Relational Interventions for Maltreated Children

Abstract

Child maltreatment may be best characterized as a pathogenic relational experience which primarily occurs in the mother-child relationship. As such, enhancing the mother-child relationship is the key process that should be targeted in intervention approaches for child maltreatment. Two salient and modifiable components of the mother-child relationship are highlighted: maternal sensitivity and attachment organization. It is argued, from a developmental psychopathology perspective, why focusing on these issues hold the most promise for interrupting negative developmental cascades and promoting resilience among maltreated children. Utilization of a tiered approach to delivering increasingly intensive relational interventions is recommended as are future directions for translational research and dissemination.

Child maltreatment has significant and enduring damaging consequences for individuals in multiple developmental domains, in addition to enormous costs for society. In 2013, approximately 3.9 million children were subjects of maltreatment reports to child welfare agencies; nearly 700,000 cases were substantiated leading to a national victimization rate of 9 per 1,000 children (US DHHS, 2015). Critically, more than 90% of maltreated children are victimized by one or both parents (US DHHS, 2015); as such, child maltreatment may be best characterized as a pathogenic relational experience which primarily occurs in the parent-child relationship. That is, in the context of child abuse and neglect, the maladaptive nature of the parent-child relationship is largely capable of increasing risk for psychopathology and other negative sequelae for maltreated children. In contrast, positive aspects of the parent-child relationship, such as positive internal representations of parents, have a significant buffering effect in the association between maltreatment and child emotional and behavioral adjustment (i.e., Toth, Cicchetti & Kim, 2002). Addressing the parent-child relationship, therefore, emerges as a central focus of interventions for child maltreatment, especially considering that the majority of child victims ultimately remain in the custody of their parents (US DHHS, 2015). To date, intervention efforts focus primarily on enhancing the mother-child relationship because mothers are most likely to be identified as perpetrators of maltreatment, often in single-parent roles, and because when fathers are present they tend to have a peripheral role in caregiving (Howes & Cicchetti, 1993; US DHHS 2015).

In this paper, I focus on the mother-child relationship as the key process that developmental scientists should enhance in intervention approaches for child maltreatment. More specifically, maternal sensitivity and attachment organization are highlighted as central components of the parent-child relationship that can be improved by relationally based interventions for maltreated children. From a developmental psychopathology perspective, the reasons that these relational issues hold the most promise for interrupting negative developmental cascades and promoting resilience among maltreated children are explained. Interventions that have been successful at addressing these mechanisms are reviewed as examples and are critically evaluated in terms of sustained improvements and breadth of outcomes, child age range targeted, levels of support for the evidence-based status, and issues related to cost effectiveness and dissemination. Given the existing evidence, a tiered delivery approach is encouraged wherein brief relational models focused on sensitivity are provided as the first line of treatment and may be followed by increasingly intensive relational treatments as needed.

Positive parenting is central for scaffolding young children’s cognitive development, and in providing external emotional and physiological regulation for the child (e.g., Bowlby, 1969; Spangler et al., 1994). Child maltreatment represents the most extreme case of failure to provide the parenting experiences necessary to facilitate competent development. Thus beyond specific incidents of abuse and/or neglect, maltreating families are often characterized by a pervasive pathogenic relational environment that undermines children’s ability to successfully resolve stage-salient developmental tasks. A developmental psychopathology perspective argues that development is hierarchically organized and the achievement of each developmental task is critical for facilitating the ability to resolve future stage-salient issues. Unsuccessful resolution of early stage-salient tasks, however, may result in disruptions in subsequent functioning throughout the lifespan (Cicchetti & Valentino, 2006). As such, interventions for child maltreatment should seek to not only decrease future incidents of maltreatment, but also to enhance components of the mother-child relationship that will promote the resolution of early stage-salient developmental tasks and positive developmental trajectories among maltreated children.

