Maternal intimate partner violence victimization and ...

Child Abuse & Neglect 82 (2018) 23–33

Contents lists available at ScienceDirect

Child Abuse & Neglect journal homepage: www.elsevier.com/locate/chiabuneg

Maternal intimate partner violence victimization and child maltreatment

T

⁎

Zohre Ahmadabadia, , Jackob M. Najmana,b, Gail M. Williamsa, Alexandra M. Clavarinoc, Peter d'Abbsa,d, Amanuel Alemu Abajobira a

School of Public Health, The University of Queensland, Herston, Queensland, 4006, Australia School of Social Sciences, The University of Queensland, St Lucia, Queensland, 4072, Australia c School of Pharmacy, The University of Queensland, Woolloongabba, Queensland, 4102, Australia d Menzies School of Health Research, Spring Hill, Queensland, 4000, Australia b

A R T IC LE I N F O

ABS TRA CT

Keywords: Maternal intimate partner violence child maltreatment

There is some limited evidence of an association between maternal intimate partner victimization (IPV) and children’s experience of maltreatment. Using data from a longitudinal study, we examine whether this relationship is independent of range of potential confounders including socioeconomic, familial and psychological factors. Data were taken from the 14 and 30-year followups of the Mater-University of Queensland Study of Pregnancy (MUSP) in Australia. A subsample of 2064 mothers and children (59.0% female) whose data on maternal IPV and child maltreatment was available, were analysed. In families with maternal IPV, two in five children reported being maltreated, compared to one in five children maltreated in families without maternal IPV. Except for sexual maltreatment which was consistently higher in female offspring, there was no gender differences in experiencing different types of maltreatment in families manifesting maternal IPV. Although both males and females were at increased risk of child maltreatment in families where mothers were victimized by their male partners, male children were more likely to be emotionally maltreated. The main associations were substantially independent of measured confounders, except for father’s history of mental health problems which attenuated the association of maternal IPV victimization and male offspring’s physical abuse. Our findings confirm that there is a robust association between maternal IPV and child maltreatment. Both maternal IPV victimization and child maltreatment co-occur in a household characterized by conflict and violence. Consequences of IPV go beyond the incident and influence all family members. Efforts to reduce child maltreatment may need to address the greater level of IPV associated with the cycle of family violence.

1. Introduction Intimate partner violence (IPV) and child maltreatment are major public health concerns worldwide (Devries et al., 2013; Gilbert et al., 2009; Krug, Mercy, Dahlberg, & Zwi, 2002). There are controversies about whether these phenomena are distinct, causal, sequential, or co-occurrent (Bidarra, Lessard, & Dumont, 2016). Children’s exposure to parents’ IPV has been described as a type of child maltreatment (MacMillan, Wathen, & Varcoe, 2013). Literature about the association between IPV and child maltreatment has

⁎

Corresponding author at: School of Public Health, The University of Queensland, Herston Road, Herston, Queensland, 4006, Australia. E-mail addresses: [email protected] (Z. Ahmadabadi), [email protected] (J.M. Najman), [email protected] (G.M. Williams), [email protected] (A.M. Clavarino), peter.d'[email protected] (P. d'Abbs), [email protected] (A.A. Abajobir). https://doi.org/10.1016/j.chiabu.2018.05.017 Received 8 February 2018; Received in revised form 4 May 2018; Accepted 15 May 2018 0145-2134/ © 2018 Elsevier Ltd. All rights reserved.


