Home-Based Parent Child Therapy for Young Traumatized Children Living In Poverty: A Randomized Controlled Trial

Abstract

A randomized control trial was used to evaluate the effectiveness of a home-based, parent-and-child therapy program specifically developed for toddlers and preschoolers living in poverty with trauma symptoms. Sixty-four children 5-years of age and younger were referred to a community-based clinic for behavior problems and emotional difficulties. All children had experienced one or more potentially traumatic events and met the DSM-5’s criteria for Post-Traumatic Stress Disorder in Children Six Years of Age and Younger. All families received government assistance indicating that their income met the federal definition for poverty. Participants were randomly assigned to either immediate treatment or wait list control groups. Significant between-group differences on all post-treatment measures were found. After the waitlist group completed treatment, significant improvements for both groups were found on all measures at six-weeks follow-up. Outcomes included reductions in challenging behaviors and emotional symptoms of trauma, improved caregiver-child relationships, and increased caregiver adherence to treatment strategies. This study offers support for early intervention of children with trauma symptoms and identifies the clinical challenges and advantages of providing therapy services in a home setting for very young children in poverty.

Many children are adversely affected by traumatic experiences. In 2014, 700,358 children experienced substantiated abuse or neglect in the United States (U.S. Department of Health and Human Services 2016). It has been estimated that about 60% of children are victims of physical abuse, 5% of children are victims of sexual abuse; and 40% of children have witnessed domestic or community violence (Finkelhor et al. 2009). Unfortunately, even very young children are not protected from experiencing traumatic events. Rather, children under the age of five-years-old are one of the highest risk groups for experiencing traumatic incidents, including physical and sexual abuse, neglect, and witnessing violence (Lieberman et al. 2011). One report estimated that 26% of children in a healthy birth cohort would witness or experience a potentially traumatic event before the age of four-years-old (Briggs-Gowan et al. 2010). Complicating the picture further, young children from families exposed to urban poverty face a disproportionate risk of exposure to trauma due to factors associated with poverty such a living in unsafe neighborhoods, racial discrimination, and the challenges associated with daily living (Collins et al. 2010). These early stressful experiences can alter the brain’s architecture and physiologic stress response systems of a developing child. Moreover, the resulting toxic stress from trauma in early childhood can have significant, harmful long-term psychological and physical health consequences (Shonkoff et al. 2012). Therefore, early intervention that addresses the impact of toxic stress is essential for the long-term well-being of our children.

The field of early childhood mental health is evolving, but there still is a relative lack of empirical research regarding treatment for very young children (birth to five years of age) who have experienced trauma (Chu and Lieberman 2010). The need for effective trauma-informed therapy is even more salient for children living in poverty, who are disproportionately more likely to experience potentially traumatizing events (Lieberman et al. 2011). Some evidence-based trauma therapy programs have been developed for use with children such as Trauma-Focused Cognitive Behavioral Therapy (TF-CBT; Cohen et al. 2006) and the Attachment, Self-Regulation, and Competency program (ARC; Arvidson et al. 2011; Blaustein and Kinniburgh 2010). An even smaller number of trauma-informed therapy programs are designed specifically for children under five years old, such as Child-Parent Psychotherapy (CPP; Lieberman and Van Horn 2008) and Attachment and Biobehavioral Catch-up (ABC; Bernard et al. 2012). While some of these trauma therapy programs have conducted efficacy research with diverse, low-income populations, these treatment programs rarely provide additional details regarding how to use a culturally-informed approach with families living in poverty. For example, research has found that effective therapy with families living in poverty often involves using a collaborative approach in setting goals for treatment and allows for flexibility within the therapy process. While some of the evidence-based treatment programs are intended to be tailored depending on the needs of the family, goals are typically determined by the nature of the program. Moreover, the completion of some long-term programs (such as CPP and ARC, both which recommend about 50 treatment sessions) may not be feasible for families with multiple life stressors. Furthermore, the design of some treatment programs requires a stricter adherence to a treatment manual, which may not allow for the kind of flexibility necessary when working in the context of families living in poverty. For example, the use of TF-CBT is reportedly contraindicated in situations where a caregiver also has untreated trauma, mental health concerns, or high levels of distress, or in families where there is household instability, serious ongoing conflict in the home, or basic needs are not being met (Chadwick Center for Children and Families 2008; Cohen et al. 2006; Lang et al. 2010). Exclusion criteria such as these may unintentionally preclude children and families experiencing ongoing stress who are most at need of therapeutic interventions.

