A Review of Mediators in the Association between Child Sexual Abuse and Revictimization in Romantic Relationships

ABSTRACT

Child sexual abuse (CSA) is a widespread adverse experience that has multiple detrimental impacts in the long-term, affecting mental, sexual, and physical health of survivors. In addition, CSA may impede interpersonal functioning, and be associated with a heightened risk of revictimization. Recent review articles have summarized the possible risk factors associated with sexual revictimization, yet an increasing body of literature suggests that CSA may be a key risk factor for multiple forms (i.e., psychological, physical, sexual) of victimization occurring specifically in the context of intimate relationships, either in adolescence (dating violence) or adulthood (partner violence). Our understanding of the mechanisms linking CSA and dating violence or intimate partner violence is still limited. This systematic review of the literature is aimed at summarizing the identified mediators of the association between CSA and revictimization in romantic relationships in past empirical reports. A total of 18 studies meeting criteria were identified. Apart from PTSD, few potential mediators have been explored in more than one study. In addition, few studies have investigated protective factors that may reduce the risk of revictimization. Several limitations in the current literature were identified including issues related to definitions and measurement. Studies relying on longitudinal designs with representative samples are clearly needed to orient future prevention efforts and break the revictimization trajectory.

KEYWORDS:

• Child sexual abuse
• dating violence
• partner violence
• revictimization
• review

Child sexual abuse (CSA) is associated with a plethora of negative outcomes such as anxiety, depression, suicidal ideation, and physical health problems (Amado et al., 2015; Irish et al., 2010). Post-traumatic stress disorder (PTSD) involving flashbacks, avoidance, and hypervigilance has also been one of the frequent repercussions associated with CSA (Hillberg et al., 2011). CSA has also been linked to a higher risk of sexual revictimization later in life (Walker et al., 2019). Indeed, a recent meta-analytic review summarizing prevalence rates across 80 studies revealed that close to half (47.9%) of victims of CSA experienced later sexual victimization in their life (Walker et al., 2019).

Researchers have also explored other types of revictimization among CSA survivors beyond the increased specific risk of sexual assault/abuse. Empirical studies have reported that CSA is associated with different forms of victimization, both in the context of dating relationships in adolescence (Hébert, Moreau et al., 2017), and in intimate romantic relationships in adulthood (Capaldi et al., 2012). Dating violence (DV) qualifies the experiences of violence in adolescence while intimate partner violence (IPV) refers to experiences in adulthood. DV and IPV refer to two different phenomena, since IPV takes place in longer relationships, and tends to be part of a crystalized pattern instead of being ineffective attempts in the learning context of conflict resolution (Cui et al., 2013), and is a distal outcome of CSA in comparison with DV. Available data shown that even after accounting for other forms of child maltreatment, CSA was found to be associated with a heightened risk of psychological, physical, and sexual DV for both teenage girls and boys (Hébert, Moreau et al., 2017). As well, findings have shown CSA to be linked to different forms of IPV in adulthood. For instance, in a large survey involving 15,000 adults in Canada, women with a history of CSA were at two to five-fold risk of experiencing psychological, physical, or sexual victimization by an intimate partner (Daigneault et al., 2009).

The connection between CSA and DV/IPV is well-established, but there has been a dearth of theoretical models tested that can provide insights into the mechanisms behind the revictimization process. Yet, knowledge regarding the mechanisms linking CSA to DV/IPV is essential to inform evidence-based prevention practices. While current efforts aim to provide universal prevention programs to prevent violence in romantic relationships, identification of mediators may provide essential cues to orient more targeted initiatives. The aim of the current review is to summarize the mediators of the association between CSA and revictimization in romantic relationships in adolescence or adulthood.

Different theoretical models have been proposed to frame the exploration of the sexual revictimization phenomenon in CSA victims or the risk of DV/IPV victimization (Hébert et al., 2012). These models could help to clarify the mechanisms involved in the revictimization risk for multiple forms of violence in CSA survivors’ intimate relationships.

Conceptual models of CSA revictimization

One of the first conceptual models on the impact of sexual abuse is the traumagenic dynamic model presented by Finkelhor and Browne (1985). The model suggested four processes by which sexually abused victims may come to develop negative outcomes, including a heightened vulnerability to revictimization: powerlessness, stigmatization, betrayal and traumatic sexualization. Powerlessness may lead CSA survivors to believe in the inefficiency of their own actions to manage future acts of aggression. Stigmatization may increase the risk of isolation and substance/alcohol abuse, and has been associated with elevated risk of revictimization. Stigmatization may also impede on the development of efficient self-protection strategies and impact survivors’ abilities to identify high-risk situations. Sense of betrayal and disillusionment may lead the child to search for or avoid significant relationships and thus may be associated with dependent or conflictual dynamics in later intimate relationships. Finally, traumatic sexualization may be linked to risky behaviors of frequent casual sex or compulsive sexual behaviors that may expose CSA victims to a heightened risk of revictimization.

