Severity of Psychological Maltreatment and Accumulative Risk for Psychopathology in Children of Mothers Exposed to Intimate Partner Violence
- First Online:
- Cite this article as:
- de la Vega, A., de la Osa, N., Granero, R. et al. J Fam Viol (2013) 28: 427. doi:10.1007/s10896-013-9521-1
- 524 Views
Psychological maltreatment (PM) is an extremely heterogeneous phenomenon that includes several subtypes. The aim of this work is to explore whether the accumulation of different subtypes of PM has a greater impact on the child’s psychopathology and functional impairment. One hundred and sixty-eight children and adolescents aged between 4 and 17 whose mothers had been exposed to intimate partner violence (IPV) participated. Psychopathology was assessed through a rating scale and a diagnostic interview. Polynomial contrasts by means of Generalized Estimated Equations explored linear and quadratic trends. The greater the number of PM subtypes suffered by children, the greater the adverse effects in psychopathology and functioning. When a child suffers four PM subtypes, the number of DSM disorders is, on average, twice as high compared with children who are suffering only one PM subtype. Linear trends were mainly found in internalizing problems. The importance of accurately assessing characteristics and severity of PM, and design efficient programs of prevention and treatment, is highlighted.
KeywordsAccumulative riskChildrenIntimate partner violencePsychological maltreatment
The American Professional Society on the Abuse of Children guidelines (APSAC 1995, page 2) state that psychological maltreatment (PM) involves “a repeated pattern of caregiver behavior or a serious incident which transmits to the child that he/she is worthless, flawed, unloved, unwanted, endangered, or only of value in meeting another’s needs”. PM has also been conceptualized and proposed as the core component of all forms of child abuse and neglect (Garbarino et al. 1986). This perspective is supported by the facts that: (a) PM is present in nearly all other acts of abuse and neglect since those acts have a negative psychological meaning (b) it appears to be the strongest predictor of the impact of child maltreatment; and (c) PM may have the longest-lasting and strongest negative effects on survivors of child abuse and neglect (Binggeli et al. 2001). PM is an extremely heterogeneous phenomenon, occurring in a wide variety of contexts. No two cases of PM will contain exactly the same elements, and many will be quite different from one another (Binggeli et al. 2001).
The APSAC (1995) proposes six PM subtypes: spurning, terrorizing, exploiting or corrupting, denying emotional responsiveness, isolation and neglect. Several studies have found that different forms of PM lead to different outcomes. For instance, spurning has been related to negative outcomes such as an angry and non-compliant behavior, negative emotion, lack of impulse control, hyperactivity and distractibility, difficulties in learning and solving problems, lack of enthusiasm and persistence, and low creativity (Egeland and Erikson 1987), and also predicted features of borderline personality disorder (Allen 2008). Denying emotional responsiveness has been considered the most devastating subtype, and can lead to anxious avoidant attachment, non-compliance, lack of impulse control, decline in competence from infancy through preschool years, low self-esteem, negative emotion, lack of enthusiasm, high dependence, self-abusive behavior, serious psychopathology (Egeland and Erikson 1987), depression and borderline personality disorder features (Allen 2008). Terrorizing significantly predicted somatic complains and anxiety in early adulthood (Allen 2008). A high rate of co-occurrence among various forms of maltreatment is observed (Dong et al. 2004; Knutson et al. 2005), which results in multiplicative effects between the number of different maltreatments and developmental consequences in the child.
Furthermore, experiences of PM are likely to increase the effect of other types of abuse or neglect, as proposed by Hart (1992). A revision of the second Canadian Incidence Study found that cases of co-occurrence of maltreatment present more negative effect, and more risk factors are observed (e.g., emotional harm, alcohol abuse, housing problems, chronicity, and referral to other services) when substantiated psychological maltreatment is also observed (Chamberland et al. 2012). This idea is in line with the classic model of additive main effects proposed by Rutter (1985, 2006); according to which, it is important to take into account the accumulated risk factors, given that the sum of the number of different types of maltreatment has poorer outcomes than each factor individually. Hahm et al. (2010) reported that experiencing different kinds of maltreatment during childhood led to an extensive range of risky behaviors in domains such as sexual risk behavior, [sexual transmission disorder (STD) diagnosis, regret getting into a sexual situation after alcohol use, multiple sex partner in the past 12 months, sex for money or sex before age 15], delinquency and suicidality. Boxer and Terranova (2008) found an association between the number of different types of maltreatment and high psychopathology. Furthermore, some combinations of PM subtypes have poorer outcomes than others; for instance, the combination of spurning and denying emotional responsiveness frequently appeared in the combinations of abuse that tended to produce the most devastating outcomes such as negative effects on feelings and perspectives about enjoyment of living, purpose in life, prospects for future life and chance of having a happy marriage (Loeber and Strouthamer-Loeber 1986; Ney et al. 1994).