Z. Ahmadabadi et al.

mostly focused on adult experiencing IPV as a consequence of childhood maltreatment and not vice versa (Kong, Roh, Easton, Lee, & Lawler, 2016; McMahon et al., 2015; Whitfield, Anda, Dube, & Felitti, 2003). Our knowledge about whether IPV victimization might be associated with or predicts child maltreatment is limited (Ko Ling, 2011). Two major approaches have been developed to explain domestic violence, which includes both partner-to-partner and parent-tochild violence: one is the feminist perspective (Dobash & Dobash, 1979; Hester, Kelly, & Radford, 1996; Yllo, 1993; Yodanis, 2004) which views the domestic violence essentially male-perpetrated and a reflection of unequal power relationships in the family. From this perspective, gendered social norms are at the intersection of intimate partner violence and child abuse (Namy et al., 2017). An alternative approach is the gender symmetry model of family violence which focuses on bidirectional and common couple violence and suggests that both men and women engage in domestic violence (Capaldi, Kim, & Shortt, 2007; Dutton, Nicholls, & Spidel, 2005; Langhinrichsen-Rohling, 2010; Straus, 2008). Given pathways to child maltreatment are complex (Munro, Taylor, & Bradbury-Jones, 2014), the association between parents’ IPV and their children’s experience of maltreatment may reflect a broader context for both exposure and outcomes of interest. IPV and child maltreatment may occur as consequences of common social, environmental, familial or individual risk factors (prevention, 2015). These different conceptual orientations can be incorporated into a three-level social-ecological model: At the societal level are social, cultural and economic factors which create a context for family violence. Beside gender inequality, limited or unequal resources that family members possess (e.g., income or education) may enhance competition, disagreement and abuse (Resource theory; Ahmadabadi, Najman, Williams, & Clavarino, 2017; Allen & Straus, 1980; McCloskey, 1996). Stressful life events might contribute to a sense of frustration and powerlessness, diminish family’s coping mechanisms, and ability to problem solve and contribute the level of interpersonal violence (Family stress theory; Cano & Vivian, 2001; Conger et al., 1994). Social isolation and living in a non-supportive, disadvantaged and high-crime residential neighborhood is associated with a greater risk of family violence (Social disorganization theory; Gracia & Musitu, 2003; Pinchevsky & Wright, 2012). The familial-level, risk factors for spousal and child abuse include marital instability, changes in family structure, single-parent or step families and large family size (Family composition perspective; Campbell et al., 2003; McLanahan & Sandefur, 1994). It has been suggested that members of a fragile, incomplete and transitional household experience difficulties in attachment, communication, adjustment and conflict resolution (McLanahan & Beck, 2010; van Ijzendoorn, Euser, Prinzie, Juffer, & Bakermans-Kranenburg, 2009). Finally are individual characteristics including age, psychopathology and behavioral factors (Ali & Naylor, 2013; Dutton, 1995). Violence perpetration – toward a partner or children- is associated with poor psychological health including depressive symptoms, anxiety, impulsivity, and emotional dysregulation (Conron, Beardslee, Koenen, Buka, & Gortmaker, 2009; Shorey et al., 2012). IPV victimization lmay ead to devastating and long-lasting outcomes including psychological impairment (Fergusson, Horwood, & Ridder, 2005; Goodman, Smyth, Borges, & Singer, 2009). There is evidence showing that women’s poor mental health may be an independent risk factor for partner’s violence (Capaldi & Crosby, 1997; Lehrer, Buka, Gortmaker, & Shrier, 2006). In addition, some behavioral difficulties manifested by children, including aggressive behaviors or attention deficit hyperactivity disorder (ADHD) may be less tolerated by the parents, especially those involved in IPV, and place children at increased risk of maltreatment. These problems may also stimulate stress and contribute to conflict and violence in the family (Johnston & Mash, 2001). Although the literature suggests that socio-economic, familial and individual risk factors may all contribute to IPV and child abuse (Smith Slep & O’Leary, 2001), no empirical study has been carried out to test this possibility. Understanding the nature of the association between maternal IPV victimization and child maltreatment should contribute to better efforts and resources to initiate successful child maltreatment prevention (Conron et al., 2009). There is increasing concern that despite the existence of a range of programmes for the reduction of child maltreatment, effective interventions targeted at reducing parental violence are not easily found (Macmillan et al., 2009). Previous research has a number of limitations that need to be addressed: An influential body of the relevant literature has been based on reports of government agencies or the clinical settings, neither of which comprises a representative sample of the population (Dutton, Hamel, & Aaronson, 2010; Langhinrichsen-Rohling, 2010). Moreover, concerns have been raised about the accuracy of retrospective reports of inter-parents’ violence made by children and that there is a need to use women’s own reports of victimization (Hungerford, Ogle, & Clements, 2010; Maynard, 1994). Importantly, evidence about gender differences in the association of maternal IPV victimization and child maltreatment is complicated and inconsistent (Davies & Lindsay, 2004; Suh et al., 2016). With the exception of sexual abuse which is more often experienced by female children (Briere & Elliott, 2003), gender differences in other forms of child maltreatment have not consistently been reported (Thompson, Kingree, & Desai, 2004). Particularly, it is still not known whether male and female offspring are maltreated differently in a family characterized by maternal victimization. Using data from a longitudinal birth cohort study of a large population-based sample, this research investigates the association between maternal IPV victimization in a heterosexual relationship and child maltreatment, separately for male and female offspring. We also examine whether this association is independent or reflects a range of potential confounders including socio-economic, familial and psychological factors. 2. Method 2.1. Participants We used data from the Mater Hospital and University of Queensland Study of Pregnancy (MUSP) (Najman et al., 2005). MUSP is a prospective study of 7223 consecutive women and their offspring who were recruited initially at their first antenatal visit to the Mater 24