The lifestyle/routine activities theories posit that all kinds of victimization are more likely to occur when there is a particular conjunction in time, place, and characteristics of the individuals that are there (Cohen & Felson, 1979). A “suitable” victim (e.g., isolated, under the influence of alcohol) in a non-monitored area, encountering an offender seeking aggression opportunities will be at higher risk to sustain violence than if the convergence of these time, location, and characteristics are not in place. Adolescents who adopt risky behaviors (such as frequenting deviant peers, using alcohol and drugs) under a low parental monitoring may be more likely to be victimized because of increased odds to encounter and interact with people that use violence.

The ecological model (Bronfenbrenner, 1979) has been adapted by Grauerholz (2000) to the specific context of CSA revictimization to broaden or expand the comprehension of factors involved in the lifespan trajectories of child victims. This model included four spheres (i.e., ontogenic, micro, exo and macro systems) that organize individual, relational, environmental, and societal risk factors for CSA revictimization. Characteristics of the victim (e.g., being a girl – ontogenic system), relational context of the aggression (e.g., family abuse – micro system), characteristics of the environment (e.g., low parental monitoring – exo system), and social norms (e.g., attitude of acceptation toward violence – macro system) interact together to heighten risk of sexual revictimization. The ecological model is a meta-model that could be used to draw interactive pathways from CSA to DV/IPV revictimization.

The use of conceptual models has helped to identify risk and protective factors associated with revictimization in CSA victims. In recent years, more complex models exploring the mechanisms involved in revictimization have been developed and researchers have investigated potential mediators in the association between CSA and DV/IPV. The identification of such mediators is a necessary step to develop evidence-based prevention programs that target the specific factors contributing to the heightened risk of revictimization in CSA survivors. In one recent review, Cascardi and Jouriles (2018) analyzed the data regarding four mediators (i.e., maladaptive attachment, emotion dysregulation, hostility, and emotional distress) in the association between various forms of child maltreatment and DV. While the review has provided insightful insights, it did not examine potential mediators involved in the specific links between CSA and experiences of dating violence, and did not include experience of violence in adult intimate relationships in their outcomes. Scoglio et al. (2019) conducted a systematic review of risk and protective factors involved in the link between CSA and subsequent interpersonal trauma in CSA survivors. Yet, the authors did not examine the mechanisms underlying the processes from the risk and protective factors and revictimization, and did not specifically focus on revictimization in the context of intimate relationships.

The aim of this systematic review is to provide a comprehensive summary of available data regarding the mediators of the association between CSA and revictimization in romantic relationships. The mechanisms that have been identified to date in empirical studies will be considered in light of the conceptual models presented (i.e., traumagenic dynamics, lifestyle/routine activities, and ecological models).

Methods

Literature search

The most commonly-used databases (PsycInfo, Psycarticle, Google Scholar, and Medline) were used to identify articles that met the inclusion criteria: peer-reviewed articles, published in English or French, that examined a mediator in the association between CSA and DV/IPV victimization in adolescence or adulthood. Multiple keywords were used to identify articles, e.g., child sexual abuse, CSA, child maltreatment, child abuse, dating violence, DV, IPV, couple violence, relational violence, co-occurrence, revictimization, polyvictimization, indirect effect, mechanism, mediation, mediator, path analysis. Titles and abstracts of manuscripts were then screened for eligibility. Relevant articles meeting inclusion criteria were retrieved and the content was coded to ensure eligibility. This process led to a final sample of 18 studies, as shown by the flowchart presented in Figure 1.

Figure 1. Flowchart of the Systematic Review Process

Coding procedures and data extraction

A codebook was developed to gather information about the study’s characteristics: sample characteristics (i.e., sample size, gender and age of the participants, sample), methodological design, measures and prevalence of CSA and DV/IPV, information about mediators (i.e., measures used, main results), and quality. All articles that were retrieved after the titles/abstracts screening process (n = 92) were coded with an inter-rater procedure to evaluate eligibility. Then, two members (first and second author) of the research team coded the 18 articles retained to ensure the evaluation of their quality and to extract all available data. The assessment of study quality was based on recommendations from the Cochrane collaboration (Higgins & Green, 2011) using seven criteria: sampling, representativeness, sample size, missing data, quality of the measurement tools, appropriateness of the analyses, and selective reporting.

Inclusion and exclusion criteria

To be included, studies needed to have examined the indirect effect of some factor(s) (i.e., individual, interpersonal, situational, or environmental characteristics) on the relationship between CSA and DV or IPV victimization. To obtain an overview of the mediators that have been explored in the association between CSA and DV/IPV, studies that operationalized CSA victimization in a composite score of child maltreatment experiences were retained as well as studies measuring CSA specifically. This strategy was used to capture all hypotheses that have been explored to better inform future studies and intervention policies.

Whereas many studies assessed both DV/IPV perpetration and victimization, the focus of this investigation was specifically on victimization. Therefore, for a study to be included, DV/IPV victimization had to be assessed separately from perpetration, and it was also a requirement that the measure of violence referred to acts perpetrated specifically by a romantic partner; studies in which the identity of the perpetrator could not be determined were excluded. Finally, only studies available in English or in French were reviewed, and studies that did not present original data (editorial, commentary or literature review) were excluded. For further details after the screening process, please refer to the flowchart presented in Figure 1.