Additionally, children living in poverty are less likely to have access to appropriate mental health services due to barriers including the lack of transportation or childcare, difficulty keeping regular appointments, variable schedules for work or school commitments, caregiver physical or mental health problems, child illness, mistrust of mental health services, high costs of mental health care, and/or inadequate insurance coverage for appropriate mental health services (Rowan et al. 2013). The practical challenges associated with accessing therapy services can be partially addressed by using an in-home treatment model. However, most evidence-based trauma therapy programs are conducted in an outpatient setting, rather than in the client’s home environment.

The Early Pathways (EP) Program was developed and tested through randomized controlled studies to address significant behavior problems in very young children (5 and younger) living in poverty using a home-based delivery system (Fung and Fox 2014; Harris et al. 2015). It also received high ratings as an evidence-based program by the Substance Abuse and Mental Health Services Administration’s National Registry for Evidence-Based Programs and Practices. Moreover, the California Clearing House for Evidence Based Programs rated EP as highly relevant for use by child welfare agencies. For the present project, starting with EP as the core treatment program, the trauma literature and available trauma-informed programs were reviewed (e.g., Briere and Scott 2012; Cohen et al. 2006; Ford et al. 2005; National Child Traumatic Stress Network 2012; Herman-Smith 2013; Thompson 2014). Practical treatment strategies based on attachment theories to address parent-child relationships and cognitive-behavioral therapy to focus on symptom reduction and emphasize positive parenting strategies for caregivers were integrated with EP. Recent research in the treatment of toxic stress in early childhood, following an ecobiodevelopmental framework (Shonkoff et al. 2012) also was included. For example, this research suggested treatment of early toxic stress should include a safe and stimulating home environment. While few existing evidence-based programs include a specific focus on the home environment, we added multiple strategies for assisting caregivers to accomplish this goal.

Method

Participants

This study was approved by the Internal Review Board of a Midwestern University. Participants were 64 children ages one-to-five years old referred to a clinic that specializes in serving very young children in poverty with emotional and behavioral problems (Fox et al. 2007). Eligibility criteria for participation in the research study were the following:

1. The child was five years old or younger at the start of treatment.

2. The child experienced some type of potentially traumatizing event, as indicated on the Traumatic Events Screening Inventory - Parent Report Revised (TESI-PRR). To qualify for participation the study, at least one response on the TESI-PRR was endorsed positively.

3. The child exhibited at least four symptoms of posttraumatic stress disorder (PTSD) as defined by the DSM-5; at least one symptom was an intrusion symptom and one was an avoidance and negative alterations in cognition symptom. This approach is consistent with previous research using clinical trials which required four DSM-IV-TR defined posttraumatic stress symptoms for study eligibility with at least one symptom of re-experiencing/intrusions and one of avoidance (Meiser-Stedman et al. 2008; Sheeringa 2011).

4. The family received public assistance, indicating that the household income was below the federal poverty level.

5. Signed consent was obtained by the child’s legal guardian.

6. The child and primary caregiver completed the comprehensive intake evaluation and at least five treatment sessions. Previous studies conducted with a similar population with behavior problems without trauma used a minimum number of 3 treatment sessions as part of treatment completion criteria (Fung et al. 2014). Given the complex nature of trauma, it was hypothesized that more minimum treatment sessions would be needed to produce change, and therefore participants were included in the final sample only if they completed at least five or more treatment sessions. Participants in the wait list control group were included in primary data analyses if they completed the initial intake and a second intake 4–6 weeks later. Participants in the wait list (WL) control group who subsequently completed at least 5 treatment sessions were also included in follow-up analyses. Children with autistic spectrum disorders, severe to profound intellectual disabilities, or serious physical illnesses were not included in this study and were referred to more appropriate services. Children who were eligible for in-home counseling services but did not meet all inclusion criteria for this research study and/or refused to be part of a research study received the full complement of mental health services offered.

