Trauma Exposure and Intimate Partner Violence Among Young Pregnant Women in Liberia

Abstract

Intimate partner violence (IPV) is a global threat to women’s health and may be elevated among those exposed to traumatic events in post-conflict settings, such as Liberia. The purpose of this study was to examine potential mediators between lifetime exposure to traumatic events (i.e., war-related trauma, community violence) with recent experiences of IPV among 183 young, pregnant women in Monrovia, Liberia. Hypothesized mediators included mental health (depression, posttraumatic stress symptoms), insecure attachment style (anxious and avoidant attachment), and attitudes indicative of norms of violence (attitudes justifying wife beating). We tested a parallel multiple mediation model using the PROCESS method with bias-corrected and accelerated bootstrapping to test confidence intervals (CI). Results show that 45% of the sample had experienced any physical, sexual, or emotional IPV in their lifetime, and 32% in the 2 months prior to the interview. Exposure to traumatic events was positively associated with recent IPV severity (β = .40, p < .01). Taken together, depression, anxious attachment style, and justification of wife beating significantly mediated the relationship between exposure to traumatic events and experience of IPV (β = .15, 95% CI = [0.03, 0.31]). Only anxious attachment style (β = .07, 95% CI = [0.03, 0.16]) and justification of wife beating (β = .05, 95% CI = [0.01, 0.16]) were identified as individual mediators. This study reinforces pregnancy as an important window for both violence and mental health screening and intervention for young Liberian women. Furthermore, it adds to our theoretical understanding of mechanisms in which long-term exposure to traumatic events may lead to elevated rates of IPV in Liberia, and points to the need for trauma-informed counseling and multilevel gender transformative public health approaches to address violence against women.

Introduction

Intimate partner violence (IPV) is a global public health problem with severe consequences for physical, sexual, mental, and social wellness (World Health Organization [WHO], 2017). A total of 30% of ever-partnered women experience IPV in their lifetime (WHO, 2017), making IPV a significant contributor to morbidity and mortality for women globally. Moreover, it is well documented that IPV is associated with a myriad of adverse mental health outcomes, including depression, posttraumatic stress disorder (PTSD), substance abuse, and suicide attempts (Devries et al., 2014; Devries et al., 2013; Kessler et al., 2017). IPV has also been linked to poor reproductive health outcomes, such as unintended pregnancy, infant morbidity and mortality, sexually transmitted infections, and HIV (Li et al., 2014; Pallitto et al., 2013; Sarkar, 2008). Although IPV is one of the most common forms of violence against women occurring across settings and cultures (WHO, 2017), it is especially elevated in low-resource countries that are recovering from conflict-related violence, trauma, and disease outbreaks, such as Liberia (Kelly, Colantuoni, Robinson, & Decker, 2018; Rubenstein, Lu, MacFarlane, & Stark, 2017; Stark & Ager, 2011).

Understanding Liberia’s history of civil war, political instability, and disease outbreak is central to understanding the country’s high prevalence of IPV. Between 1989 and 2003, civil war displaced a third of the population and resulted in 250,000 deaths (Ellis, 2006). The civil conflict was characterized by widespread violence and infractions against humanity (National Transitional Government of Liberia, 2004), including killing, torture, forced labor, rape, sexual assault, and forced marriages (Annan, Blattman, Mazurana, & Carlson, 2008; Liebling-Kalifani et al., 2011). Violence against women reached unprecedented levels during the civil war. Reports estimate that up to 70% of women in Liberia have experienced some form of gender-based violence in their lifetime, with intimate partners reported as the perpetrators of more than 95% of reported gender-based violence cases (Liberia Institute of Statistics and Geo-Information Services, Ministry of Health and Social Welfare, National AIDS Control Program, & Macro International Inc, 2008; Stark & Ager, 2011). Similar to other conflict-affected settings such as Uganda, Thailand, Palestine, Sierra Leone, and Côte d’Ivoire (Annan & Brier, 2010; Clark et al., 2010; Falb, McCormick, Hemenway, Anfinson, & Silverman, 2013; Gupta et al., 2014; Gupta, Reed, Kelly, Stein, & Williams, 2012; Saile, Neuner, Ertl, & Catani, 2013), Liberian women and men who have experienced higher levels of conflict-related violence/trauma also report higher levels of IPV victimization and perpetration, respectively, both during and after conflict (Kelly et al., 2018; Vinck & Pham, 2013). These associations have been observed even 10 years post-conflict in Liberia (Vinck & Pham, 2013).