Data analysis

Main results

Prevalence rates of CSA and DV/IPV victimization were collected and are presented in Table 1 for cross-sectional and longitudinal studies. For prevalence rates, no summary could be computed because of the vast differences observed in the measures used to assess CSA and DV/IPV. Mediators examined by the studies are descriptively presented in Table 1. Significant effects of mediators are identified in the table with an asterisk.

Table 1. Methods and Main Results from the Studies Included in the Review

Reference	N for sample (% women)	Measures	Prevalence of CSA and DV/IPV	Mediators tested	Quality of study
Cross-sectional studies
Alexander (2009)	93 women seeking services in IPV agencies (100%)	CSA: 1 homemade item IPV: CTS-2	CSA: 50.0% of multiply IPV victimized 17.5% of singly IPV victimized	Affect dysregulation, dissociation	High risk of bias
Atmaca and Gençöz (2016)	222 women (100%)	CSA: CTQ IPV: CTS-2	CSA: 12.2% IPV (phy): 41.9% IPV (psy): 82.9% IPV (sex): 35.6% IPV (injuries): 23.9%	Young domains: Disconnection and rejection^†, impaired autonomy and performance, impaired limits, other-directedness, overvigilance and inhibition, emotional Young 18 Schemas: emotional deprivation^†, vulnerability to harm or illness^†	High risk of bias
Bender et al. (2003)	362 African American women seeking services in public hospitals (100%)	CSA: CTQ IPV: Index of Spouse Abuse (Hudson & McIntosh, 1981)	CSA: n = 360 IPV (phy): n = 288 IPV (other than phy): n = 298	Social support for physical IPV^†, social support for nonphysical IPV*	High risk of bias
Hébert, Daspe et al. (2017)	8 194 adolescents (57.8%) (unweighted sample) 6 531 (weighted sample)	CSA: 2 items adapted from Finkelhor, Hotaling, Lewis & Smith (1990) DV: adapted from CADRI	CSA: 10.3% CSA (girls): 14.9% CSA (boys): 3.9%	PTSD symptoms*	Low risk of bias
Herrero et al. (2018)	Subsample of 23 863 women	CSA: 5 homemade items IPV: 14 homemade items	CSA: 8.6% IPV (psy): 15.9% IPV (phy): 7.6% IPV (sex): 0.2%	Traditional partner^†, violent partner^†	Low risk of bias
Kahya (2018)	288 married women from Turkey (100%)	CSA: short form of CTQ IPV: CTS-2	CSA: 34% IPV (psy): 83.3% IPV (phy): 14.6% IPV (sex): 31.6% IPV (injuries): 2.1%	Hostility, rejection sensitivity*	High risk of bias
McMahon et al. (2015)	25 778 adults (54%)	CSA: adapted from CTQ DV: adapted from CTS (Straus, 1979)	IPV (women): 1.6% IPV (men): 2.7%	Child maltreatment^†	High risk of bias
Messing et al. (2012)	1 150 female nurses (100%)	CSA: 1 homemade item IPV: Abuse Assessment Screen (Rabin, Jennings, Campbell, & Bair-Merritt, 2009)	CSA: 17.3% IPV (psy): 23.4% IPV (phy): 19.4% IPV (sex): 8.3%	PTSD symptoms^†	High risk of bias
Rothman et al. (2011)	456 pediatric emergency patients (55.7%)	CSA: adapted from Adverse Childhood Experiences (Felitti et al., 1998) IPV: adapted from Safe Dates (Foshee et al., 1998; 2001)	CSA (3 times or +): 32% DV: 65% DV (psy): 68% DV (phy): 58% DV (sex): 13%	Drinking style, problem behavior (boys) ^†; (girls)*	High risk of bias
Ulloa et al. (2009)	Subsample of 327 undergraduate women (100%)	CSA: adapted from CTS and SES (Roosa et al., 1998) IPV: CTS-2	Not reported	Fearful dating experiences*	High risk of bias
Wekerle et al. (2001)	1 329 youth (58%) 224 youth from CPS (55%)	CSA: CTQ IPV: CIRQ	Not reported	PTSD symptoms^†	Low risk of bias
Wekerle et al. (2009)	408 adolescents from CPS (51.8%)	CSA: CTQ DV: CADRI	CSA (girls): 30% CSA (boys): 8% DV (girls): 63% DV (boys): 49%	PTSD symptoms*	Low risk of bias
Longitudinal studies
Feiring et al. (2013)	121 sexually abused children/adolescents from CPS (73%)	CSA (severity): score (Chaffin et al., 1997) derived from records of CPS DV: CIRQ	DV: 80.2% DV (psy): 79.3% DV (phy): 24.0%	Externalizing behaviors, abused-related shame, anger	Low risk of bias
Ihongbe and Masho (2018)	1 163Add Health study (52%), (unweighted sample)	CSA: 1 homemade item IPV: 3 homemade items	CSA: 5.3% IPV (phy): 23.2% IPV (sex): 6.4%	Age of sexual initiation^†	Low risk of bias
Iratzoqui and Watts (2019)	11 621Add Health study (non-available)(weighted sample)	CSA: 1 homemade item IPV: 4 homemade items	Not reported	DV victimization^†	Low risk of bias
Maas et al. (2010)	941 adolescents (46.5%)	CSA: 2 items from CTQ DV: 1 homemade item	CSA: 10%DV: 12%	Externalizing behaviors, internalizing behaviors, alcohol consumption	Low risk of bias
Rich et al. (2005)	551 college women (100%)	CSA: adapted from Child Sexual Victimization Questionnaire (Finkelhor, 1979) IPV: Dating conflict scale of CTS (Straus, 1979)	CSA (other than rape): 6.2% CSA (rape): 1.4% IPV (moderate, 2-months reference period): 4.2% IPV (severe, 2-months reference period): 1.3%	Adolescent dating violence, adolescent sexual victimization, depression	High risk of bias
Zamir et al. (2018)	80 women (100%)	CSA: medical or CPS records IPV: CTS	CSA: 27.5% IPV: 56.2%	Dissociation*	High risk of bias