Procedure

A comprehensive intake evaluation was completed for each participant in their home settings. A semi-structured parent interview was conducted to gain information regarding the child’s background, strengths, family composition and mental health history, child’s health history, daily routines and living skills, specific problem behaviors, and trauma history. The pretest assessment using the measures described below was also completed. All items were read to caregivers unless they preferred to answer them on their own. Any child meeting the PTSD diagnostic criteria from the DSM-5 (APA 2013) was given a psychiatric diagnosis that was reviewed by a qualified professional (e.g., licensed psychologist, licensed professional counselor, or a licensed clinical social worker). Treatment sessions were scheduled to occur once per week for one hour or longer if needed. The caregiver and therapist collaborated to identify treatment goals at the first session. Each week a treatment plan was reviewed with treatment goals and strategies to be practiced during the following week with a copy provided to the caregiver. Treatment sessions began by reviewing and documenting progress toward treatment goals and completing the weekly assessments. EP with the added trauma component was originally designed to take an average of 16 weeks to complete, depending on the individual needs of the child and family. However, the number of sessions varied depending on the needs of the child and the availability and motivation of the child’s primary caregiver. Additional booster sessions were sometimes provided after the 4–6-week follow-up session, depending on the needs of the family and clinical judgment of the therapist.

Treatment Program

The treatment program involved an integration of the evidence-based EP along with the new trauma-informed components. A copy of the integrated treatment program manual can be downloaded at www.marquette.edu/education/early-pathways. The program was piloted with three separate therapists and families prior to implementation. The five core elements of the EP were retained and included: (a) strengthening the parent-child relationship through child-led play, attunement practices and other nurturing activities; (b) helping parents maintain developmentally appropriate expectations for their child (c) helping parents learn cognitive strategies to respond calmly and thoughtfully to their child’s challenging behaviors; (d) using positive reinforcement to strengthen the child’s pro-social behaviors and listening sessions to improve the child’s compliance to parent requests; and (e) using limit-setting strategies to reduce the child’s challenging behaviors, such as redirection, ignoring, or time-out (Fox 2017). Limit-setting strategies were modified to reflect best practices in trauma-informed care. For example, a Time-In strategy was used in place of time-out or ignoring in cases where a child’s emotional outburst was triggered by a trauma reminder rather than a functional temper tantrum, or in cases where the child had not developed the ability to self-regulate emotions.

In addition to these core elements, the trauma-informed treatment components included: Basic Safety, Caregiver-Child Relationship, Predictable and Nurturing Environment, Trauma-Informed Limit Setting Strategies, Calming Strategies, Naming and Practicing Feelings, Healthy Thoughts and Feelings, Identifying Sources of Support, Building Prosocial Skills, Trauma Narrative Development and Implementation and Seeking Closure. Many of these treatment components were adapted from existing evidence-based programs such as Trauma-Focused Cognitive Behavioral Therapy (Cohen et al. 2006) or were developed to meet the unique needs of very young traumatized children in poverty. Parent handouts were developed to match the educational levels of the families.

Finally, in addition to the treatment program, we also routinely provide advocacy services for the family by referring them to available community resources (e.g., rent assistance, food pantries, pediatricians who accept Medicaid insurance, counseling for parents who were traumatized themselves, etc.). Often, we also provided direct advocacy services such as providing safety locks for cabinets, doors and windows, toys for the children, among other needed items in the home.