In addition to exposure to conflict-related trauma, Liberians have also endured trauma related to the more recent 2014–2016 Ebola epidemic. The epidemic devastated Liberia, resulting in 10,678 suspected and laboratory-confirmed cases, approximately 4,810 deaths, and widespread fear and stigma (Centers for Disease Control and Prevention, 2017; Van Bortel et al., 2016). Fear of contagion, the loss of loved ones, separation from family, and exposure to mass death was experienced by those in areas affected by the epidemic (Rabelo et al., 2016; Schwerdtle, De Clerck, & Plummer, 2017). The epidemic also resulted in a rise in community violence (Cohn & Kutalek, 2016; Cousins, 2018). Although not systematically monitored, multinational organizations reported an increase in gender-based violence in Liberia and other Ebola-affected countries during and after the epidemic (Korkoyah & Wreh, 2015). Thus, exposure to trauma associated with the Ebola epidemic could have similar population-level effects as conflict-related trauma on IPV incidence.

Collectively, the trauma associated with the occurrence of multiple devastating events may have lasting effects on the incidence of IPV in Liberia. However, explorations of pathways between exposure to traumatic events with IPV in post-conflict settings are limited. Among the complex multilevel pathways theorized (Devakumar, Birch, Osrin, Sondorp, & Wells, 2014; Kinyanda et al., 2016; Levy, 2002; Mootz, Stabb, & Mollen, 2017), mental health outcomes are one mechanism for further investigation. In Liberia, individuals exposed to war-related traumatic events report high rates of depression and PTSD, which in some cases have persisted long after the end of conflict (Galea et al., 2010; Johnson et al., 2008; Rockers, Kruk, Saydee, Varpilah, & Galea, 2010; Vinck & Pham, 2013). High rates of depression, PTSD, and psychological distress have also been observed among Ebola survivors (Rabelo et al., 2016; Schwerdtle et al., 2017). These mental health outcomes, in turn, have been found to be risk factors for IPV victimization and perpetration in non-conflict settings (Reingle, Jennings, Connell, Businelle, & Chartier, 2014; Stith, Smith, Penn, Ward, & Tritt, 2004; Yakubovich et al., 2018). While there are less studies focused in conflict-affected settings, a few studies associate negative mental health outcomes and risk of IPV in post-conflict settings (J. Gupta et al., 2014; Kinyanda et al., 2016; Rees et al., 2016; Usta, Farver, & Zein, 2008). In particular, one study found that childhood trauma, and PTSD, depression, and disability together mediated conflict-related trauma exposure and IPV among Afghan women (Jewkes, Corboz, & Gibbs, 2018).

Relationship and family dynamics are changed by exposure to community violence and associated traumatic events. Intergenerational effects include changes to family structure and stability (Levey et al., 2017), increased perpetration of child abuse and neglect (Catani, 2010; Crombach & Bambonyé, 2015), and diminished parental capacity and mental health (Berckmoes, De Jong, & Reis, 2017; Betancourt, McBain, Newnham, & Brennan, 2015; Panter-Brick, Grimon, & Eggerman, 2014). These factors and other forms of interpersonal trauma are considered risk factors for the development of insecure attachment, including anxious and avoidant attachment in adulthood in developed settings, refugee populations, and war-affected populations (Dalgaard, Todd, Daniel, & Montgomery, 2016; Fraley, Roisman, Booth-LaForce, Owen, & Holland, 2013; Morina, Schnyder, Schick, Nickerson, & Bryant, 2016; Widom, Czaja, Kozakowski, & Chauhan, 2018). Anxious attachment is characterized by fear of rejection and abandonment in intimate relationships, while avoidant attachment is characterized by discomfort with intimacy and the desire for independence. Although evidence is mixed (Velotti, Beomonte Zobel, Rogier, & Tambelli, 2018), both attachment styles have been linked to an increased likelihood of IPV victimization, revictimization, and perpetration in developed settings (Doumas, Pearson, Elgin, & McKinley, 2008; Lewis et al., 2017; McClellan & Killeen, 2000; Sandberg, Valdez, Engle, & Menghrajani, 2019). Thus, for young women in Liberia who grew up during and post-civil war, attachment style could be an additional mechanism explaining the relationship between lifetime exposure to traumatic events and recent experience of IPV.