The APSAC (1995) indicates that the number of subtypes suffered by the child is an indicator of the severity of the PM. Previous research has not focused on the multiplicative effects that different forms of PM might have in a risky population, such as that of children exposed to interpersonal violence (IPV). The study of this topic in this population is especially important, because PM of children and IPV are a closely related phenomena (Appel and Holden 1998; Osofsky 1999). IPV is a feature of many families in which emotional abuse co-occurs (Butler-Sloss 2001). The current trend is to consider exposure to IPV as a form of abuse, because witnessing an assault can terrify children and significantly alter their socialization (McGee and Wolfe 1991). As defined in the APSAC (1995), ‘witnessing IPV’ is a form of ‘Corrupting’. Given the high prevalence of children exposed to IPV, it is relevant to study the accumulative effect of suffering different types of maltreatment in this population.
Little research has been carried out on children exposed to IPV. Spurning has been the PM subtype with the greatest global effect on the children exposed to domestic violence, as it was significantly associated with internalizing and externalizing problems; and denying emotional responsiveness specifically increased the risk of internalizing psychopathology and impairment in the emotional area (de la Vega et al. 2011). The aim of this study is to assess the accumulative risk of the different types of PM in the mental health and functioning of children exposed to IPV. In accordance with the evidence from research carried out in different populations, the hypothesis of this work is that suffering a larger number of PM types will increase the negative consequences for the psychopathology and functioning of children living in an IPV context. Studying the relationships between the number of PM subtypes and the risk for psychopathology has relevant clinical applications for the prevention of appearance of different PM subtypes and the detection and treatment of populations at risk.
The present study is an original contribution to the field beyond previous knowledge. Unlike previous research, which is focused in the accumulation of risk factors or types of maltreatments, the present study focuses in the accumulation of a specific type of maltreatment: PM. It also contributes to the modeling of trends of the accumulation of PM (linear, quadratic or cubic) and to report on the different associations of each trend. These objectives pretend to fill the gap that exists in the study of a high risk population such as children exposed to IPV, which is poorly studied in regards to psychological maltreatment.
One hundred sixty-eight children and adolescents aged between 4 and 17 years whose mothers were battered women attending an outpatient Gender Violence Centre for Women in the Barcelona area participated in the study. The inclusion criterion was that both, mothers and children, were exposed to intimate partner violence, at least, in the last year. Of the 131 mothers invited to participate, 116 accepted. No statistical differences emerged for the comparison between participants and refusals by children’s sex (p = .944), age (p = .777), ethnicity (p = .070) or socioeconomic status (p = .133) (Hollingshead 1975). Mothers’ mean scores on the physical and non-physical scales in the Index of Spouse Abuse (Hudson and Rau 1981) were also similar for participant and non-participants (p = .115 and p = .817).