Clinician Training

Clinicians included licensed mental health professionals and graduate students enrolled in mental health programs. Spanish-speaking clients received the treatment program from either a bilingual therapist or an English-speaking therapist with a translator; all parent handouts were translated into Spanish for these families. All therapists trained in the new trauma-informed components of the treatment program had already received extensive training and experience with EP. The didactic training component for trauma-informed components included formal workshops, weekly staff meetings and additional training sessions as well as an ongoing review of the integrated treatment manual. All staff and graduate students received weekly group supervision sessions with a licensed psychologist; students also received weekly individual supervision. A treatment fidelity checklist was used with each case to ensure that the program was implemented with fidelity. Therapists were asked to indicate which treatment components were discussed with the family or implemented in each treatment session; these checklists were reviewed during supervision to ensure adherence to EP. Not every individual treatment activity was necessary for each family. For example, providing psychoeducation to parents who were victims of intimate partner violence was a necessary treatment component for children who had witnessed this violence, but may not be a relevant component for families who had not experienced this form of violence. For each of the categories of topics (e.g., Family Safety) therapists were trained to use clinical judgment to determine the extent to which each specific topic needed to be addressed with each family.

Measures

Traumatic Events Screening Inventory - Parent Report Revised (TESI-PRR)

The TESI-PRR (Ghosh-Ippen et al. 2002) includes 24 items such as “Has your child experienced the death of someone close to him/her” and is answered by a caregiver with either Yes, No, or Unsure. Inter-rater reliability for the original TESI (Ford and Rogers 1997) was reported to range from .73 to 1.0 for the different types of traumatic events (Gray and Slagle 2006).

The Early Childhood Behavior Screen

The ECBS (Holtz and Fox 2012) is a 20-item caregiver-report measure which assesses the frequency of a young child’s prosocial behaviors (e.g., “Shares toys”) and challenging behaviors (e.g., “Hits others”). Items are rated by the primary caregiver on a three-point frequency scale (3 = often, 2 = sometimes, 1 = almost never). Only the Challenging Behavior Scale (CBS) was used in the present study. Total scores on the challenging behavior scale (CBS) range from 10 to 30 with higher scores indicating a greater frequency of challenging behaviors. This tool was developed for use with children from low-income families and is written at a 3.9 grade level. Field-testing was conducted with a representative, diverse sample of 439 parents from low socioeconomic status in an urban community. The internal consistency using coefficient alpha was .87. The CBS demonstrated adequate levels of concurrent validity (r = .75) with the Eyberg Child Behavior Inventory (ECBI; Eyberg and Pincus 1999), as well as adequate levels of sensitivity (82%) and specificity (80%) based on the relationship with the ECBI. Harris et al. (2016) conducted a confirmatory factor analysis that provided support for the ECBS’ factor structure. Receiver operating characteristics (ROC) curve analyses were used to test the CBS’s ability to distinguish between 428 clinic-referred children and 245 non-clinic referred children. Results showed an acceptable fit model for the ECBS, providing further evidence of its construct validity. Optimal cut-scores by child age derived from the ROC curve analyses were provided. Sensitivity rates for cut scores ranged from 0.76 to 0.83 and specificity rates ranged from 0.88 to 0.95. Acceptable test–retest reliability and good internal consistency also was observed.

The Pediatric Emotional Distress Scale (PEDS)

Two subscales from the PEDS (Saylor et al. 1999) were used to assess for possible trauma symptoms: Anxious/Withdrawn (PEDS-AW) and Fearful (PEDS-F). The PEDS was designed for use with children ages two to ten years old and the items were written at a 4.0 grade level. The PEDS-AW includes six items (e.g., “Seems worried”), with subscale scores ranging from 6 to 24, and the PEDS-F includes five items (e.g., “Has bad dreams”) with subscale scores ranging from 5 to 20. Items are rated on a four-point Likert-type scale (1 = Almost Never, 2 = Sometimes, 3 = Often, 4 = Very Often). Four geographically and developmentally diverse samples were used to determine preliminary psychometrics. The authors reported adequate alpha coefficients for the PEDS-AW (r = .74) and the PEDS-F (r = .72). Concurrent validity was demonstrated with significant correlations between parents’ reports of PTSD symptoms on the Child Posttraumatic Stress Reaction Index (CPTS-RI; Frederick 1985) with both PEDS-AW (r = .62) and PEDS-F (r = .59).

