Progress in Developing Responsive Parenting Programs for Child Welfare–Involved Infants

A significant proportion of young children will be maltreated and become involved with child welfare services. Of all American children investigated for child maltreatment, 20.6% are aged 0–2 years (Ringeisen, Casanueva, Smith, & Dolan, 2011. Typically, involvement in child welfare includes in-home services although placement into foster care is not unusual; 20% of children in foster care were younger than 3 years old in 2011 (U.S. Department of Health and Human Services [U.S. DHHS], 2012). Putnam-Hornstein and Needell (2011) recently estimated that 14% of children born in California in 2002 were reported for maltreatment during the next 5 years. Nationally, the highest rate of child maltreatment is observed during the first year of life (U.S. DHHS, 2010), with about 2.1% of children reported during that first year. The harm to children is, arguably, the greatest during that time (Kotch et al.,, 2008). Children and families who are not successfully served at the time of birth are highly vulnerable to longer term developmental problems and, not uncommonly, a later admission to foster care.

Despite two decades of trend data indicating the need to refocus our efforts to better serve the youngest children (Berrick, Needell, Barth, & Jonson-Reid, 1998; Wulczyn, Barth, Yuan, Jones-Harden, & Landsverk, 2008), our child welfare services have not kept pace with programs to address the needs of very young children and their caregivers. A recent national study by Horwitz and colleagues (2012) showed that only 19.2% of young children who were reported to child maltreatment agencies, and whose parents were investigated for maltreatment, subsequently received a mental health or parent training program. I believe that few readers would disagree with the conclusion of this study that “the lack of services received by these young, multi-challenged children is a services systems and social policy failure and should serve as a call to action” (p. 580).

One reason for service failure is the scarcity of available, feasible, evidence-based practices to help maltreating parents caring for infants and toddlers. The California Evidence-Based Practice Clearinghouse (CEBC) is arguably the best source of information about the effectiveness of practices involving maltreated children. CEBC considers research criteria for identifying programs at various levels of efficacy and at various levels of testing with child welfare populations. The CEBC identifies only one program, for 3- to 6-year-old children, Parent Child Interaction Therapy (PCIT), that receives its highest rating of “well-supported by research” (http://www.cebc4cw.org/topic/infant-and-toddler-mental-health-0-3/, retrieved September 3, 2012). Only one treatment program, Child Parent Psychotherapy, that addresses children younger than 3 is listed in the next highest category “supported by research evidence.” This program has recommended service delivery parameters of weekly meetings for approximately 52 weeks—hardly a good fit with the much shorter period of services typical for child welfare–involved families.

In the parent training area of the CEBC, a few more programs provide evidence of effectiveness with young children and parents involved with child welfare services. There are just two programs with a rating of well-supported by research for children 0–2 years of age: Triple P, Positive Parenting Program; and Oregon Social Learning Center Parent Management Training: PMT-O™. Although neither program has examined effectiveness with the child welfare population, each has demonstrated benefits with similar populations (de Graaf, Speetjens, Smit, DeWolfff, & Tavecchio, 2008; Forgatch, Bullock, & Patterson, 2004; Phelan, Howe, Cashman, & Batchelor, 2012).

The recent publication of a randomized controlled trial of Safe Care that shows benefit for a sample primarily comprised of preschoolers (about 75%) suggests that this intervention has a very good fit with parents of young children at risk of maltreatment or who have experienced maltreatment (Chaffin, Hecht, Bard, Silovsky, & Beasley, 2012). In this study, the SafeCare group had a significantly lower 6-year recidivism rate than the 45% rate observed in the in-home services as usual condition. SafeCare is not currently rated by CEBC at the highest level of evidence because there are not yet two randomized clinical trials, a common criterion for establishment of evidence-based practices (Chambless & Ollendick, 2001).