Number of PMs suffered
(N = 168)
N = 64
N = 54
N = 34
N = 16
Child’s age (years); Mean (SD)
Years of exposure to IPV; mean (SD)
Sex: male (%)
Other forms of maltreatment (%)a
Schedule for the Assessment of Intimate Partner Violence Exposure in Children
(SAIPVEC) (Developmental Psychopathology Epidemiological and Diagnostic Unit, unpublished). The different types of PM suffered by children were assessed using this instrument, created ad hoc for this study and based on Holden (2003). The SAIPVEC is a rating scale with interview format for collecting information. For this kind of assessment, the clinician working at the Gender Violence Center, asked mothers and rated the information. This measure has five sections: 1) child’s degree of exposure to IPV; 2) type and degree of aggression against the mother; 3) characteristics of the aggressor; 4) The role of the mother in the aggression and 5) the types of child maltreatment. The degree of exposure to IPV (section 1) was assesseddichotomously (yes/no) through 11 non-exclusive items (prenatal exposure, victim, participation, eyewitness, auditory/not visual witness, seeing the initial effects, experiencing the consequences, hearing about it, unaware of the situation, provided with some explanation about what happens). The type of aggression against the mother (section 2), was assessed dichotomously (yes/no) through 3 non-exclusive items (sexual, psychological, and physical), the degree (1 = slight, 2 = moderate, 3 = severe) of the exposure; and child age at the beginning of each kind of violence were assessed also on section 2. The characteristics of the aggressor (section 3), as reported by the mother, were assessed dichotomously (yes/no) (aggressive only in the family environment, antisocial, dysphoric/limit, substance abuse, impulsive, extremely jealous, inhibited/explosive, male chauvinist/dominant, psychopathic, other), The aggressor’s age and the relationship of the aggressor with the child were also assessed on section 3. The role of the mother in the aggression (0 = passive, 1 = Mother defends himself by attacking the aggressor, 2 = mother calms the aggressor, 3 = escape, 4 = assertive behavior), and the resolution (mother says sorry, aggressor is sorry, denial of conflict, minimization of conflict, acceptation, conflict goes on, submission/compliance, mother leaves family home) were assessed on section 4. The last section, which is the mainly used in this study, registers the types of child maltreatment with 8 dichotomous items (physical maltreatment, sexual maltreatment, physical neglect, ant the APSAC subtypes of PM: terrorizing, corrupting/bad socialization, spurning, lack of emotional responsiveness and isolation). Each type of child maltreatment has a definition that the clinician must follow in order to assess if this condition is present or absent: 1) Terrorizing, meaning behavior such as threatening to injure, kill or abandon the child or someone he/she cares about, or his/her pets; 2) Corrupting, that is, allowing or encouraging antisocial or inappropriate behavior, misogyny, violent behavior, verbal or physical aggression or substance abuse; 3) Spurning, including rejecting, scorning, ridiculing or criticizing the child; 4) Denying emotional responsiveness, ignoring the emotional needs of the child and his/her attempts to interact, or not showing positive emotions towards the child, being detached and uninvolved or being unable to display affection; 5) Isolation, described as unreasonably restricting contact with other children, not providing opportunities for socialization; and 6) Physical Neglect, which is lack of attention to the physical and educational needs of the child. Children from the sample could be simultaneously exposed to more than one subtype of PM. Physical Neglect was included in the analyses as a control variable, since it is not a form of PM. This instrument showed good internal consistency (Cronbach’s alpha equal to 0.70 in the section used for this study, in his case the last section of the instrument) and convergent validity with other standardized instruments (Ezpeleta et al. 2007).
Child Behavior Checklists
(CBCL/1½–5; CBCL/6–18) (Achenbach and Rescorla 2001) were used as dimensional measures of psychopathology answered by mothers. There are two different instruments: CBCL/1½–5, used for mothers of children aged from a year and a half to five years old and CBCL/6–18, used for mothers of children and adolescents aged 6 to 18 years old. They contain 100 and 113 items, respectively, with three response options (0 “Not True”, 1 “Somewhat or Sometimes True” and 2 “Very True or Often True”), and cover a wide range of emotional and behavioral problems in children and adolescents (such as Anxiety-Depression, Withdrawal, Somatic complains, Attention Problems and Aggressive Behaviour). These two instruments have eight scales in common: Anxiety- Depression, Withdrawal, Somatic Complains, Attention Problems, Aggressive Behavior, Internalizing Problems, Externalizing Problems and Total Score. There are 3 scales specific for the 6-18 version: Social Problems, Though Problems, Breaking Rules Behavior; and 2 scales are specific for the 1 ½-5 version: Emotional Problems and Sleep Problems. Standardized T-scores were analyzed in order to include in the analysis the common scales for young (4 and 5 years) and older (6 to 17) children. T-scores are chosen since they allow comparison by sex and age. Specific scales of the CBCL/6-18 version were analyzed separately (but not of the CBCL/1½–5, because of the small sample size for this range of ages). The reliability coefficients (Cronbachs’ alpha) of the CBCL scales in this sample are good, the lower alpha is 0.647 corresponding to the CBCL 1 ½–5 somatic complains scale 0.647 and the higher is 0.936 corresponding to the CBCL ½–5 total scale.