Parent-Child Relationship Scale (PCRS)

This scale was used to measure the clinician’s subjective assessment of quality of the caregiver-child relationship as well as the caregiver’s adherence to the treatment program (Fox and Nicholson 2003). The PCRS uses a scale of 0–100 with five anchors at 20-point intervals: poor (ranging from 0 to 20), below average (ranging from 20 to 40), average (ranging from 40 to 60), good (ranging from 60 to 80), and exceptional (ranging from 80 to 100). Multiple descriptive markers are provided for each interval to improve inter-rater reliability (e.g., “Parent is often thoughtful when interacting with child” or “Parent can be responsive to child’s needs and set appropriate limits on child’s behavior, but not consistently”). Inter-rater reliability was determined based on 101 cases, and a kappa coefficient of .57 was reported (Fung and Fox 2014), indicating moderate agreement between observers (Viera and Garret 2005).

Therapist Treatment Report (TTR)

The TTR was completed during or immediately following each weekly treatment session. This report includes clinical notes, observations, and progress toward parent and child goals. The treatment report also included a four-item scale based on the primary objectives of EP, designed to assess caregiver adherence to program strategies. These items are: (a) “Does the parent maintain appropriate expectations?” (b) “Does the parent stop and think before responding?” (c) “Does the parent utilize rewards appropriately?” and (d) “Does the parent utilize appropriate discipline?” Items are rated by the clinician using a three-point Likert-type scale (1 = rarely/not at all, 2 = sometimes, 3 = most times). The four scores were combined for a composite score that ranged from 4 to 12, with higher scores representing greater caregiver adherence to treatment. For the present study, therapists were trained to rate these items in the context of the child’s trauma. For example, parent use of appropriate discipline refers to appropriate trauma-informed discipline strategies. Reliability for this scale was determined from 102 observations (Fung and Fox 2014). Two clinicians independently scored the items, and kappa coefficients were computed for each scale item: appropriate expectations = .89, stop and think = .92, utilized rewards = .95, utilized discipline = .89. The coefficient alpha of the entire scale for the sample was .88, indicating good agreement between observers (Viera and Garret 2005).

Family Satisfaction Survey (FSS)

A 7-item survey was used to assess caregiver satisfaction with the treatment services. This scale is used anonymously to facilitate caregivers providing honest feedback. On a 7-point Likert-type rating scale, caregivers were asked to rate: the quality of services received (1 = poor to 7 = excellent), how the services contributed to their child’s improvement (1 = not at all to 7 = a lot), how the clinic helped them to improve management of their child (1 = not at to 7 = a lot), if caregivers would use the clinic again if needed (1 = no, definitely not to 7 = yes, definitely), current status of the child’s referral concern (1 = considerably worse to 7 = greatly improved), if caregivers would recommend the clinic to others (1 = no, definitely not to 7 = yes, definitely), and the caregiver’s confidence in managing their child’s behavior in the future (1 = not at all confident to 7 = very confident). Total scores range from 7 to 49, with higher scores indicating greater satisfaction with services. In a similar study, internal consistency for this measure was reported to be r = .82 (Fung and Fox 2014).

Research Design

A convenience sample of children referred to the clinic for participation in the trauma therapy program was used. Eligibility for the trauma study was determined after the completion of the intake evaluation. Therefore, consecutively referred participants meeting full criteria for inclusion were randomly assigned to immediate treatment (IT) or wait list (WL) groups using a computer-derived random numbers table upon completion of the intake evaluation. Participants in the IT group were scheduled to start treatment within one week of their initial intake. Participants randomly assigned to the WL group waited four to six weeks for treatment services after their initial intake. Using a six week wait list was based on previous studies with populations living in poverty (Harris et al. 2015; Fung and Fox 2014). This shortened length of time on the wait-list was to avoid the risk of increased attrition due to a longer wait times to receive services.