In addition, exposure to gender-based violence and other forms of mass violence have lasting effects on gender inequitable norms and the acceptance of violence against women. Liberia ranked 154 out of 160 countries on the United Nations 2017 Gender Inequity Index (United Nations Development Program, 2017), a measure associated with both acceptance and occurrence of IPV within and between country analyses (Redding, Ruiz-Cantero, Fernández-Sáez, & Guijarro-Garvi, 2017; Tran, Nguyen, & Fisher, 2016; Willie & Kershaw, 2019). In a study with 592 young women in post-conflict Liberia, attitudes accepting physical IPV were prevalent and associated with a greater likelihood of experiencing IPV (Callands, Sipsma, Betancourt, & Hansen, 2013). Thus, attitudes justifying violence against women may also mediate the relationship between exposure to traumatic war/community-violence-related events and IPV experience among women.

Although there is increased recognition and concern about elevated incidence of IPV in post-conflict settings, more research is needed to understand how exposure to traumatic events in this context increases women’s risk of IPV. We examined potential mediators between lifetime exposure to traumatic events (i.e., war-related trauma, community violence) with recent experiences of IPV among young, pregnant women in Monrovia, Liberia. We focused on this population because young women between the ages of 15 and 24 are more likely to experience IPV than any other age group in Liberia (Liberia Institute of Statistics and Geo-Information Services et al., 2008), and pregnancy represents a time of increased risk for both mental health outcomes and IPV in vulnerable populations (Osok, Kigamwa, Huang, Grote, & Kumar, 2018; Rees et al., 2016; Taillieu & Brownridge, 2010). Moreover, experiencing IPV during pregnancy is of significant public health concern as it increases the risk for infant mortality and morbidity and negative mental health outcomes in mothers (Alhusen, Ray, Sharps, & Bullock, 2015; Mahenge, Likindikoki, Stockl, & Mbwambo, 2013; Rurangirwa, Mogren, Ntaganira, Govender, & Krantz, 2018). Based on the literature cited above, we hypothesize mental health outcomes (depression, PTSD), insecure attachment style (anxious and avoidant attachment), and attitudes indicative of norms of violence (attitudes justifying wife beating) will be independently associated with exposure to traumatic events (related to war, community violence) and IPV and may mediate the trauma-IPV relationship. These hypotheses are displayed in the conceptual model depicted in Figure 1.

Figure 1.

A conceptual model of the hypothesized parallel multiple mediation model of the relationship between experience of war-related trauma and intimate partner violence.

Method

This cross-sectional study included a researcher-administered survey with 183 pregnant women recruited from a community health clinic in Monrovia, the capital city of Liberia. The data used for this analysis were part of a larger study focused on sexual and mental health of young pregnant women. Clinicians referred pregnant women receiving prenatal care from this clinic to a research assistant to learn about the study. The research assistant then conducted a ten minute assessment of eligibility administered through computer-assistant personal interviewing (CAPI) software due to low literacy in the population. Eligibility criteria included (a) receiving or had received prenatal services from the local community health clinic, (b) 18 to 30 years old, (c) residing in Monrovia, (d) between 13 and 24 weeks of gestational age, (e) and no pregnancy-related medical problems. The research assistant obtained written informed consent from women who were eligible and interested in participating, and then administered a questionnaire that lasted 90 minutes in a private setting in the community health clinic using CAPI software. In total, 195 women completed the questionnaire. However, 12 women did not answer any of the IPV items; therefore, only 183 women were included in this analysis. Participants received $6 and a meal to compensate them for their time and travel. All study procedures were approved by Institutional Review Boards in the United States and Liberia.

Measures

The study measures were culturally adapted through a four-stage process (Beaton, Bombardier, Guillemin, & Ferraz, 2000): (a) we began with the initial translation of the measures from English to Liberian English (the primary language/dialect of our sample) by local community members; (b) next, we worked with key informants to ensure key concepts in the materials were relevant to the Liberian context and cultural norms; (c) measures were then back-translated by key informants to ensure that the meaning and context of material was maintained through translation; and, finally, (d) an expert panel of Liberian key informants, fluent in English and Liberian English and experienced in behavioral research, reviewed and came to consensus on semantic, idiomatic, experiential, and conceptual equivalence.

Of the sociodemographics measured, we included the following as potential covariates in our analysis: age (continuous), education (dichotomized for analysis as primary grade or less, and secondary or greater), in a relationship (yes/no), any living children (yes/no), employed (yes/no).

