of women said that they had been the victim of intimate partner violence at least once in the last 12 months; 21% had perpetrated violence during that period. Victimization and perpetration are clearly connected: Sixty-six percent reported neither victimization nor perpetration, 5% reported only perpetration, 13% reported only victimization, and 15% reported both victimization and perpetration (not shown).

Predictors of Partner Violence

In uncontrolled analyses (Table 2), middle school aggression is associated with about a doubling of the odds of violence victimization (odds ratio, 2.0) and perpetration (2.5). Lifetime number of sex partners is also positively associated with risk (1.3 for both victimization and perpetration). However, early sexual initiation is not.

Victimization is also associated with several pregnancy-related measures. Women who were currently trying to get pregnant were at increased risk (odds ratio, 1.8), as were those who had had at least one unintended pregnancy (1.6) or one abortion (1.5), and those who had had a problem getting pregnant (2.2). By contrast, perpetration is associated only with having had an abortion (1.8) or multiple abortions (1.7). Both victimization and perpetration are positively associated with women's number of unintended pregnancies and pregnancy problems (1.3 for each violence measure). Raising a child is not related to the likelihood of intimate partner violence, although living with a partner is associated with an increased risk of perpetration (1.9).

In the adjusted analyses, middle school aggression and lifetime number of sex partners remain risks for intimate partner violence victimization (odds ratios, 1.9 and 1.3, respectively), along with number of unintended pregnancies and pregnancy problems (1.3). For perpetration, middle school aggression remains significant (odds ratio, 2.5), as do lifetime number of sex partners (1.3), number of unintended pregnancies and pregnancy problems (1.3), and living with a partner (1.8). In addition, early sexual initiation is negatively associated with perpetration (0.5). No other measures predict the risk of involvement in intimate partner violence.

DISCUSSION

Data on the connections between violence and reproductive health issues have been limited, despite the importance of such information for informing effective interventions and identifying opportunities for reaching women at risk.³⁷ Addressing this limitation, this study draws attention both to a population of young women at risk and to a factor—a broad scope of pregnancy problems—that can help target prevention as well as intervention programs.

About one in three young women in the United States become pregnant before age 20, about half of all U.S. pregnancies are unintended and about half of unplanned pregnancies are ended by abortion.³⁸ Further, nearly a third of women of childbearing age have been a victim of intimate partner violence at some point in their lives.³⁹ Our findings suggest that for at least some young women, unintended pregnancies, pregnancy problems and partnership violence are common and entwined, and that greater integration of reproductive health and community violence prevention efforts are called for. In our study, unintended pregnancies and fertility problems were associated with an increased risk of intimate partner violence. While current pregnancy was not identified as a predictor of increased risk, neither was it associated with a reduced risk; this is a concern, given the potential for physical harm during this period.

For at least some young women, unintended pregnancies, pregnancy problems and partnership violence are common and entwined.

In addition to examining characteristics of the context in which intimate partner violence occurs, our results confirm the established correlation between victimization and perpetration. As previous work confirms,⁴⁰ disputes between domestic partners often involve reciprocal aggressive acts that tend to place women at greater risk of serious injury. For the most part, characteristics that predict the risk of victimization also predict the risk of perpetration, and these include not only a history of risk behaviors (aggression in middle school, greater number of lifetime sex partners), but also pregnancy problems. Interestingly, however, early initiation, by itself, is not independently related to victimization. Perpetration is related to living with a partner, which may increase opportunities for violence and strain relationships.

Limitations

Several limitations of our study must be considered. As in any longitudinal study, there may be selective bias in attrition. However, three-quarters of the sample were successfully tracked—a notable proportion, given evidence that the problem of sampling attrition is increasing.⁴¹ Further, attrition analyses indicate that middle school students lost to subsequent surveys and those successfully followed up did not differ in levels of baseline risk behaviors.

The sample was drawn from three schools in one city; so caution should be used in generalizing results, especially to more socioeconomically or ethnically diverse populations. And while the original sample represented about three-quarters of all females at the participating schools, those who did not have parental permission to participate may differ from those who did—for example, in terms of truancy. Evidence from substance use surveys indicates that students who do not get parental permission may engage in higher levels of risk behavior, but do not present a major source of bias.⁴²

In addition, reports of pregnancy experiences were obtained at the last survey wave and are retrospective. Women might have answered differently about some issues—such as whether pregnancies were unintended—had they been surveyed at the time of the actual events.⁴³ Data on unintended pregnancies and pregnancy problems are lifetime measures, and many of these events likely occurred prior to the reported intimate partner violence. However, we cannot examine timing in relation to intimate partner violence; disentangling the complex causal and temporal relationships among these factors is an important area for future research. We examined unintended pregnancies and pregnancy problems together because, from a conceptual standpoint, we are interested in the consequences of increasing numbers of potentially adverse events. In future studies with sufficiently large samples, it will be important to examine different types of events, and combinations of events, to better understand relationships with intimate partner violence.

Information about intimate partner violence was collected for the past year, and was not considered in the context of a specific relationship or only the current relationship. Further, accounts of intimate partner violence include lesser acts of aggression as well as less frequent, more serious acts. While different types of intimate partner violence are correlated, additional factors may differentiate women at greatest risk of serious harm, either in relation to unintended pregnancies and pregnancy problems or not.

Conclusion

Taken together, our findings suggest the importance of incorporating education related to intimate partner violence into reproductive health services, including screening and treatment for pregnancy, prenatal care and abortion services. Such education would include information that helps women assess whether they are at risk and, if they are, helps them understand the importance of changing partners or their own behavior to avoid persistent or escalating violence as they move through adulthood. The connection of number of lifetime sexual partners to intimate partner violence risk suggests that agencies whose clients are likely to have multiple sexual partners, such as STD clinics, also should take a more active part, moving beyond providing general social marketing information to more targeted screening, referral and linkages to appropriate services, such as counseling and legal aid.

The findings that both perpetration and victimization are associated with the number of lifetime sexual partners may suggest that young adult women who have had many partners may not yet have developed the relationship skills or had the experience of a stable relationship in which interactions are nonviolent. However, even as these young women form longer term relationships in middle adulthood, they may need assistance in developing the nonviolent conflict resolution skills required to avoid intimate partner violence. This possibility, together with the connection between middle school aggression and victimization and perpetration more than a decade later, underscores the importance of early prevention and intervention. In addition to overall community efforts to address violence and all of its negative consequences for women and their families, there is the need for effective prevention curricula that develop skills for behavior management both with peers and with partners. Ensuring that such programs are successfully implemented, especially for women and communities at high risk, requires advocacy, public will and a sustained budgetary commitment to prevention.