Maternal Sensitivity & Attachment Organization

The formation of attachment relationships is a central developmental task during the first two years of life. Parent–child interactions throughout this period shape children’s emerging affect and physiologic regulation abilities, as well as their biobehavioral patterns of response (Gunnar and Vazquez, 2006), from which the capacity for attachment emerges. Infants rely on their primary caregivers, typically mothers, to derive a sense of security and they use this relationship as a base from which to explore the environment (Sroufe, 1979). Consistent caregiver sensitivity and emotional support are essential for the development of secure attachment representations because infants rely on these regularities throughout early childhood to develop internal models of themselves and others, and to create expectations for the future (Bowlby, 1973). In the absence of these critical parenting behaviors, the development of a secure attachment relationship is often impeded, and attachment insecurity results. Moreover, the inconsistent, harsh and sometimes frightening behaviors of maltreating mothers often leave their infants without an organized behavioral response to stress, known as Disorganized attachment (Main & Soloman, 1990). Indeed among maltreating families, rates of attachment insecurity, and disorganized attachment in particular, have been reported as high as 80–90% of mother-child dyads (Barnett, Ganiban, & Cicchetti 1999; Cicchetti, Rogosch, & Toth, 2006).

Attachment security has a critical role in supporting subsequent positive development including affect regulation, coping, emotional and behavioral functioning, peer relationships, and physiological regulation (see Shore, 2001; Toth et al., 2013 for review). In contrast, longitudinal evidence demonstrates that the developmental outcomes associated with disorganized attachment, including dissociative and externalizing symptoms, are often even more pathological and maladaptive than are those associated with other forms of attachment insecurity (i.e., Carlson, 1998; Lyons-Ruth, Easterbrooks, and Cibelli, 1997). As such, intervening to promote more positive early mother-child interactions among maltreating families is the goal of relational intervention approaches, and is the process through which negative developmental sequelae associated with disruptions in attachment organization may be averted.

Relational interventions targeting maternal sensitivity and/or attachment organization among maltreating families may generally be divided into shorter (5–16 weeks) or longer term (20 weeks-1 year) programs, all with weekly visits and involving both mother and child. In the brief models, the treatment typically focuses on enhancing caregiver sensitivity and modifying behavioral interactions with the child (Dozier et al., 2006; Moss et al., 2011). Longer term approaches often target improving attachment security more broadly, and also include psychotherapy to facilitate mothers’ reinterpretation of their own childhood experiences and development of more positive representations of their children, or training on specific behavioral management techniques (Cicchetti et al., 2006; Chaffin et al., 2004). Importantly, a meta-analysis of over 80 studies demonstrated that interventions are effective in improving insensitive parenting (d = .33; in general, values of .2, .5, and .8 represent small, medium, and large effect sizes respectively) and attachment insecurity (d = .20; Bakermans-Kranenberg, van Ijzendoorn, & Juffer, 2003) with small to medium effect sizes; although included studies were not exclusive to maltreating families, these important parenting mechanisms appear to be modifiable. Intervention components that were associated with the largest effects included length, where briefer treatments focused on sensitivity (< 16 sessions) were more effective than long term models. Both approaches, including their advantages and disadvantages for work with maltreating families are reviewed in the following sections.

Brief Relational Models

As noted, sensitive caregiving is central for the development of secure attachment relationships and for providing an external scaffold for children’s developing stress regulatory system (Bowlby, 1969, Spangler et al., 1994). Thus, sensitivity may be considered a more proximal parenting mechanism to enhance during relational interventions relative to the more distal mechanism of attachment organization. There are a number of brief relational models that have been successful in improving maternal sensitivity among maltreating families including Attachment and Biobehavioral Catch-Up (ABC; Dozier et al., 2006), Promoting First Relationships (PFR; Kelly, Zuckerman, Sandoval, & Buehlman, 2008) and a brief attachment-based intervention (Moss et al., 2011). These models are short in duration (8–10 sessions), and are typically provided in home. Clinicians in all models focus on assisting mothers to respond sensitively and appropriately to child affective cues and provide frequent, positive, in the moment feedback to highlight parents’ strengths. Video feedback is also used to highlight positive changes in parents’ and children’s behaviors.

To date, ABC is the brief relational intervention for child maltreatment with the most empirical support. ABC was evaluated in an randomized controlled trial (RCT) which randomized mothers and children (ages 6 months to 2 years) at high risk for neglect into ABC to or to developmental education intervention. Families were ethnically diverse including 61% African American, 15% Hispanic, 15% White/non Hispanic and 9% biracial families. Both treatments were 10 sessions, provided weekly for 1 hour each in the home. Results indicated that children in the ABC treatment had lower rates of disorganized attachment (d = .52) and higher rates of secure attachment (d = .38) posttreatment relative to children in the other intervention despite having comparable rates of attachment insecurity pretreatment (Bernard et al., 2012). Furthermore, children who received the ABC intervention had a more normative pattern of diurnal cortisol regulation than children who received the control intervention (Bernard, Dozier, Bick & Gordon, 2015). These physiological effects were sustained three years following treatment (Bernard, Hostinar, & Dozier, 2015). Similar positive effects have been noted in samples of foster parents receiving ABC, including increased maternal sensitivity (Bick & Dozier, 2013).