Notes. CPS = Child Protective Services; CSA = child sexual abuse; CTQ = Childhood Trauma Questionnaire (Bernstein et al., 1994); IPV: intimate partner violence; CTS = Conflict Tactic Scale (Straus, 1979); DV = dating violence; CIRQ = Conflict in Relationship Questionnaire (Wolfe et al., 1998); CADRI = Conflict in Adolescent Dating Relationship Inventory (Wolfe et al., 2001); IPV (phy) = physical IPV; IPV (psy) = psychological, verbal or emotional IPV; IPV (sex) = sexual IPV; *p <.05 for full mediation. ^† p <.05 for partial mediation.

Quality of the studies

The recommendations of the Cochrane Collaboration’s guidelines (Higgins & Green, 2011) were used to assess the quality of the studies reviewed. To be considered at low risk of bias, a study has to meet good standard of quality for most criteria. When it did not, or when a major flaw on one criterion was likely to influence the validity of the results, studies were considered at high risk of bias. When the information was not available for several criteria, studies were classified as unclear risk of bias to avoid an uncertain judgment of quality. A categorical classification of the studies was used rather than a computation of a continuous summary score that assumes equal weights for each criterion, which can lead to misleading evaluations of the study quality (Higgins & Green, 2011).

Results

Description of the study set

A summary of the characteristics of the studies’ sample, design, method, results and quality is presented in Table 1. For both cross-sectional and longitudinal design studies, most of the studies were conducted in the United States. In cross-sectional studies (n = 12), most of the samples were composed of participants exclusively identified as female (n = 7), and the majority of the studies were carried out with samples of adults. Studies involving only female participants included a sample of undergraduates (Ulloa et al., 2009), emergency patients (Rothman et al., 2011), women seeking services in IPV agencies (Alexander, 2009), African American women seeking services in public hospitals (Bender et al., 2003), and two with adolescents or with a proportion of adolescents recruited in child protective services (CPS) (Wekerle et al., 2001, 2009).

For longitudinal studies (n = 6), samples differed widely across studies: being conducted with a clinical sample of sexually abused youth (Feiring et al., 2013), with a college sample (Rich et al., 2005), with adolescents and one with women from general population, or with data from the Add Health study (Ihongbe & Masho, 2018; Iratzoqui & Watts, 2019).

Mediators of the association between CSA and DV/IPV victimization

Cross-sectional studies

In the cross-sectional studies analyzed for this review, CSA was reported by 8.6 to 50% of the participants, with higher rates being reported in clinical samples. DV/IPV was reported by 0.2 to 83.3% of the participants, the most prevalent form of DV/IPV being psychological victimization and the least prevalent form being sexual DV/IPV.

One of the cross-sectional studies examined the direct effect of CSA on IPV victimization, controlling for the effect of maltreatment (McMahon et al., 2015), and this separately for men and women. In their sample, 1.6% of women and 2.7% of men reported IPV. The results obtained with a sample of 25,778 adults indicated that for both men and women, CSA was directly associated with IPV victimization, and indirectly associated with it by the latent construct of childhood maltreatment.

Traumagenic model

According to this model, sexually abused victims become more vulnerable to sexual revictimization through the sequelae of CSA, such as PTSD and dissociation symptoms, as well as cognitive and emotional processes.

Sequelae of CSA

PTSD symptoms were the most frequently examined mediators of the relationship between CSA and DV/IPV victimization. In our study set, four studies tested the impact of this variable. Hébert, Daspe et al. (2017) conducted a study with a large representative sample of teenagers to examine how PTSD symptoms mediated the association between CSA and different forms of DV victimization. In their sample, 10.3% of youth (14.9% of girls and 3.9% of boys) were categorized as victims of CSA. Three different mediational models, one for physical DV, one for psychological DV and one for sexual DV were examined. Results showed that PTSD symptoms fully mediated the association between CSA and each form of DV.