Public Hospital in Brisbane between 1981 and 1983. The cohort was followed up when the study children were 6 months, 5 years, 14 years, 21 and 30 years old. At each phase of the study, written informed consent was obtained from mothers and /or their children. The analyses for the present study were conducted on the data from baseline, 5, 14 and 30 year follow ups. At the 14 year follow up, 4444 mothers completed questionnaires about their occupational, financial and marital status as well as their own experience of IPV. At the 30 year follow up, 2425 offspring completed the Childhood Trauma Questionnaire (Bernstein et al., 2003). The main analysis for this study was conducted on a subsample of 2064 mothers and children whose data on IPV victimization and child maltreatment was available. Overall 59.0% of the offspring participants were females and the mean age was 30.30 (SD = 1.13). The average age of mothers at the 14 year follow up was 39.8 years (SD = 4.9). 2.2. Measurement 2.2.1. Maternal IPV victimization At the 14-year follow-up, mothers completed a 7-item questionnaire about their last year experiences of victimization by a male partner. They were asked whether their partners had done any of the following during a disagreement: yelled at, insulted, sulked or refused to talk, threw something at, pushed grabbed or shoved, tried to hit and hit then. These items are taken from other measures of domestic violence, such as the Conflict Tactics Scale (Straus, 1979). Response options comprised never ( = 1), Sometimes ( = 2) and Often ( = 3). Items were summed and averaged (α = 0.70; mean: 1.31 (SD = 0.2)) and cases with score one standard deviation above the mean (≥1.51) were considered as victimized. 2.2.2. Childhood Trauma Questionnaire For measuring child maltreatment we used the Childhood Trauma Questionnaire (CTQ–SF) (Bernstein & Fink, 1998; Bernstein et al., 2003). The CTQ is a retrospective and self-administered questionnaire which has been used worldwide. It has good psychometric properties in general population and clinical samples (Bernstein & Fink, 1998; Bernstein et al., 2003; Lobbestael, Arntz, Harkema-Schouten, & Bernstein, 2009). It contains 25 items to measure five dimensions of childhood maltreatment: emotional abuse (5 items including being hated, hurt or insulted; α in the current study = 0.85), physical abuse (5 items including being beaten, bruised or punished with object ; α = 0.76), sexual abuse (5 items including being sexually touched, molested or abused; α = 0.95), physical neglect (5 items including not having enough food or wearing dirty clothes; α = 0.55), and emotional neglect (5 items including not felt loved or important or close to family members; α = 0.89). Participants were asked to respond to each item with regard to “when they lived with parents”. Response options comprise from never ( = 1) to very often ( = 5), producing scores of 5 to 25. We used the recommended cut-off scores to categorize cases into abused and not-abuse (emotional abuse (≥ 13), physical abuse (≥ 10), sexual abuse (≥ 8), emotional neglect (≥ 15) and physical neglect (≥ 10). We combined two forms of neglect to have a more stable and consistent measure of child maltreatment. 2.2.3. Possible confounders 2.2.3.1. Socio-economic factors. Maternal education was measured at first clinic visit and included high school or less, diploma and college and university. At the 14 year follow up mothers were asked about their own and their partner’s gross income. The Australian poverty line at each phase was used to categorize the pooled family income into two categories of poor and higher. Present occupational status was separately asked for mothers and their partners. Social problems of the residential environment at 14 years was assessed using the question How much are the following a problem in the area where you live? Problems included vandalism, house burglary, car theft, drug abuse, violence on streets, unemployment, noisy driving, alcohol abuse and school truancy. Areas with 8 or 9 problems were considered as problem area. Family social network was measured by 5 questions which comprised how many friends and relatives to which they felt close, met regularly and from whom they received support. For each question respondents chose from five-point scale (none = 1 to more than 15 = 5). After summing the items up, network size below 2 was categorized as small. 2.2.3.2. Familial factors. At 14 years, maternal marital status was measured by the question what is your present marital status. Single and separated/divorced/widowed women were excluded and present marital status included two categories of living together and married. Length of relationship was assessed by the question how long has the present relationship lasted (recoded to less than 10 years and more). Mothers were also asked if their present partner was the father of the child (stepfather vs. biological father). Number of children who usually lived with the mother at the 14 year follow up were classified into nil to two children and 3 children and more. 2.2.3.3. Psychological factors. Maternal depression was assessed at first clinic visit, 5 and 14 year follow-ups using a modified version of Delusions-Symptoms-States Inventory (DSSI (Bedford & Foulds, 1977). Seven questions measured how frequently the respondent had recently felt such symptoms as hopelessness, loss of interest or difficulty sleeping. A symptom was counted if the respondent sometimes and more experienced the feeling (depression scores ranged from 0 to 7). To create the variable Maternal chronic depression over 14 years, three depression scales were averaged and mothers were classified into chronically depressed (> 2.5) moderately depressed (1–2.5) and non-depressed (< 1). At 14 years mothers were asked whether their partners had ever been treated for mental or emotional problems. For measuring child mental health, mothers completed a modified child behavior checklist (CBCL (Achenbach & Edelbrock, 1991) at 5 years. This questionnaire measures internalizing behavior (anxious/ depressive, withdrawn and somatic behaviors), aggression and attention deficit hyperactivity disorder (ADHD). In the current study, a total score was used and those who scored above the 90th percentile were considered as having behavioral problems. All offspring respondents in the 14 and 30 year follow ups were about 14 years old (male = 13.90, SD = 0.32; female = 13.91, 25