Moss’ brief intervention was evaluated in an RCT where mother-child dyads with children 1–5 years of age from maltreating families were randomized into an 8-week intervention or a control condition consisting of treatment as usual by the child welfare agency (Moss et al., 2011). Parents in the intervention group demonstrated significantly greater sensitivity than did the parents in the control group post-treatment (d = .47), despite being comparable at baseline. The intervention also facilitated attachment security, as a greater proportion of insecure children in the intervention group became secure (effect size r = .36, small effect), and a greater proportion of children with disorganized attachment in the intervention group became organized (r = .37) in comparison with the control group.

Finally, a randomized trial of PFR with children between 10–24 months and their caregivers (including foster parents, biological parents and/or other relative caregivers), demonstrated that caregivers in the PFR condition had greater improvements in sensitivity (d = .41) and understanding of children’s emotional needs (d = .36) postintervention than did those in the educational comparison condition (Spieker et al., 2012). Although successful in improving sensitivity, postintervention and 6-month follow up revealed no changes in attachment security.

Of these brief models, only ABC has been disseminated and evaluated in effectiveness studies. For example, families who participated in ABC had lower child abuse potential scores (parent reported), less parenting stress and children had lower internalizing and externalizing behaviors compared to families in a wait-list control condition (Sprang, 2009). Results from an RCT with substance abusing mothers found that mothers who received ABC demonstrated greater sensitivity than mothers in a control condition of the same length that inquired about mother-child wellbeing and provided age-appropriate books (d = .67; Berlin, Shanahan, & Appleyard Carmody, 2014).

Overall, brief relational models are effective in improving maternal sensitivity among maltreating families following a short duration of treatment, and are associated with additional positive effects including child physiological regulation and emotional adjustment. Moreover, these models can be provided by paraprofessionals, which may facilitate widespread dissemination and implementation. ABC, for example, has no minimum educational level requirement for parent trainers. Training costs are approximately $5000 per trainee, and include a 2–3 day workshop. This is followed by a year of twice-weekly supervision for fidelity checks and certification. A disadvantage to this approach, however, is the lack of outcome data beyond 1 year (aside from maintenance of physiological effects). Also, despite reductions in (self-reported) parent child abuse potential, there are no data regarding whether these models reduce reinvolvement with the child welfare system.

Intensive Relational Models

As opposed to focusing narrowly on maternal sensitivity, alternate relational intervention models seek to improve the parent-child relationship by focusing more broadly on attachment security. Examples of more intensive relational interventions include Child-Parent Psychotherapy (CPP; Lieberman & Van Horn, 2008) as well as Parent-Child Interaction Therapy (PCIT; Chaffin, Silovsky, dunderburk, Valle, Brestan, & Balachova, 2004). Although both models focus on improving the quality of parent-child relationship and enhancing mothers’ ability to engage in more positive interactions with the child, these models differ in their therapeutic approach, as well as the age ranges targeted.

The primary goal of CPP is to strengthen the parent-child relationship as a vehicle through which children’s attachment organization, sense of safety, and appropriate affect may be restored. It is used with children ages 0–5 and is typically implemented in weekly in-home sessions for approximately one year. CPP is a nondirective model and focuses on the development of a strong therapeutic relationship between mother and therapist, through which the mother may then reflect on past relationships, differentiate current and past relationships, and develop more positive internal representations of herself and her child (Lieberman & Van Horn, 2008). The mother and child participate together in sessions and play is often used to facilitate parent-child interactions. Intervention targets include identifying maladaptive internal representations as well as behaviors that interfere with the parent-child relationship and encouraging more developmentally appropriate mother-child interactions.