Messing et al. (2012) conducted a study with 1,150 female nurses. They examined the mediational role of PTSD in the association between CSA and psychological, physical, and sexual IPV. CSA was reported by 17.3% of the sample, lifetime psychological IPV by 23.4%, physical IPV by 19.4%, and sexual IPV by 8.3%. They found a direct association between CSA and PTSD symptoms and an indirect effect of CSA through PTSD symptoms on psychological and sexual IPV. Finally, CSA was associated with physical IPV in their model, but no mediational effect of PSTD was found.

Wekerle et al. (2001) carried out a study with 1,329 adolescents recruited from a community sample and 224 adolescents recruited from child protective services. They examined the mediational role of PTSD symptoms in the association between child maltreatment and victimization in boys and girls with a multiple hierarchical regression. Entering PTSD symptoms in the models for the general population sample and the child protective services sample made the effect of maltreatment become non-significant only for girls within the school sample, suggesting a mediational contribution of this variable for this subpopulation. Wekerle et al. (2009) examined PTSD symptoms in another study with an independent sample of 408 youth recruited in CPS. In their sample, 30% of girls and 8% of boys reported CSA. Concerning DV, 63% of girls and 49% of boys endorsed at least one experience of victimization in the last 12 months. In this study, the addition of PTSD symptoms in the multiple hierarchical regression model for girls DV victimization did not make the CSA effect become non-significant, suggesting no mediational effect (a victimization model was not tested for boys).

Dissociative symptoms were examined in one cross-sectional study as a mechanism that might explain the association between CSA and revictimization in intimate relationships. Alexander (2009) studied this association in a sample of singly and multiply victimized women and found that, contrary to expectations, dissociation did not mediate the association.

Emotional and cognitive processes

Atmaca and Gençöz (2016) hypothesized that cognitive vulnerabilities associated with child abuse and neglect may play a role in the repetition of early negative experiences of victimization. They tested two parallel multiple mediation models with the five domains and 18 associated schemas proposed in Young’s model, with a sample of 222 married women. In the sample, 12.2% of the women reported a history of CSA and 23.9% to 82.9% endorsed some form of IPV victimization, the most prevalent form being psychological IPV. With regard to the larger domains of cognitive/emotional schemas, the model explained 24% of the variance in IPV victimization; only Disconnection/rejection showed a significant path from child victimization to partner victimization in adulthood. The mediational model including Young’s schemas predicted 25% of the variance in IPV victimization. Among the indirect effects of the 18 schemas on the link between child victimization and IPV victimization, only Emotional Deprivation and Vulnerability to Harm or Illness were significant.

The mechanism of affect dysregulation was examined by Alexander (2009) in a study of 93 women victims of IPV. Her results showed no indirect effect of affect dysregulation between CSA and multiple IPV experiences. Kahya (2018) examined how hostility traits and sensitivity to rejection could mediate the association between child abuse and neglect and IPV victimization in a sample of 288 Turkish women. In this sample, 34% of women reported having sustained CSA, 83.3% psychological IPV, 14.6% physical IPV, and 31.6% sexual IPV. While child abuse and neglect were found to be associated with both hostility and rejection sensitivity, only the mediational path through rejection sensitivity was significant.

Lifestyle/routine activities theories

These theories suggest that revictimization is more likely to occur when there is a particular conjunction in time, place and in the characteristics of the victim and the aggressor, including their behaviors.

Rothman et al. (2011) conducted a study with 456 young emergency pediatric patients to examine the mediational role of drinking style and of problem behavior in the association between childhood victimization and DV victimization. In their sample, 32% of youth reported CSA on three or more occasions and 65% reported experiences of DV. When examining only drinking styles in the model, a partial mediation effect was observed for boys and girls in the association between child victimization and DV victimization. When entering both drinking styles and problem behaviors into the model, problem behaviors partially mediated the association between childhood victimization and DV victimization among boys, while it fully mediated the association among girls. Both for boys and girls, drinking style had no effect.

Ecological model

According to this model, characteristics of the victim, relational context of the aggression, characteristics of the environment, and social norms interact together to create risk for sexual revictimization.

Characteristics of the romantic partner

Herrero et al. (2018) explored the mediational role of the partner’s endorsement of traditional gender role beliefs and engagement in violent behaviors in other relationships in the association between CSA and IPV among 23,863 European women. Partner gender roles was assessed with one item referring to whether the participant felt they had an equal say with regard to the use of the household income while partner’s general violence was also measured with one item (i.e., Has your partner even been physically violent toward anyone outside the family?). CSA was reported by 8.6% of the sample, psychological IPV by 15.9%, physical IPV by 7.6% and sexual IPV by 0.2%. Their results demonstrated that having been sexually abused in childhood was associated with having a more violent partner that endorsed more traditional gender roles, which in turn was associated with higher levels of IPV. In their model, 22% of the total effect of CSA on IPV was explained by the partner’s traditional gender role endorsement and general violence.

Ulloa et al. (2009) examined how recent fearful dating experiences (e.g., coercion, lack of control, fear or shame in intimate relationship in the past year) mediated the association between CSA and IPV victimization with a sample of 327 college women. The model showed a fully mediational effect of fearful dating experiences.