All measures were administered for both IT and WL at intake (Time 1), except the satisfaction survey, which was only administered at the completion of the program. The TESI-PRR was only administered at Time 1 to screen for the presence of potentially traumatizing events. The primary assessments (ECBS-CBS, PEDS-AW, PEDS-F, and therapist Treatment Report (TR) were completed at each treatment session and at treatment completion. The WL group completed the measures again when beginning the treatment program (Time 2), and again at the completion of the program (Time 3). For IT and WL groups, a follow-up occurred six weeks after the final posttest to assess for maintenance of treatment gains using the study’s primary measures (ECBS-CBS, PEDS-AW, PEDS-F, TR, and PCRS).

Results

Participants

The final sample included 64 participants with 32 in the immediate treatment (IT) group and 32 in the wait list (WL) control group (see Table 1 for demographic information). There were 44 males (68.8%) and 20 females (31.3%). Children were African American (42.2%), multiracial (28.1%), Latina/o or Hispanic (18.8%), and European American (10.9%). Children were an average of 39.11 months old (SD = 13.32), and 20.3% of children had been previously diagnosed with a developmental delay. Families were asked to identify their primary referral concern at intake and again at post-treatment. At intake, the most common primary referral concerns were temper tantrums (40.6%) and aggression toward others (37.5%). Twenty-three percent of children had been exposed to two different potentially traumatic events in their lifetime, and 73% of children had been exposed to three or more different traumatic events.

Table 1.

Between Group Comparisons of Demographic Variables for Immediate Treatment and Wait List Groups

	Immediate (n = 32)			Wait List (n = 32)
Variable	%	X	SD	%	X	SD
Child Age (months)		40.91	14.31		37.31	12.20
Child Gender
Males	78.1			59.4
Females	21.9			40.6
Has developmental delay	18.8			21.9
Child Race
African American	31.3			53.1
Multiracial	34.4			21.9
Latina/o	21.9			15.6
European American	12.5			9.4
Caregiver
Biological mother	53.1			62.5
Both parents	25.0			12.5
Foster/kinship	15.6			15.7
Other relative	6.3			9.4
Caregiver age		32.16	10.25		30.91	10.95
Caregiver married	15.6			15.6
Caregiver employed	50.0			46.9
Mother finished 12th grade	84.0			84.6
Father finished 12th grade	76.5			58.8
Children in home		2.63	1.41		2.56	1.08
Number of traumatic events		5.06*	2.72		3.78	1.84

*p < .05

Caregivers were biological mothers (57.8%), both biological parents (18.8%), foster/kinship caregiver (15.6%), or other relatives (7.8%). Caregivers were more likely to be single (57.8% never married, 14.1% were separated, and 12.5% were divorced), and about one-half of caregivers were unemployed (51.6%). Of the children’s biological parents, 15.7% of mothers and 32.4% of fathers had completed less than a 12th grade education, 70.6% of mothers and 61.8% of fathers had completed 12th grade, and 10.9% of mothers and 5.9% of fathers had completed at least some post-high school education. Average caregiver age was 31.52 years (SD = 10.55).

The immediate treatment (IT) and wait list (WL) groups were compared on demographic variables using independent-group t-tests for continuous variables and chi-square tests for categorical variables (see Table 1). No significant differences were found on demographic variables. However, participants in the IT group endorsed more potentially traumatic events in the child’s lifetime based on the TESI-PRR [t (62) = 2.20, p = .031].