CPP has been evaluated in a number of RCTs with maltreated children. For example, Cicchetti, Rogosch, and Toth (2006) randomized 12-month old infants from maltreating families and their mothers to one of three conditions: CPP, a psychoeducational parenting intervention (PPI), or a community standard comparison group. A fourth group of dyads from nonmaltreating families were included as an additional comparison. Families were predominately low income, and 74.6% were minorities. Mothers and infants in the intervention conditions received weekly therapy sessions for one year. Prior to treatment, attachment insecurity was higher among infants from maltreating families than from the nonmaltreating families. At post-intervention children in the CPP and PPI groups showed significant increases in secure attachment, and decreases in disorganized attachment (d = .84–1.13) whereas attachment in infants in the community standard and nonmaltreatment comparison groups did not improve. At one-year post intervention, children in the CPP condition had higher rates of secure attachment (d = .33) and lower rates of disorganized attachment (d = .34) than children in the PPI condition (Pickreign Stronach, Toth, Rogosch, & Cicchetti, 2013). Thus, only CPP was efficacious in maintaining attachment security over time. Additional outcomes associated with CPP include physiological regulation (Cicchetti et al. ,2011), decreased behavior problems (d = .24) and symptoms of posttraumatic stress disorder (d = .63; Lieberman, Van Horn, & Ghosh Ippen, 2005) as well as decreased maternal distress (d = .38; Lieberman, Ippen, & Van Horn, 2006).

Parent-child interaction therapy (PCIT) is based in both attachment theory and social learning theory (Eyberg & Robinson, 1982). Although originally developed for children with disruptive behavior disorders, it has been applied to improving parent-child interactions among maltreating families with children 4–12 years of age (Chaffin et al., 2004). PCIT is didactic and teaches parents to increase positive interactions and to use specific behavior management skills to decrease negative behaviors. Sessions also involve the use of bug-in-the-ear technology so clinicians may coach parents to implement the skills during sessions. Rather than assessing attachment security, positive and negative behaviors are tracked and charted each session to provide the parent with immediate feedback regarding progress.

Chaffin and colleagues (2004) evaluated the efficacy of PCIT in preventing rereports of physical abuse among abusive parents of 4–12 year olds. Families were randomized to receive PCIT, PCIT plus individualized enhanced services, or a standard community-based parenting group. The interventions were implemented across 22–24 weeks of sessions. Results revealed that those who received PCIT were less likely to be re-reported for abuse approximately 2 years later than those in the standard community comparison group, and decreased negative parent-child interaction behaviors mediated that effect.

Overall, long term relational models have demonstrated success in promoting secure attachment among several other positive outcomes and they have been evaluated among diverse samples representative of those involved in the child welfare system. Post-treatment evaluations (limited to 1–2 years) have revealed maintenance of effects, including decreased re-involvement with the child welfare system. Nonetheless, the costs of disseminating and implementing these long-term models are high. Masters level education is a requirement for clinicians in both CPP and PCIT. Training in both models involves 3–5 days of workshops, but then includes up to 18 months of consultation and supervision support. As such, these models present some practical challenges for widespread dissemination.

Recommendations

The mother-child relationship has been highlighted as the key mechanism to be modified in relational interventions for child maltreatment. Intervening to enhance components of the mother-child relationship, including maternal sensitivity and attachment security, is critical for interrupting negative developmental cascades and for promoting resilience among maltreated children (Toth et al. 2013). Empirical research randomizing maltreating families with similar levels of maladaptation into shorter vs. longer term treatment options is needed before firm conclusions may be made regarding length of treatment. In the absence of such data, and in light of the limitations on resources for mental health services in the child welfare system, a tiered service delivery model may be the most prudent option. In a tiered model, brief relational treatments that target maternal sensitivity could be provided first, and only followed by more intensive services in cases that do not sufficiently respond to the initial treatment. One such model of tiered service delivery, Building Healthy Children, is currently being implemented at Mt. Hope Family Center. Young at-risk mothers are recruited through pediatric clinics and offered support to address basic needs such as food, housing, employment. Depending on need, tiered services are available, ranging from didactic parenting support to treatment for depression and CPP (Paradis, Sandler, Manly, & Valentine, 2013). Because brief treatments require fewer resources in terms of provider training, educational background and time relative to intensive models, this flexible, tiered approach would allow for more families to receive treatment, and could facilitate more targeted identification of those who may require more intensive services. Additional recommendations are provided below:

First, an important process component of relational interventions for child maltreatment is the provision of support for the maltreating parent, including both emotional support and basic needs assistance (Thompson, 2015). In several interventions discussed, the therapist aims to identify and praise mothers for specific moments of positive mother-child interactions rather than providing critical feedback (i.e., Bernard et al., 2012; Chaffin et al., 2004; Moss et al., 2011). Such support is essential for building trust, rapport and engagement among maltreating mothers who may be especially sensitive to negative feedback about their parenting and competence. It will be important for future research to determine whether treatments need to be delivered in home or one-on one to provide this support or if group based interventions may be successful. Group relational interventions, such as Circle of Security (Hoffman, Marvin, Cooper & Powell, 2008), have demonstrated improvements in attachment security among at-risk families, but have not yet been examined in RCTs. Another group relational intervention, Relational Psychotherapy Mother’s Group (RPMG; Luthar & Suchman, 2000; Luthar et al., 2007), has been evaluated in two RCTs. Although attachment was not assessed as a post-treatment outcome, RPMG was associated with other improvements including reduced risk for child maltreatment and improved child socio-emotional functioning, and thus may serve as a cost-effective group option.

Additionally, maltreating families are often facing multiple problems (i.e. poverty, housing instability, unemployment, etc.), and those struggling most with basic needs are often the most difficult to engage and retain in treatment (Thompson, 2015). Providing assistance with basic needs alone is insufficient to address child maltreatment. Integrating this support, however, into a network of tiered relational services may be most ideal for improving the functioning of maltreating families and flexibly addressing their needs. Extended basic needs assistance may be especially critical in supporting families in the months following brief relational treatments.

Second, brief relational interventions for child maltreatment that are appropriate for delivery beyond infancy and toddlerhood are needed (Toth et al., 2013). During the preschool years sensitive parenting typically shifts from physical behavior to include more verbal displays. In particular, parents’ abilities to co-construct elaborative and emotionally supportive narratives about children’s emotional experiences become critical in shaping children’s representational models of self and others and in supporting their development (Fivush, Haden & Reese, 2006; Thompson, 2006). Deficiencies in this behavior have been documented among maltreating mothers and serve as a mechanism through which maltreatment is associated with cognitive, emotional, and physiological dysfunction among maltreated preschool aged children (Valentino et al., 2015). Reminiscing and Emotion Training (RET; Valentino et al., 2013) was developed for maltreated preschool-aged children and their mothers, and focuses on improving maternal emotionally supportive communication about every day past events as the primary mechanism of change. A pilot study randomizing dyads to receive RET or a wait-list control condition indicated that maltreating parents who received RET demonstrated greater elaborative and emotion-rich reminiscing at post-test than did parents in the control condition (d = .64–.97, respectively), and their children showed enhanced memory and ability to discuss emotions (d = .94, 1.4). Support for RET is preliminary and pending the results of an ongoing RCT. Nonetheless, it serves as an example of a relational intervention with potential to be developmentally appropriate for delivery during the preschool years. RPMG (Luthar & Suchman, 2000; Luthar et al., 2007), is another brief relational intervention that is appropriate beyond early childhood and has been used with children up to age 18.

Toward the goal of disseminating and implementing these models, it is important for providers to track progress by evaluating the mechanisms being targeted in treatment and their associated outcomes. In brief relational models targeting maternal sensitivity, clinicians should track the frequency of appropriate, sensitive parental responses to child behavior across sessions. Additionally, a brief 25 card maternal sensitivity Q-sort (Tarabulsy et al., 2009) could be implemented immediately following observation to provide an overall rating of sensitive parenting, as has been done in prior effectiveness trials (Berlin et al., 2014). Tracking improvement in attachment security is difficult, however successful implementation of CPP in the community has involved brief semi-structured observations of mother-child play (i.e., the Crowell (Crowell & Feldman, 1988)) to assess attachment security pre- and post-treatment along with parent reports of parenting stress and child emotional and behavioral functioning to track treatment progress (Barnett et al., 2014). Alternately, as is done in PCIT, the frequency of goal parenting behaviors (sensitivity, developmentally appropriate interactions, etc.) or child behavior (proximity, contact seeking and maintenance, avoidance) could be tracked session by session to monitor progress. Unfortunately continuous report scales of child attachment security do not currently exist.