Our primary independent variable, exposure to traumatic events, was measured with the Harvard Trauma Questionnaire (HTQ; Mollica et al., 1992). This scale was designed to assess torture, trauma, or trauma-related events as they relate to mass violence or conflict. The scale was adapted for the Liberian context in partnership with the aforementioned expert panel of Liberian key informants. Respondents indicated whether they had experienced particular traumatic events (yes/no) assessed through 39 items (e.g., have you seen people fighting war, been taken from your family by force, seen the killings of people), which were summed for a total score (Cronbach’s α in present sample = .90).

Of the hypothesized mediators for this analysis, we measured two mental health factors (depression, PTSD), two relationship factors (anxious and avoidant attachment style), and one measure of norms of violence (attitudes justifying wife beating). Depressive symptoms were measured using the Patient Health Questionnaire–9 (PHQ-9; Kroenke, Spitzer, & Williams, 2001), a nine-item scale designed for clinicians to diagnose depression and monitor treatment response. The items assess the frequency and severity in which participants experience different depressive symptoms over the past 2 weeks. We adapted the scale responses from a 4- to a 3-point scale for our study population: severity: 0 = not at all difficult, 1 = sometimes difficult, 2 = too difficult; frequency: 0 = not at all, 1 = some days, 2 = every day (Cronbach’s α in present sample = .70). PTSD symptoms were measured with the 17-item PTSD Checklist (PCL-C), for which participants rated the severity of PTSD symptoms over the prior 30 days (Ruggiero, Ben, Scotti, & Rabalais, 2003; Weathers, Litz, Huska, & Keane, 1994). We amended the response scale from 5 to 4 points: 0 = not at all, 1 = sometimes, 2 = most of the time, 3 = all of the time (Cronbach’s α in present sample = .76).

Two subscales of the 34-item Experiences in Close Relationships–Revised (ECR-R) Questionnaire were used to measure anxious attachment style and avoidant attachment style (Fraley, 2002; Fraley & Shaver, 2000; Fraley, Waller, & Brennan, 2000). Anxious attachment items measure the participants’ fear of rejection and abandonment in intimate relationships (e.g., I worry that my boyfriend doesn’t love me/won’t stay with me). High scores on avoidant attachment items represent individuals who are uncomfortable with intimacy and seek independence (e.g., I prefer not to be too close to boyfriends/I find it hard to allow myself to depend on boyfriends). The original scale used a 7-point Likert-type scale, which was adapted to a 4-point Likert-type scale for our study population: 0 = never, 1 = sometimes, 2 = most of the time, 3 = always (Cronbach’s α in present sample for anxious attachment = .91 and avoidant attachment = .93).

A scale developed by Briere (1997) and used in the Liberian Demographic and Health Survey (Liberia Institute of Statistics and Geo-Information Services et al., 2008) was used to measure attitudes justifying wife beating in five different situations: (a) going out without telling husband, (b) not taking good care of children, (c) arguing with husband, (d) denying husband in bed, and (e) not cooking well. Responses were on a 4-point scale: 0 = never, 1 = sometimes, 2 = most of the time, 3 = always. The total scale was summed for analysis (Cronbach’s α in present sample = .74).

The study outcome, IPV, was measured with the Severity of Violence Against Women Scale (SVAW; Marshall, 1992). A total of 43 items of the original 49-item scale were included to evaluate the seriousness of violence experienced by women by an intimate partner in the prior 12 months, which included a range of emotional (e.g., threatened physical violence, spoiling something you own), physical (e.g., being slapped, hit, punched, pushed, or experienced other use of bodily force), and sexual abuse experiences (e.g., forced sex). Women were asked how often in the prior year they experienced these acts. Response options were on a 4-point scale ranging from 0 = never to 3 = 3 or more times. The total scale was summed for analysis (Cronbach’s α in present sample = .91). In addition to the SVAW, used as our outcome measure, respondents were also asked in single items about their experience of physical, sexual, or emotional IPV in their lifetime, as well as in the prior 2 months (while pregnant). We present these results within participant characteristics to add more description to women’s experience with IPV.

Data Analysis Approach

In SPSS version 24, we first tested bivariate correlations between all variables to determine if all hypothesized mediators should remain in the final model (i.e., were independently associated with both trauma and IPV) and to identify covariates to control for in the final model. We tested age, education, relationship status, any living children, and employment as potential covariates, as we expected them to associate with our model variables. Those that remained statistically significant (p < .05) in their relationship to exposure to traumatic events or IPV would be retained in the final model. We then tested a parallel multiple mediation model using the PROCESS method for SPSS version 24 developed by Hayes (2013) to examine mediators of the relationship between exposure to traumatic events and IPV. We chose to use a parallel multiple mediation model over simple mediation approach, as it allows for comparison of the sizes of the indirect effects through different mediators. Our hypothesized conceptual model (see Figure 1) included the independent variable (exposure to traumatic events), the dependent variable (IPV), and five proposed mediators (depression, PTSD, anxious attachment, avoidant attachment, wife-beating attitudes).