Diagnostic Interview for Children and Adolescents-IV
(DICA-IV) (Reich 2000). The DICA-IV, a semi-structured diagnostic interview that covers the most common DSM-IV (American Psychiatric Association 1994) diagnostic categories in children and adolescents, was used to assess child psychopathology. It was adapted and validated for the Spanish population with satisfactory psychometric properties (Ezpeleta et al. 1997a). Agreement obtained between interviewers was good to excellent (kappa values of between 0.65 and 1) (Ezpeleta et al. 1997b). There are three versions of this interview DICA-IV: One for children aged 8 to 12, one for adolescents aged 13 to 17, and one for parents (assessing children aged 8 to 17). Also the Diagnostic Interview for Children and Adolescents for Parents of Preschool and Young Children (DICA-PPYC) was answered by mothers of children aged 4 to 7. It was considered as a different instrument, rather than a version of the DICA, and has good statistical properties (Ezpeleta et al. 2011). The interview was carried out by trained psychologists. DSM diagnoses were derived, where applicable, by combining the information from the mothers and the children. Different psychologists interviewed the mother and the child separately. A symptom was considered to be present if any of the two informants reported it as present. For children aged 4 to 7, the information was obtained from the mothers only.
Children’s Global Assessment Scale
(CGAS; Shaffer et al. 1983). The CGAS, adapted for Spanish population (Ezpeleta et al. 1999), was used to measure functional impairment during the last year. Scale scores were in a range between 1 (maximum impairment) and 100 (best performance). The lowest score combining the information from the child and the mother was chosen.
The study was approved by the Ethics Review Board of the authors’ institution. Women attending to a center for battered women, whose children had been exposed to IPV at least during the last year, were informed and invited to participate in the research. Informed written consent to participate was obtained from the mother and oral consent from the children. Confidentiality was guaranteed.
The SAIPVEC was completed by a trained clinical psychologist at intake of women to the Gender Violence Centre. After an interview with the mother, the clinician handling her case rated whether or not the child met each type of abuse. The age-corresponding version of the DICA-IV interview, which lasts approximately 1 hour, was carried out by trained personnel, simultaneously and separately with mothers and children able. After the interviews, clinicians rated the degree of impairment (CGAS) associated with the psychological symptoms identified through a semi-structured diagnostic interview (DICA). Lastly, mothers completed the CBCL questionnaire version adequated for the children’s age. After their participation, women received an oral report about the mental state of their children. The return of information was done with care and sensitivity, supporting and guiding mothers on the best intervention options, in order to avoid distress in this sensitive group.
The statistical analysis was carried out with SPSS 19 for Windows. All PM subtypes proposed by APSAC were assessed, but one subtype was excluded from the analysis: isolation, because of the low prevalence in the sample (4.8 %). This research refers to nested structure data (some siblings had the same mother), but a low level of hierarchy was observed (58 % of mothers included only one child, 38 % two children and 4 % three children: mean number of children per family was 1.47), so that multi-level models were inadequate because they did not permit satisfactory adjustment (Hox 2002). To account for data dependency at the lower data level and prevent some estimation bias, the random factor “family” was included in multiple mixed models through Generalized Estimating Equations (GEE). Family was considered as a random factor in the GEE models because our study includes children pertaining to different families, and its aim was to generalize the results to all the families of the larger population of families exposed to domestic violence.
Linear, quadratic and cubic trends of number of maltreatments in the impact on psychopathology and functioning were explored though polynomial contrasts in GEE analysis. Number of PMs was the independent variable, and mean scores of the CBCL/1½–5 and CBCL/6–18, CGAS scores and diagnosis derived from the interview were the dependent variables. Linear trends rated the global increase–decrease of mean scores, quadratic trends explored whether the change comparing children with 1 and 2 PMs was statistically equal to the change comparing children with 2 and 3 PMs, and cubic trends did the same for changes between 2–3 PMs and 3–4 PMs.
All the analyses were adjusted for the covariates children’s sex and age, duration of exposure to IPV, presence of other forms of maltreatment (physical, sexual abuse or physical neglect), and type and severity of IPV suffered by the mother. Given the multiple comparisons in the study, the Bonferroni-Finner correction (Finner 1993) was used to control type-I error and to avoid spurious results. Bonferroni’s corrections were applied to adjust to .05 the level of significance for the set of comparisons.
Regarding the prevalence of PM in the sample, 38.1 % (N = 64) of the children were exposed to terrorizing, 27.4 % (N = 46) to spurning, and 35.7 % (N = 60) to denying emotional responsiveness. All the children in the sample were exposed to PM subtype corrupting because they all were exposed to IPV. In 38.1 %, corrupting was the only type of PM suffered. Those suffering only corrupting (38.1 %; N = 64) are the group of number of PM =1. A total of 32.1 % (N = 54) of the children suffered two types of PM (N = 54), 20.2 % (N = 34) suffered three types (group of number of PM = 3) and 9.5 % four types (terrorizing, spurning and denying emotional responsiveness, plus corrupting).