Apart from measurement of sensitivity or attachment, assessing improvements in child trauma symptoms, emotion regulation, internalizing and externalizing symptoms can also indicate the extent to which progress is being made. Additionally, new incidents of maltreatment are important progress variables to be tracked. Maltreatment recurrence information (i.e., number and content of new reports, investigations, substantiations, etc.) can be obtained via official state child welfare records with maternal consent, and subsequently evaluated with operational definitions for child maltreatment such as the Maltreatment Classification System (Barnett, Manly, & Cicchetti, 1993). Clear operational criteria for maltreatment subtypes and severity, independent of child welfare designations, are necessary because the determinations included in official child welfare records (i.e., substantiated physical abuse) are subject to evidentiary standards that vary across locales (Manly, 2005). Tracking maltreatment occurrence can also be done via maternal self-report; however, this approach to the assessment of maltreatment has high potential for nondisclosure because reporting new incidents of maltreatment could result in prosecution. Alternately, one could assess the potential for future maltreatment with self-report measures such as the Child Abuse Potential Inventory (CAPI; Milner, 1990) with the goal of documenting significant decreases in child abuse potential across treatment. Finally, clinicians working with maltreating families are always monitoring for suspected maltreatment and are mandated to report those concerns; however a lack of new reports during treatment is not a sufficient marker of treatment effectiveness.

Biobehavioral regulation is an additional mechanism through which child maltreatment leads to dysfunction, and the ability to restore children’s regulatory capacity is an important measure of effective treatments (Toth et al., 2013). Several relational interventions for maltreated children have promoted more normative child cortisol regulation (Bernard et al., 2015; Cicchetti et al., 2011). It will be important for future research to delineate the specific mechanisms and direction of associations among maternal sensitivity, attachment and physiological regulation, and to determine whether biological change produces the greatest sustained benefits for child well-being. Considering first, our evolving understanding that interactions between mothers and children are complex exchanges of both physiology and behavior, and second, that the harms of child maltreatment derive, in part, from how chronic stress biologically undermines self-regulation in both the parent and child (Repetti, Robles, & Reynolds, 2011), future research should seek to include biological assessments of both the mother and child into RCT designs. Examining intervention outcomes across different levels of analysis will allow for explication of mechanisms of change, including for whom positive change is most likely (Cicchetti & Dawson, 2002; Toth et al., 2013). Measurement of diurnal cortisol, for example, provides critical information about individual physiological regulation that cannot be readily observed or self-reported. Such data may inform treatment selection or may even be utilized as a marker of treatment progress in the future; however, at the present time, measurement of cortisol level and/or diurnal variation is not diagnostic. Further, it is important to note that biological data should not be used to withhold treatment. Instead biomarker data including genotyping may become useful for identifying who will be more/less responsive to interventions and for guiding decisions such as the immediate initiation of intensive treatment, rather than the initial provision of a brief model that is less likely to be effective for that individual. Given the associated costs and the nascent state of our knowledge regarding how this information may inform individual, actionable treatment recommendations, inclusion of biomarkers should not be brought to scale as interventions are implemented outside the lab at this time.

Finally, it should be noted that nearly all of the relational interventions reviewed were evaluated with mothers. With the exception PCIT, which has been evaluated in a RCT with 35% male caregivers (Chaffin et al., 2004), fathers represented less than 5% of participating caregivers in all other intervention studies reviewed. Recent research suggests that there are important similarities and important differences in associations among parent sensitivity, parent-child attachment and child behavior between mothers and fathers (Braungart-Rieker, Zentall, Lichenbrock, Ekas, Oshio, & Planalp, 2014). Thus, because the vast majority of the work on relational interventions has been developed and implemented with mothers, we cannot assume the same strategies will work as well with fathers.

In conclusion, relational intervention approaches to child maltreatment that focus on improving the mother-child relationship hold the most promise for improving the developmental trajectories of maltreated children. To serve the greatest number of maltreating families with the limited funding available, a tiered service delivery model is recommended wherein families may first receive brief relational models that target maternal sensitivity and well as assistance with basic needs. Subsequently, additional intensive relational services could be offered on a more targeted basis. Ultimately, enhancement of maternal sensitivity and attachment security among maltreating families are the key processes upon which to focus our efforts to facilitate healthy development among maltreated children.