Our mediation test required the assessment of multiple pathways. The a path included the assessment of the effect of the independent variable (traumatic events) on the mediators. The b path consisted of analyzing the effect of each mediator variable on the dependent variable (IPV), controlling for the independent variable and all mediator variables. For the c′ path, we regressed traumatic events and all mediators onto IPV to assess the indirect effect of exposure to traumatic events on IPV through our mediators. We used the bias-corrected and accelerated bootstrapping method to test confidence intervals (CIs), which does not assume normality of the sampling distribution of the indirect effect and offers relatively greater power and better Type I error rates compared with other mediation approaches (Hayes, 2013). Following Preacher and Hayes (2008), 95% CIs were used for the indirect effects, and 10,000 bootstrapping samples were generated.

Results

Table 1 displays sociodemographic characteristics of the sample. Descriptive data and bivariate correlations of the variables included in the multiple mediation model are displayed in Table 2. PTSD was not independently associated with IPV, and anxious attachment was not associated with trauma or IPV. Therefore, we excluded both variables from our final model. The final tested model is displayed in Figure 2, which controlled for age; all other hypothesized covariates (education, relationship status, any living children, employment) were trimmed from the model (p < .10).

Table 1.

Participant Characteristics, Liberia 2016 (N = 183).

	n (%)	Range
Age (M, SD)	22.78 (3.68)	18–30
Education
No education	15 (8.20%)
Primary (Grade 1–6)	52 (28.42%)
Secondary (Grade 7–9)	46 (25.14%
High school (Grade 10–12)	63 (34.42%)
Some college or vocational training	7 (3.82%)
Relationship status
Not in a relationship	15 (8.20%)
In a relationship	168 (91.80%)
Any living children
No	99 (50.8%)
Yes	96 (49.20%)
Employed
No	91 (49.70%)
Yes	92 (50.30%)
Any lifetime experience of IPV
No	100 (54.60%)
Yes	83 (45.40%)
Any recent experience of IPV (prior 2 months)
No	124 (67.80%)
Yes	59 (32.30%)

Note. IPV = intimate partner violence, inclusive of emotional, physical, and sexual intimate partner violence.

Table 2.

Descriptive Information and Bivariate Correlations Between the Assessed Variables, Liberia 2016 (N = 183).

	Severity of IPV	T rauma Exposure	Depression	PTSD	Anxious Attachment	Avoidant Attachment	Justification of Wife Beating	M	SD	Range
Severity of 1PV	—	—	—	—	—	—	—	3.68	5.65	0–19
Trauma exposure	.40**	—	—	—	—	—	—	6.60	3.69	1–21
Depression	.23**	.39**	—	—	—	—	—	4.22	2.41	0–10
PTSD	.12	49**	.46**	—	—	—	—	5.40	4.14	0–22
Anxious attachment	.31 **	22**	.17*	.19*	—	—	—	9.94	6.88	2–35
Avoidant attachment	.06	.1 1	.08	21 **	.34**	—	—	24.22	8.16	0–47
Justification of wife beating	.23**	.18*	.1 1	.18*	.08	−.04	—	0.63	1.16	0–6

Note. IPV = intimate partner violence; PTSD = posttraumatic stress disorder.

^*p < .05.

^**p < .01.

Figure 2.

The tested parallel multiple mediation model of the relationship between exposure to traumatic events and experience of intimate partner violence.

Multiple Mediator Model of the Relationship Between Exposure to Traumatic Events and IPV

Figure 3 illustrates the findings of our multiple mediator model’s direct effects (indirect effects are reported in Table 3). The three proposed mediators were first regressed onto traumatic events (a path). We found greater exposure to traumatic events was associated with greater depression (β = .28; 95% CI = [0.18, 0.37]; p < .001), greater anxious attachment style (β = .40; 95% CI = [0.12, 0.67]; p = .005), and greater justification of wife beating (β = .07; 95% CI = [0.03, 0.15]; p = .002).

Figure 3.

Multiple mediation model of relationships between experience of wartime trauma, multiple mediators, and experience of intimate partner violence.

Note. Age is controlled for in this model.

*p < .05. **p < .01.

Table 3.