Dimensional Psychopathology (CBCL)
Linear, quadratic, and cubic trends of number of forms of psychological maltreatment and CBCL psychopathology
Number of PM suffered
N = 168
N = 64
N = 54
N = 34
N = 16
Only CBC6-18 scales
N = 134
N = 48
N = 44
N = 27
N = 15
Breaking Rules Behavior
Functional Impairment and DSM Variables
Linear, squared, and cubic trends of number of psychological maltreatments and functional impairment, and DSM symptoms and disorders
Means for different number of PM
N = 168
N = 64
N = 54
N = 34
N = 16
CGAS (mean score)
N of Externalized Symptoms
N of Internalized Symptoms
N of Total Symptoms
N of DSM Disorders
According to the hypothesis, a larger number of PMs suffered by a child exposed to IPV increases linearly the severity of psychopathology and functional impairment. These results are in agreement with those of previous models indicating that accumulated risk factors, such as the different forms of PM, lead to poorer outcomes (Boxer and Terranova 2008; Hahm et al. 2010; Hart 1992; Rutter 1985). These results are especially relevant given the high prevalence of children exposed to IPV (Queen Sofia Center for the Study of Violence 2007) and the multiple exposures to maltreatment that they suffer (Appel and Holden 1998; Cawson et al. 2000; Mitchell 2005; Mullender and Morley 1994; Osofsky 1999). These results indicate that each form of psychological maltreatment is relevant, and allow to identify children exposed to IPV as a target group for risk reduction.
Although the area of problems in which this increase mostly occurs is internalizing (anxiety-depression, withdrawal), accumulative risk of different forms of PM affects different areas of psychopathology, and also increases the likelihood of attention and social problems, as well as externalizing problems (aggressive behavior). Different forms of PM have an unspecific effect (Binggeli et al. 2001; Brassard and Donovan 2006; Iwaniec 2006), so that the prevention of PM in IPV must include components directed towards emotion, mood, and also behavior. Accumulative risk of PM also affected global functioning in daily life. As the number of PM increases, the level of functioning of children exposed to IPV is increasingly poorer. Briere (2004) also reported in an abused population that more severe and prolonged abuse increases subsequent mental health impairment. Glaser (2002) linked emotional abuse and neglect to impairment of the child’s development in all domains of functioning.
The fact that all the significant trends were linear implies that, as the child is exposed to an increasing number of maltreatments, child’s psychopathology also increases and child’s functioning decreases. Therefore, prevention programs for children exposed to domestic violence must be addressed to decrease the number of psychological maltreatments they suffer. This research has important clinical applications for the design of prevention and treatment programs, and also for the assessment of children living in circumstances of IPV. IPV and child PM are closely associated. This is the first approach to assessing the accumulative risk of PM in a Spanish population exposed to IPV. However some limitations must be considered on interpreting the results of the study. The size of some maltreatment groups (isolation) made it impossible to include this form of PM in the analysis. The voices of children under 8 years of age are missing from this study and the data from mothers speaking for these younger children under eight could be subjective and may not mirror what the children think or feel; however, interview schedules are not appropriate for children under age 8 (Ezpeleta 2001). Also, the results can be generalized only to children of mothers exposed to IPV seeking help.
In conclusion, the results show that the more abuse suffered, the greater the adverse effects. This paper highlights the relevance of taking into account the different forms of PM, given that each type is important in considerations of child psychopathology and functioning. Each PM that clinicians might prevent will mean a significantly smaller number of DSM symptoms and disorders. It is important to assess the presence of the different components of PM and to use instruments that enable us to do so. The careful assessment of PM in IPV will permit the accurate identification of exposed cases and the application of prevention programs that help to avoid the appearance of new PM subtypes.
This work was supported by grant SEJ2005-01786 and by a grant from the Research Training Program (Spanish Ministry of Education and Science). University Training Program (Formación del Profesorado Universitario, FPU) reference AP2007-01614.
And grant y the Science and Education Ministry of Spain for the proyect "Effects of Domestic Violence i Children" (Efectos de la exposición a violencia doméstica en niños) reference SEJ2005-01786.