Social support

Only one study considered the role of a protective factor, perceived social support, as a mediator in the association between CSA. Bender et al. (2003) conducted a study with 362 African American women who were seeking services in urban public hospitals. They examined the predictive value of child maltreatment forms separately for adult physical IPV and adult nonphysical IPV. The results showed that social support partially mediated the association between CSA and physical IPV, while it fully mediated the association between CSA and nonphysical IPV.

Longitudinal studies

In studies retained for this review, only six utilized a longitudinal design. These studies examined internalizing and externalizing behaviors, teen sexual initiation and victimization, dissociation, depression, anger and abuse-related shame as mechanisms involved in the association between CSA and victimization in intimate relationships. In the longitudinal studies selected, CSA was reported by 5.3 to 27.5% of participants and DV/IPV by 1.3 to 80.2%, depending of the form of intimate violence considered.

Traumagenic model

Feiring et al. (2013) completed a study with 121 victims of CSA recruited from CPS and specialized clinics; victims were aged 8 to 15 years old during the first wave of the study. They followed their development across six years. In their sample, 80.2% of youth reported having sustained at least one act of DV at wave 3 of the data collection; specifically, 79.3% reported psychological DV while 24.0% reported at least one occurrence of physical DV victimization. The authors examined pathways between CSA abuse severity (measured at Time 1), externalizing behaviors and abuse-related shame (measured at Time 1 and Time 2), anger (measured at Time 3) and DV victimization (measured at Time 3). The results of their structural equation model showed no association between CSA severity and other variables. Direct effects were observed for externalizing behaviors (Time 1) and anger (Time 3) on DV victimization (Time 3) and indirect effect of anger (Time 3) was observed from abuse-related shame (Time 2) to DV victimization (Time 3). Rich et al. (2005) used a 2-wave assessment design with 551 college women, 6.2% of which reported CSA other than rape, and 1.4% reported rape in childhood, to examine the mediational role of DV victimization, adolescent sexual victimization, and depression measured at Time 1 in the association between CSA (measured at Time 1) and DV victimization (measured at Time 2). Their results revealed no direct effect of CSA and no mediational effect was found. Zamir and colleagues (2018) used a subsample of 80 females from the Study of Risk and Adaptation to examine dissociation as a mediator (measured at age 19) in the association between child physical/sexual abuse (composite score measured before 17.5 years old) and IPV (measured at ages 20 to 32). In their sample, CSA was reported by 27.5% of women and IPV by 56.2% of the participants. Their results showed a partial mediation effect of dissociation in the association between child maltreatment and IPV.

Lifestyle/routine activities theories

Maas et al. (2010) conducted a longitudinal study with 941 youth to examine how externalizing behaviors, internalizing behaviors and alcohol consumption mediated the association between child maltreatment and DV victimization. In their sample, 10% of youth retrospectively reported CSA at wave 2 and 12% of them DV victimization at wave 1. They performed path analyses separately for men and women. Results showed only one direct association in the model for women, between maltreatment and internalizing behavior, and no path from child maltreatment in the model for men. No mediation effect was found in the association between child maltreatment and DV.

Iratzoqui and Watts (2019) used longitudinal data from the Add Health study. They formed a subsample of 11,621 participants that were interviewed and surveyed in waves I, II and III to examine how the association between CSA before the sixth grade (retrospectively reported at wave III) and IPV victimization (reported for the last 12 months at wave III) was mediated by DV victimization (measured at wave II). Their results revealed that both direct and indirect effects were significant, which supported a partial mediation.

Ihongbe and Masho (2018) also used a subsample of 1,163 participants from waves I, III and IV of the Add-Health study. CSA was assessed retrospectively at wave 1 and was reported by 5.3% of 18 to 26 years old participants. IPV was assessed at wave IV (when participants were 24–32 years old). Physical IPV was reported by 23.2% of the sample while sexual IPV was reported by 6.4%. The authors examined the mediating role of age at sexual debut in the association between CSA and physical and sexual IPV among boys and girls. Their results indicated that while CSA was associated with age of sexual initiation in both boys and girls, mediation of the relationship between CSA and physical and sexual IPV victimization was only present in girls. These results suggested that mediators may play a different role in boys and girls processes of revictimization over time.

Ecological model

No longitudinal studies explored variables that could be integrated within the ecological model.

Discussion

The aim of this review was to summarize the mediators of the association between CSA and physical, psychological and sexual revictimization in romantic relationships identified in past empirical reports. Through a systematic process of review, 18 studies were identified for analysis. Although this review contributes to a better understanding of the mechanisms involved in the intimate partner revictimization experiences of CSA victims, results remain inconclusive regarding a number of factors due to limited research in the area. Indeed, in the present review, PTSD symptoms is the mediator that has been the most investigated, and was found to be significant in three of the four studies reviewed.