Mediation of the Effect of Exposure to Traumatic Events on Experience of Intimate Partner Violence Through Depression, Anxious Attachment Style, and Wife-Beating Attitudes, Liberia 2016 (N = 183).

			Bootstrapping
	Product of Coefficients		Bias-Corrected and Accelerated CI
	Point Estimate	SE	95% Lower	95% Upper
Depression	0.029	0.047	−0.067	0.117
Anxious attachment	0.073	0.033	0.026	0.161
Justification of wife beating	0.050	0.038	0.001	0.159
Total	0.152	0.070	0.031	0.3141

Note. Age is controlled for in this model. SE = standard error; CI = confidence interval.

In our assessment of all hypothesized mediators with IPV, which controlled for all mediators and exposure to traumatic events, anxious attachment style was associated with greater severity of IPV (β = .18; 95% CI = [0.07, 0.29]; p = .001), as were greater scores on the wife-beating attitudes scale (β = .67; 95% CI = [0.15, 1.33]; p = .04). Depression was not associated with IPV at a statistically significant level (β = .11; 95% CI = [–0.23, 0.44]; p = .52).

Finally, for the c′ path, we regressed traumatic events and all mediators onto IPV to assess the indirect effect of exposure to traumatic events on IPV through our mediators. The results show a positive and significant association (β = .15, 95% CI = [0.26, 0.73], p < .001).

Table 3 includes the results of the multiple mediation analysis. Statistical significance of mediators is indicated by 95% CIs that do not contain zero. The results show that taken together, depression, anxious attachment style, and justification of wife beating significantly mediated the relationship between exposure to traumatic events and experience of IPV (see total in Table 3). However, as individual mediators, only anxious attachment style and justification of wife beating were found to mediate this relationship.

Discussion

Liberia’s history of multiple devastating events over the prior three decades has left several generations of the population exposed to significant trauma. All women in our sample of pregnant young women in Monrovia, Liberia experienced one or more traumatic events, primarily related to war and community violence in their lifetime, even though most were born toward the end or after Liberia’s civil war. The results of the study show that greater exposure to traumatic events is associated with greater likelihood and severity of IPV victimization among young pregnant women. These findings support a growing body of work demonstrating elevated incidence of IPV in post-conflict settings and among individuals who have experienced trauma related to war and community violence (Annan & Brier, 2010; Clark et al., 2010; Falb et al., 2013; J. Gupta et al., 2014; J. Gupta et al., 2012; Jewkes et al., 2018; Kelly et al., 2018; Saile et al., 2013; Vinck & Pham, 2013). Importantly, this study adds to our understanding of pathways that potentially explain the trauma-IPV relationship using mediation analysis. We found greater exposure to traumatic events was independently associated with greater rates of depression, PTSD, anxious attachment, and attitudes justifying wife beating, all of which, apart from PTSD, were independently associated with experience of IPV among young pregnant women. However, only anxious attachment and attitudes justifying wife beating were identified as statistically significant individual mediators of the trauma-IPV relationship.

This study supports our hypothesis that greater exposure to traumatic events in a post-conflict setting may influence the development of insecure attachment, which in turn increases women’s risk of IPV victimization. Experiencing war and the long-term consequences of such events on communities (e.g., poverty, family separation, continued community violence) can influence a child’s sense of attachment through a loss of security, predictability, family unit stability and structure, and supportive community networks and institutions (Betancourt, McBain, Newnham, & Brennan, 2013). For example, Liberia’s civil war resulted in an estimated 340,000 orphaned children (Collins, 2015; Liberia Institute of Statistics and Geo-Information Services, 2008). Research has demonstrated childhood trauma, including separation from parents, is predictive of insecure attachment in adults (Bryant et al., 2017). Qualitative research in post-conflict Liberia reported adolescents who experienced disruptions in early relationships suffered from disconnection from their families and communities, in addition to other negative psychosocial and emotional health consequences (Levey et al., 2017).

This finding also contributes to our understanding of the relationship between attachment style and IPV. Insecure attachment styles have been linked to an increased likelihood of IPV victimization and perpetration in developed settings (Doumas et al., 2008; Lewis et al., 2017; McClellan & Killeen, 2000; Sandberg et al., 2019). However, a systematic review and meta-analysis examining this relationship found a considerable number of studies reporting no association, and very few studies with this aim conducted in low-income countries (Velotti et al., 2018). In our study, anxious attachment was identified as a mediator between trauma exposure and IPV. However, avoidant attachment was not independently associated with IPV. Future research should continue to explore the intergenerational effects that war and other forms of community trauma can have on attachment and interpersonal relationships to better understand elevated IPV incidence in post-conflict settings.