Attrition

A Consort Diagram (see Fig. 1) was used to show the flow of participants in each group throughout the entire study from intake through follow-through. As shown in Fig. 1, a total of 12 (27.2%) participants in the IT group dropped out prior to completing five treatment sessions and 5 (13.5%) participants in the WL group dropped out prior to completing a second intake. Treatment completers and non-completers were compared on demographic variables and pretest measures using independent-group t-tests for continuous variables and chi-square tests for categorical variables. There were no significant differences in demographic or pretest variables between treatment completers and non-completers.

Fig. 1.

Participant flowchart from random group assignment through short-term follow-up evaluations.

The average program duration and number of treatment sessions also were compared between the IT and WL treatment groups. The average program duration was 21.84 weeks (SD = 10.01) for the IT group and 20.31 weeks (SD = 9.54) for the WL group. Within the WL group, there was an average wait time of 7.25 weeks (SD = 5.52) from first intake (Time 1) to second intake (Time 2). The IT group completed an average of 10.22 sessions (SD = 5.10). For the combined sample, there was an average length of 7.5 weeks (SD = 8.85) between posttest session and short-term follow-up session.

Data Analyses

The IBM SPSS Statistics for Windows, Version 21.0 program was used to conduct the quantitative statistical analyses for this study. For all participants who met inclusion criteria, intention-to-treat (ITT) analyses were used with the last observation carried forward (Gupta 2011). This means that families in the immediate treatment (IT) group who dropped out of treatment after the fifth treatment session were still included in statistical analyses and their final measures were used for data analyses. Participants in the IT group who did not complete an intake and at least five treatment sessions were eliminated from the database and designated as non-completers. Similarly, participants in the wait list (WL) control group who did not complete a second intake (at Time 2) were eliminated from the database and designated as non-completers. Cohen’s effect size was used for all significant results (Cohen 1998).

Results for the study’s primary dependent measures by group are shown in Table 2. A significant group difference was found between the IT and WL groups on the ECBS-CBS with a large effect size [F (1, 61) = 25.55, p < .001, Cohen’s d = .97). Results of a MANCOVA demonstrated significant differences on the two PEDS measures [F (2, 59) = 13.08, p < .001]. Univariate results showed significant between-group differences on the PEDS-AW with a large effect size [F (1, 60) = 22.97, p < .001, Cohen’s d = 1.05), and a medium effect size on the PEDS-F [F (1, 60) = 8.04, p < .01, Cohen’s d = .59). Results of an ANCOVA revealed a significant between-group difference in the PCRS with a medium effect size [F (1, 56) = 7.70, p < .01, Cohen’s d = .52]. Results of an ANOVA also revealed a significant between group difference in TR scores at Time 2 with a large effect size [F (1, 62) = 53.11, p < .001, Cohen’s d = 1.82]. Finally, a repeated measures analyses of variance (ANOVAs) was conducted to determine if significant changes were made across the three measurement points (pretest, posttest, and follow-up) for the combined sample of both groups among participants who completed at least five treatment sessions (see Table 3). For the WL group, pretest scores from the second intake were used in analyses. Results showed a significant change on the ECBS-CBS from pretest to follow up with a medium effect size [F (2, 40) = 10.78, p < .001, Cohen’s d = .75]. Analyses also revealed significant changes with large effect sizes in both the PEDS-AW [F (2, 40) = 11.99, p < .001, Cohen’s d = 1.04) and PEDS-F [F (2, 40) = 8.57, p < .01, Cohen’s d = .80]. Results demonstrated significant changes across time with large effect sizes on the PCRS [F (2, 30) = 10.53, p < .01, Cohen’s d = .97), and the TR [F (2, 32) = 47.66 p < .001, Cohen’s d = 1.94). There were no significant between-group differences on any of the outcome measures at follow-up. We also compared the informal symptom checklist between the immediate and delayed groups. For the WL group, all children continued to display their primary referral concern at intake and the second pretest (e.g., aggression, tantrums, excessive crying, self-injury). For the immediate treatment group, only 28% of the children continued to have symptoms following treatment (e.g., nightmares, oppositional behaviors, tantrums, and aggression). Scores from each of the seven items on the Family Satisfaction Survey were summed to create a total score that ranged from 7 (low satisfaction) to 49 (high satisfaction). The mean score at posttest was 46.40 (SD = 2.38) for the IT group, indicating a high level of satisfaction.