Some mediators have been examined in two studies. For example, dissociation has shown inconsistent results. It failed to show a mediational effect in a cross-sectional study (Alexander, 2009), yet was found to act as a mechanism linking CSA to IPV in a longitudinal study (Zamir et al., 2018). Hostility/anger was also examined in two studies, but with consistent results across longitudinal and cross-sectional designs (Feiring et al., 2013; Kahya, 2018). Finally, some mediators have been shown to have a significant mediational effect but only in one cross-sectional study: emotional deprivation, vulnerability to harm and illness, having a traditional or a violent partner, fearful dating experience, and problem behaviors all mediated the association between CSA and revictimization. Some mediators have also been examined in longitudinal investigations but were found to have no effect: affect dysregulation, depression, and adolescent sexual victimization.

Cascardi and Jouriles (2018) conducted a review to examine the contribution of four mediators (i.e., maladaptive attachment, emotion regulation, emotional distress and hostility) in the association between all forms of child maltreatment and DV victimization and perpetration. Their review was focused on four specific mediators (putting aside other mediators that have been explored), and examined only DV experiences in adolescence. The current review is distinguished from the one conducted by Cascardi and Jouriles by exploring specifically the mediators identified in the relationship between CSA and later violence in romantic relationships, and builds knowledge on their review by including samples of adults, and a broader spectrum of mediators. For the most part, our results are consistent with the findings reported by Cascardi and Jouriles (2018). Maladaptive attachment, emotion dysregulation, emotional distress (such as depression, anxiety and PTSD symptoms), and hostility have been identified as mediators, sometimes in combination, in the association between maltreatment (including CSA) and DV victimization. In the present review however, affect dysregulation and depression were not found to be significant mediators, and no studies were retained based on our exclusion criteria that investigated maladaptive attachment as a mediator between CSA and DV/IPV victimization.

In the present review, several studies investigated mediators among female CSA victims. In studies that included both men and women, evidence suggested gender differences in mediators between CSA and revictimization (e.g., problem behaviors, PTSD symptoms, internalizing behaviors, age of sexual initiation); however, most of these mediators were only examined in one study each. The results of one study (Bender et al., 2003) also suggested that mediators may differ across DV/IPV outcomes (physical, psychological IPV). As such, exploring mediator in moderated meditational models would be essential in future investigations. The results of the present review also echoe the findings of Scoglio et al. (2019) in that very few studies have been found to address protective factors that may reduce the risk of revictimization. Only two of the 25 studies reviewed by Scoglio considered the role of parental care as a moderator of the risk of revictimization in CSA survivors.

Limitations and implications for future studies

This review reveals several important limitations in the current literature that hinder our capacity to draw firm conclusions regarding mediators of the relationship of CSA to DV/IPV. . First, the operationalization of DV/IPV and measures used to assess these constructs in the different studies are quite inconsistent. Questionnaires assessing DV and IPV demonstrated significant variation in how these concepts were measured and defined. Some measures used lifetime or 12-months reference period, and not all forms of DV/IPV were considered in each investigation (i.e., psychological, physical, sexual). Some studies also used an unvalidated single item or an invalidated adaptation of a questionnaire to measure DV/IPV. These disparities render comparison across studies challenging at best.

In the same way, different definitions of CSA in terms of age limit and of acts considered as sexual abuse were noted. While the CTQ (Bernstein et al., 1994) was use in most studies, some studies used a single or a few author created items to measure specific acts of sexual victimization with different time frame, for example: having private parts touched or being forced to touch someone’s private parts before the age of 17 (Alexander, 2009), being sexually touched by an adult or having a sexual relation with an adult before the 6^th grade (Ihongbe & Masho, 2018; Iratzoqui & Watts, 2019) and being touched in a unwanted way or being implied in sexual activities as a child (Messing et al., 2012). These differences may have diminished or potentiated the mediational effects, depending on the definition used. In addition, several studies rely on composite scores of maltreatment forms instead of differentiating the effect of forms of maltreatment, which may mask the role of some mediators in the association between CSA and specific forms of DV/IPV.

Regarding the samples and the designs of the studies reviewed, only a few were conducted with representative samples; this limits the generalizability of the results. Samples were often limited to women or included a small number of men. Thus, it is difficult to ascertain whether the mediators involved are gender-specific, even if the results of some studies indicate that differences could be observed. Moreover, few studies considered the role of cultural diversity, and no study retained for the review considered sexual or general minority groups; thus, results cannot be applied to these populations. This is unfortunate given that research has shown potentially high rates of victimization in these groups.

Even in cases where significant mediation was found, the variance explained by the mediators was small, suggesting that other factors not explored may play an important role in the revictimization process of CSA victims. For example, attachment has been identified as a mediator in the association between child maltreatment and DV in the review performed by Cascardi and Jouriles (2018) and thus may be a relevant factor to consider when examining the relation between CSA and DV or IPV revictimization. Cascardi and Jouriles (2018) also suggested that multiple mediator models, e.g., combining emotional regulation and emotional distress, could better explain the revictimization process compared to a unique mediator. In the present review, only a small proportion of the studies examined combination of mediators to better understand the victimization pathways from CSA to DV/IPV. While PTSD was identified as a relevant mediator, future studies will need to document whether survivors had access to therapy following disclosure, as intervention could both affect whether a child develops PTSD, and whether he or she resolves PTSD. As such, treatment received or lack thereof could represent a confounding variable.