We also found support for our hypothesis that young pregnant women who have been exposed to trauma would be more likely to endorse violence against women. As anticipated, this in part explained the relationship between trauma exposure and IPV, adding to other studies demonstrating acceptance of violence against women is associated with victimization (e.g., Allen & Devitt, 2012; Callands et al., 2013). This relationship is likely bidirectional, with IPV exposure influencing women’s learned acceptance of this behavior. However, individual attitudes endorsing violence against women are also an indicator of broader community norms that endorse gender inequitable attitudes and a culture of violence more broadly. In conflict-affected settings where gender-based violence was used as a war tactic, community violence can have lasting effects on broader acceptance of violence against women. Thus, young pregnant women who have experienced greater traumatic events in our sample may be connected to networks with more acceptance of violence against women and gender inequity, and, in turn, more IPV perpetration. Kelly et al. (2018) demonstrated Liberian women who lived in a conflict fatality-affected district 5 years post-conflict were at a 50% increased risk of IPV exposure. Those in districts with 4 to 5 cumulative years of conflict were almost 90% more likely to experience IPV than those living in a district with no conflict (Kelly et al., 2018). Our study only assessed individual-level attitudes as a proxy for community norms. More research is needed that takes a multilevel approach to understanding the effects of broader community norms resulting from community violence on IPV incidence.

In addition to contributing to our theoretical understanding of mechanisms between trauma exposure and IPV, this study has important implications for intervention. Approximately 45% of young pregnant women in our sample reported experiencing any emotional, physical, or sexual IPV in their lifetime, and 32% had experienced at least one of these forms of abuse in the 2 months prior to our interview, while pregnant. While alarming, these rates are consistent with other estimates in Liberia (Liberia Institute of Statistics and Geo-Information Services et al., 2008; Stark & Ager, 2011), reinforcing the need for violence reduction interventions for young women in Liberia and highlighting pregnancy as a particularly important time frame for intervention. The WHO recommends clinical inquiry in antenatal care for women with conditions that could be caused or exacerbated by IPV (WHO, 2018). Universal screening for IPV paired with ongoing support services is recommended for women of reproductive age including pregnant/postpartum women in developed countries (Curry et al., 2018). However, even in resource-rich settings, implementation is low (Chisholm, Bullock, & Ferguson, 2017). Thus, research is needed in settings with low-capacity health systems such as Liberia to increase the feasibility and effectiveness of integrating IPV interventions into antenatal care, and to reach women not linked to services (WHO, 2011).

Our findings also point to the potential for trauma-informed counseling approaches for violence reduction in low-income post-conflict settings. Efficacy trials on trauma-informed IPV counseling show promise for IPV prevention and reduction but are largely conducted in developed settings (e.g., Creech, Benzer, Ebalu, Murphy, & Taft, 2018; Decker et al., 2017; Machtinger, Cuca, Khanna, Rose, & Kimberg, 2015; WHO, 2013), pointing to the need for research that adapts and tests trauma-informed IPV interventions for West African settings. Our study also demonstrates the need for IPV interventions that dually address mental health conditions. Although the mental health outcomes assessed in this study were not identified as mediators between trauma and IPV, both trauma and IPV were associated with depression, and trauma was associated with PTSD. This finding is consistent with syndemics theory (Singer, 2009; Singer & Clair, 2003) and research, which demonstrates a cumulative effect of early life trauma experience on multiple mental health conditions in adulthood that require integrated public health approaches (Singer, Bulled, Ostrach, & Mendenhall, 2017). We previously found support for a cumulative effect of exposure to trauma and other mental health and psychosocial conditions on engagement in transactional sex among this sample of young pregnant women in Liberia (see Sileo, Kershaw, & Callands, 2019).