Other promising interventions are being tested with this early childhood population. A recent paper by Bernard and colleagues (2012) on their Attachment and Bio-Behavioral Catch-up (ABC) program found benefit for mother–child dyads. This 10-session intervention employs videotaping and feedback to help focus on specific behavioral targets. It aims to increase “nurturing care” and secure attachment, immediately after the intervention, for families at risk of maltreatment. Children in the ABC program had lower rates of disorganized attachment (32%) and higher rates of secure attachment (52%) when compared to the control intervention (57% and 33%, respectively). The authors conclude that “These findings are exciting given the nature of the sample (i.e., high risk for maltreatment and neglect) and given the mixed results of previous studies that have assessed disorganized attachment as an outcome of intervention” (Bakermans-Kranenburg, Van Ijzendoorn, & Juffer, 2005, p. 633).

Considered within the context of the need for better services for infants and young children and their caregivers in the child welfare system and the small number of evidenced-based programs for this population, the study by Spieker, Oxford, Kelly, Nelson, and Fleming (2012) offers significant possibilities. This anticipation is heightened by the positioning of Promoting First Relationships (PFR), which is disseminated by NCAST (Nursing Child Assessment Satellite Training), the center developed by Kathryn Barnard. This legacy surely creates the expectation that the intervention will have noteworthy capacity to help parents boost their engagement in responsive parent–child interactions. The curriculum (Kelly, Zuckerman, Sandoval, & Buehlman, 2003) also has a long history and has been available for dissemination and training, through NCAST.

The Spieker et al. (2012) PFR study is highly ambitious which is unusual in child welfare related research. This randomized clinical trial endeavored to improve the responsiveness of parent–child interactions and also attempted to show measurable change in secure attachment. As it turns out, changing infant attachment security was too difficult a challenge. Lasting effects on security were not observable. There was no change from baseline to postintervention. Nevertheless, there were certainly some important benefits demonstrated for caregivers. After participating in the PRF program, caregivers scored significantly higher on sensitivity to toddlers and understanding of toddlers, than did those in the control group. These effects remained at the 6-month follow-up.

The lack of findings for an increase in secure attachment can be considered from at least three possible perspectives. First, taking a narrow scientific view, this study can be summarized as having failed to achieve a central purpose and, therefore, there is little justification for replication or dissemination. Second, we can adopt a less narrow view of the disappointing findings on infant security by considering the many difficulties in measuring attachment outside of controlled conditions in the laboratory. In essence, we could speculate that the PFR intervention might have improved infant security of attachment, but such improvement was difficult to measure. We can also consider the evidence that the measurement of attachment is affected by a variety of transitional forces that arise when environments change (Eagle, 2006) and, therefore, measuring attachment for the highly mobile child welfare population may be especially unrealistic.

A third perspective suggests that we can take this opportunity to reconsider the use of attachment security as an outcome indicator for this or for future studies. Demonstrating the benefits of the PRF program to improve responsive parenting does not require evidence of improved attachment security. To the contrary, several investigations of interventions that make no assumptions about attachment as fundamental to improved development have shown improvements in child behavior, measures of child stress, and caregiver stress (e.g., Fisher, Gunnar, Chamberlain, & Reid, 2000).

A fundamental principle of science is parsimony. In this vein, the Spieker et al. (2012) study reminds us that the advancement of evidence-based parenting programs need not be routed through the thicket of contradictory previous studies about what measures of attachment are the best or what they really mean. The opportunity for greater parsimony in the work of Spieker et al. (2012) is raised by research done by Pears, Bruce, Fisher, and Kim (2010), also published in this Journal, on indiscriminate friendliness. This research shows a stronger relationship between indiscriminate friendliness and self-regulation than between indiscriminate friendliness and attachment security. Both the Spieker et al. (2012) and the Pears et al. studies demonstrate that measures of attachment are not necessary to understanding children’s behavior or showing improved parenting. The importance that the investigators give to making sure that their research is framed as an attachment intervention, even when the basic paradigm is not confirmed by the findings, is puzzling. In this case, more sensitive parenting is not strongly associated with greater attachment security (r = .26, p < .05). The authors attribute this to poor measurement because it is difficult to detect attachment security in applied or “real-world” settings. The psychometric literature on measuring attachment certainly confirms their conclusion (Lim et al., 2010). In all, the work to develop improved parenting programs would be more straightforward, and probably more successful, if the original conceptualization of the NCAST was left to speak for itself.