Finally, one of the main limitations of this review is that the majority of studies identified used cross-sectional design. This method has often been criticized because using cross-sectional data has been known to generate substantial bias in the estimation of the mediation effect (Maxwell & Cole, 2007; Mitchell & Maxwell, 2013; O’Laughlin et al., 2018). Only eight studies in the set selected for review met the criteria of low risk of bias and six have relied on a longitudinal design, which represent a design allowing for testing mediation effects. In the majority of longitudinal studies, factors that have been examined were found to be non-significant. Therefore, the mechanisms involved in revictimization of CSA survivors in romantic relationships remain unclear. Studies relying on longitudinal designs with representative samples are clearly needed to identify relevant variables to target in prevention efforts and interrupt the victimization trajectory of victim of CSA.

Conclusion

Prevention programs for DV and healthy relationships are often offered to youth in school settings (Cornelius & Resseguie, 2007; Whitaker et al., 2006). Adolescence is a crucial period in which relational and communicational abilities are developed in first intimate relationship experiences (Connolly & McIsaac, 2009) and conflict resolution patterns of interactions may persist into adult relationships, including those characterized by violence (Gómez, 2011). Programs are disseminated to adolescent populations in the hopes of preventing DV, and by extension IPV revictimization. While these efforts are laudable and should continue, the high prevalence of revictimization underscores the need to provide effective and timely intervention to victims of CSA. Intervention programs should address empirically-identified mediators of the relationship between CSA and revictimization, including anger, abuse-related shame, and sensitivity to rejection in relationships. The evidence also points to PTSD symptoms as a mechanism associated with revictimization; as such, treatments targeting these symptoms in sexually abused children (Trauma-focused Cognitive Behavioral Therapy, Cohen et al., 2017) or adults (Emotionally Focused Therapy, MacIntosh & Johnson, 2008) could contribute to a reduced risk of revictimization in the context of romantic relationships in adolescence or adulthood.

Declaration of interest statement

The authors disclose no conflict of interest.

Additional information

Funding

This work was supported by the Social Sciences and Humanities Research Council of Canada [950-230791].

Notes on contributors

Martine Hébert

Dr. Martine Hébert (Ph.D. in psychology) is the Tier 1 Canada Research Chair in Interpersonal Traumas and Resilience and the co-holder of the Interuniversity Research Chair Marie-Vincent on child sexual abuse. She is full professor at the Department of sexology at the Université du Québec à Montréal, and director of the Sexual Violence and Health Research Team ÉVISSA, an interdisciplinary group aiming to achieve a comprehensive understanding of sexual violence including consequences on mental, physical and sexual health in different developmental contexts (childhood, adolescence and early adulthood). She served as the leading investigator of the Youths’ Romantic Relationships Project, a representative study of over 8000 teenagers in Quebec and, with her partners, is developing a school-based program aiming to promote healthy relationship and prevent dating violence. She is also funded by the Canadian Institutes of Health Research (CIHR) to conduct a longitudinal study identifying crucial mediators involved in resilience pathways among sexually victimized youth.

Andréanne Lapierre

Andréanne Lapierre (Ph.D. candidate) holds a bachelor’s degree in special education and is a clinical psychology student at Université du Québec à Montréal. She is a member of Sexual Violence and Health Research team and is interested in interpersonal violence. Her doctoral research centers on conflict resolution in adolescent dating relationships and on the contextual processes leading to dating violence perpetration. She recently contributed to a meta-analysis studying family and peer risk and protective factors for dating violence victimization and to a national report on dating violence.

Heather B. MacIntosh

Dr. Heather MacIntosh, Ph.D. is Associate Professor in the M.Sc.A Couple and Family Therapy Program at McGill University where she is the recipient of the H. Noel Fieldhouse Award for Distinguished Teaching. Dr. MacIntosh is a clinical psychologist and principal investigator on SSHRC and FQRSC funded research projects examining the impact of early life trauma on the process and outcome of couple therapy, and the development and maintenance of healthy adult relationships. She is the author of the recently released book: Developmental Couple Therapy for Complex Trauma a Manual for Therapists by Routledge Press, a treatment manual outlining her psychoanalytically informed, evidence based treatment model for working with couples dealing with the impacts of complex trauma as well as a number of peer reviewed articles and chapters in the area of childhood trauma, interpersonal outcomes and the process of couple therapy.

A. Dana Ménard

Dr. A. Dana Ménard is a clinical psychologist and adjunct professor in the Faculty of Science at the University of Windsor. She earned a PhD in clinical psychology from the University of Ottawa in 2013, an MA in experimental psychology from Carleton University in 2007 and a BSc in psychology from the University of Ottawa in 2005. She has held previous academic appointments at the University of Western Ontario and at Wayne State University in Detroit, Michigan. As a clinician, Dr. Ménard has worked at Detroit Receiving Hospital, the London Health Sciences Centre, the Royal Ottawa Hospital and the Ottawa hospital, among others. Aside from her work on the impacts of child sex abuse on adult functioning, she has also researched and published in the areas of optimal sexual experiences, sexuality in the media, student mental health and wellbeing and academic lab safety.