SD = 0.31) and 30 years old (male = 30.39, SD = 1.17; female = 30.25, SD = 1.11), so we did not adjust for age variability in the sample. 2.3. Statistical analysis For comparing males and females by study variables, descriptive statistics and chi square were used. Univariate and multivariate logistic regression were conducted to estimate the risk of each form of child maltreatment and Odds ratio (OR) and 95% confidence intervals (CI) were reported. In the current study child maltreatment forms were not mutually exclusive. Statistical analyses were carried out using STATA-13 and SPSS-24. 3. Results Table 1 presents information about the total sample and compares male and female offspring using potential covariates, maternal IPV victimization and experience of childhood maltreatment. About 14.0% of mothers reported experiencing IPV in their relationships at 14 years. Some 8.8% of offspring reported experiencing emotional abuse, 7.5% physical abuse, 7.4% sexual abuse and 14.3% Neglect. No significant gender difference was found in covariates and maternal IPV, except female offspring reported higher rates of childhood emotional and sexual abuse than did males. Bivariate associations between possible confounders and maternal IPV victimization and child maltreatment for female and male children are represented in Table 2. Amongst socio-economic variables, being aboriginal-islander, mother’s teenage pregnancy, small social network, low family income, and mother’s partner’s unemployment significantly predicted both females’ child maltreatment and maternal IPV victimization. Residential problems was the only significant predictor of exposure and outcome in male offspring. Interestingly, mother’s low education and unemployment were not significant predictors of maternal IPV victimization. Among familial factors, being in a stepfamily, cohabiting and length of relationship less than 10 years were the most significant risk-factors for maternal IPV and child maltreatment. Number of children living with mother had no significant association with her victimization. All measured psychological problems of mother, father and child were associated with both maternal IPV victimization and child maltreatment for males and females. To examine the effect of possible confounders on the primary association between maternal IPV and child maltreatment, we progressively developed 4 multivariate models for each category of potential confounders, which were associated with both exposure and outcome of interest (Table 3). The first model presents the crude association of maternal victimization and forms of child maltreatment (model 1). For model 2 the primary association was adjusted for socio-economic factors. Family structure characteristics were then added to model 3. In the final model psychological variables, including maternal depression, mother’s partner mental problem and child behavior problems, were included. IPV was strongly associated with emotional abuse for males, but this was not observed for females. In a family male children report less emotional abuse than do females (Fig. 1), but they are more likely than females to be emotionally abused when mothers experience IPV victimization. The difference between male’s and female’s odds ratio is statistically significant (p = 0.049). Maternal experience of IPV had a consistent and significant association with physical abuse and neglect for both male and female offspring. Table 3 also shows that the associations between maternal IPV victimization and almost all forms of child maltreatment were independent of a wide range of socio-economic, familial and psychological factors. Only for male children, adjustment for psychological factors, attenuated the link between maternal IPV victimization and physical abuse. A further detailed analysis showed that amongst psychological variables, father’s history of mental health problems was confounding the association. Fig. 1 compares rates of child maltreatment within and between male and female offspring in families with and without maternal IPV victimization. Except for the sexual abuse in males, all children - either males or females - experience higher rates of emotional and physical abuse as well as well as neglect in families with maternal IPV victimization. Rates of child maltreatment are similar for male and female offspring within these families. 4. Discussion This study has investigated the association between maternal IPV victimization and child maltreatment, separately for male and female offspring. We assessed whether children in families with maternal victimization are more likely to be maltreated. We also tested three categories of variables including social and economic determinants, family structure and individual mental health problems which might confound the relationship between maternal IPV victimization and child maltreatment. The main hypothesis was that the association between maternal IPV victimization and child maltreatment may reflect confounding and common causes. We found that two in five victimized mothers (36.1%) and one in five (21%) non-victimized mothers had children who reported being maltreated (p < 0.001). Neglect followed by sexual abuse in females and physical abuse in males were the most frequent forms of maltreatment reported by children whose mothers were victims of intimate partner violence. Rates of maltreatment in female offspring appeared to be greater than rates experienced by males, however, male children in the context of maternal intimate partner violence, were at higher risk of self-reported emotional maltreatment. The main associations between IPV and most forms of maltreatment were independent of almost all measured confounders, except for father’s history of mental health problems which attenuated the primary association. Taken together, our findings show that there is a robust association between maternal IPV and child maltreatment and both cooccur in affected families. A number of possibilities need to be considered when interpreting the findings: 26



Table 1 Study variables by gender of Offspring. %

Socio-economic factors Racial background at FCV Caucasian Asian Abor-Islander Residential problem at 14 years Normal Problem area Family income at 14 years Higher ($20,800+) Low (-$20,799) Mother’s education at FCV Complete high school and above Incomplete high school Mother’s employment at 14 years Employed (fully/self/part-time) Unemployed Partner’s employment at 14 years Employed (fully/self/part-time) Unemployed Social network size at 5 years Adequate Small Mother’s age at pregnancy 20+ 13-19 Familial factors Mother’s partner at 14 years Biological father of child Stepfather Mother’s Marital changes over 7 to 14 years No change 1-2 changes 3+ changes Mother’s marital status at 14 years Married Living together Number of children at home at 14 years =