The lack of parsimony in the conceptual foundation for the PFR intervention and the related measurement problems may also limit its dissemination throughout child welfare agencies. Child welfare agencies have not, historically, used manualized interventions (Barth et al., 2005) but they are showing more interest in doing so. Yet, my experience tells me that it would be a difficult task to convince child welfare agencies or parent training providers to embrace a new training program that is marketed “to promote secure attachments” but shows no lasting impact on attachment. Even though there is much confusion among child welfare personnel about what is an evidence-based practice, the findings from this study hardly seem like a strong basis for selecting and implementing PFR. Many child welfare practices do not have research evidence of effectiveness but are used because they appear, on the face of it, to be valuable. The results of this trial, although providing a sound basis for future research, seem unlikely to meet that test in many child welfare agencies.

A related limitation that would likely impede implementation is that the attachment-related outcomes of the PFR program cannot be readily measured by parent trainers or child welfare workers. Although child welfare agencies have admittedly relied on minimal evidence of success such as reports of whether parents attended parent training, I believe that the growing uptake of PCIT by child welfare systems is, at least in part, due to the documentation of improved parenting competence that is generated by this evidence-based program. Using measures of change in outcomes as a strategy for engaging and informing parents during treatment is, in any event, part of the next generation of approaches that are likely to become more standard (Kelley & Bickman, 2009).

Furthermore, implementation of the PFR intervention by child welfare services is likely to be challenging because it is not sufficiently clear how it addresses the major goals of safety, permanency, and well-being. In the Spieker et al. (2012) study, the comparison condition leaders all had bachelor’s degrees and the PFR leaders were all mental health professionals. This is potentially a significant problem as an alternative explanation for why the intervention group improved more than the control group in parenting behaviors. The study is also unclear about which type of caregivers had increased parental sensitivity and I believe this issue would matter to professionals in child welfare making decisions about adopting the PFR intervention. Indicating that there were no differences between birth, kin, and foster parents on improvement does not demonstrate that there would have been main effects of the same size if only birth parents were studied.

In short, how would a child welfare agency manager know from the findings, whether PFR is likely to be used successfully in foster parent training to improve developmental well-being or in conventional parent training to increase safety children remaining at home. Even though Spieker et al. (2012) provided evidence that the baseline scores on a range of measures were not significantly different by caregiver types, child welfare agencies are likely to be less than confident that improving parental sensitivity will increase child safety, in the near term. Given this ambiguity, I would expect child welfare agencies to be more attracted to SafeCare^®, with its stronger track record on improving safety for young children and their caregivers, than to PFR.

Despite these concerns, the Spieker et al. (2012) research does provide promising evidence for changing parenting behavior. Important findings from this study, when considered with similar results from Bernard et al. (2012), indicate that using videotape feedback over a brief time frame (10 weeks in both cases) is a powerful tool to improve responsive parenting. The PFR program, like ABC, is conducted in caregiver’s homes, one family at a time. This is a logical setting to promote the use of new skills although it is not the most convenient setting for service delivery systems. Delivering parent training in groups is also far more typical for child welfare agencies, and probably less costly, and some interventions that have always been done individually (e.g., PCIT) are now being tested in groups.

Overall, I believe that Spieker and colleagues (2012) are to be commended for their efforts to develop an intervention that addresses the pressing need for programs to serve the youngest and arguably most vulnerable children in the child welfare system. As work on the PFR and hopefully other new interventions for this age group move forward, it is important that they focus on the extent to which such programs are effective with subpopulations of child welfare–involved families. Equally important for the wider adoption of such programs by child welfare services will be developing interventions that are clearly designed to include improvements in child safety.