Finally, our finding that attitudes toward wife beating contribute to women’s risk for IPV suggest the need for multilevel gender-transformative interventions, which broadly aim to reconfigure harmful masculine norms to improve men’s and women’s health and to shift beliefs, behaviors, and relationships toward gender equity (Dworkin, Fleming, & Colvin, 2015; Dworkin, Hatcher, Colvin, & Peacock, 2013; G. R. Gupta, 2000). Systematic reviews of gender-transformative interventions report that programs that engage individuals or groups of men and women to adopt gender-equitable attitudes have been successful at reducing IPV incidence and gender-inequitable attitudes in African settings (Casey, Carlson, Two Bulls, & Yager, 2018; Dworkin, Treves-Kagan, & Lippman, 2013). Individual-level interventions that combine community-or structural-level components may be even more effective (Dworkin et al., 2015; Dworkin, Treves-Kagan, & Lippman, 2013). Examples include campaigns to raise community awareness about violence against women (Hossain et al., 2014), and those that address young women’s economic dependence on men through conditional cash transfers (Kilburn et al., 2018). The scale up of these approaches in tandem with policy-level change (e.g., development and enforcement of laws to investigate and prosecute perpetrators of violence using due process) could together address norms that drive the IPV epidemic in Liberia and significantly reduce the high burden of morbidity, mortality, and economic costs associated with IPV (WHO, 2017). However, achieving this level of change will require concerted efforts and commitment across governmental and nongovernmental institutions to prioritize the end of IPV and gender-based violence more broadly.

Limitations

The generalizability of our findings is limited by our purposive sampling of women attending community health clinics in the capital city of Liberia. The relatively small sample size may have limited our ability to detect all existing associations. While our mediation analysis approach is an important contribution to the literature on pathways between trauma exposure and IPV, the cross-sectional and self-reported nature of the data prohibits conclusions about causality and the directional nature of the relationships examined. Only one other study that we are aware of has examined these relationships in a conflict-affected setting with path analysis, which was also cross-sectional (Jewkes et al., 2018). Future longitudinal research is needed to further elucidate these and other pathways, which can establish temporality and determine the direction of the correlations observed in our cross-sectional study, which are likely bidirectional. For example, depression may increase risk of experiencing IPV, but may also be a consequence of IPV. In addition, the measures used were adapted for this population to ensure cultural equivalence. However, adaptations may limit the ability to compare our findings with studies that have used the original scales. Despite these limitations, these data shed light on the intergenerational and long-term societal effects of trauma on IPV and contribute preliminary data to warrant more rigorous investigations on mental health, interpersonal, and normative pathways between exposure to trauma on experience of IPV among Liberian youth.

Conclusion

In this study of pregnant young women in Liberia, we found high rates of IPV, including experience of IPV during pregnancy. With dire consequences of IPV during pregnancy on both mother and infants, this study reinforces pregnancy as an important window for both violence and mental health screening and intervention for this population. In addition, this study adds to a growing literature that demonstrates cumulative exposure to traumatic events in post-conflict settings is associated with an increased risk of experiencing IPV and begins to fill a gap in the literature on potential pathways that explain this relationship. Anxious attachment style and attitudes justifying wife beating were identified as mediators between the trauma-IPV relationship. These findings not only contribute to our theoretical understanding of this relationship but point to the need for trauma-informed counseling and multilevel gender-transformative public health approaches to address violence against women.

Biography

Katelyn M. Sileo is an Assistant Professor in the Department of Kinesiology, Health, and Nutrition at the University of Texas at San Antonio. Her research focuses on the role of gender norms in health outcomes and the development of behavioral and health system interventions to improve HIV and reproductive health care engagement in global settings.

Trace S. Kershaw is Department Chair and professor in the Department of Social and Behavioral Sciences in the Yale School of Public Health. His research focuses on social and structural determinants of health among adolescents and emerging adults, with a focus on behavioral intervention development and the application of technologic methods to understand how social and geographic context influence health.

Shantesica Gilliam is a second-year doctoral student in the Department of Health Promotion and Behavior at the University of Georgia. Shantesica’s research focuses on the developing sexual health and HIV prevention strategies focused on reducing HIV stigma, racial and gender health disparities, and community violence and gender-based violence among racial and sexual minorities.

Erica Taylor is a second-year doctoral student in the Department of Health Promotion and Behavior at the University of Georgia. Her research focuses on the intersectionality of mental health and prenatal care in minority women living in low-resource areas, examining mental health implications of HIV diagnosis in pregnant women, and to explore stress reduction interventions guided by motivational interviewing in pregnant women.

Apoorva Kommajosula earned her BSc in Health Promotion at the University of Georgia and is an MPH candidate at Emory University. Her research interests in global health focus on analyzing barriers to health services in low- and middle-income countries and developing community-based interventions to promote health as well as health education in low-resource settings.

Tamora A. Callands is an assistant professor in the Department of Health Promotion and Behavior at the University of Georgia. Her research focuses on developing interventions to promote mental and sexual health among pregnant women in low- and middle-income countries (LMIC), exploring facilitators and barriers to violence prevention in LMICs, and building social support for pregnant women in rural communities via technology.