Table 2.

Analyses of Covariance for Dependent Measures for Immediate Treatment (IT) and Wait List (WL) Groups at Pretest and Posttest/Pretest 2

	Time 1				Time 2
	IT Pretest		WL Pretest		IT Posttest		WL Pretest 2
Measures	M	SD	M	SD	M	SD	M	SD	d
ECBS-CBS	23.03	4.25	22.81	4.24	17.70***	4.49	22.17	4.73	.97
PEDS-AW	11.44	3.47	11.81	4.06	8.52***	2.06	11.10	2.80	1.05
PEDS-F	12.13	3.29	11.97	3.43	8.69**	3.33	10.65	3.32	.59
PCRS	60.47*	14.67	52.03	13.13	64.77**	18.38	56.44	3.96	.52
TR	n/a		n/a		9.81***	1.99	6.47	1.66	1.82

*p < .05, ** p < .01, ***p < .001. Adjusted Time 2 scores based on analyses of covariance (ANCOVA). The notation d refers to Cohen’s d effect size of ANCOVA comparisons at Time 2 with pretest scores as covariates, based on adjusted mean scores. For TR, the notation d refers to Cohen’s d effect size of ANOVA comparison at Time 2

Table 3.

Repeated Measures ANOVAs with Pairwise Comparisons for Both Groups Combined at Pretest, Posttest, and Follow-up

	Pretest		Posttest		Follow-up
Measures	M	SD	M	SD	M	SD	d
ECBS-CBS	21.62	4.96	16.48**	5.02	17.24**	6.63	.75
PEDS-AW	10.86	2.65	8.19**	2.46	8.38***	2.38	.99
PEDS-F	9.71	3.00	7.24**	2.34	7.48*	2.54	.80
PCRS	60.00	16.53	72.81***	13.54	74.38***	14.13	.97
TR	6.53	1.66	10.53***	1.81	10.35***	2.23	1.94

*p < .05, ** p < .01, ***p < .001. The notation d refers to overall effect size from pretest to follow-up

Discussion

This study involved the development, implementation, and evaluation of the expanded EP Program, a home-based parent-and-child therapy program for very young children living in poverty who have significant behavioral and emotional problems and experienced traumatic events. Results of this study revealed that children who participated in EP decreased challenging behaviors as well as anxious/withdrawn and fearful symptoms of trauma (such as sleep disturbance, clinging behavior, or being easily startled). In addition, based on clinician observation, the quality of the caregiver-child relationship improved and caregivers improved in their abilities to use therapy strategies (such as remaining calm, maintaining fair expectations, implementing positive parenting strategies, etc.). Caregivers also reported a high level of satisfaction with the program after their participation. Moreover, these improvements were maintained at least 4–6 weeks after ending services.

These results are similar to previous studies using EP (Fox and Holtz 2009; Fung et al. 2014). The results of this present study also are comparable with the results of TF-CBT research, which also used an RCT methodology (Sheeringa et al. 2011). Both TF-CBT and EP incorporated some similar aspects to treatment such as: enhancing safety, providing psychoeducation, and developing coping skills (such as cognitive coping and relaxation techniques). Given the positive impact of both treatment approaches, the results of the present study may indicate that a more strictly manualized treatment approach with rigid exclusion criteria may not be necessary to produce positive effects.

The results of the present study indicated that trauma-informed therapy can be effective with families living in poverty. Previous research has identified poverty as a risk factor for poor treatment adherence (Armbruster and Fallon 1994; Kazdin and Mazurick 1994), yet children living in poverty are significantly more likely to be exposed to violence, abuse, or other sources of chronic stress (Lieberman et al. 2011), and therefore are in greater need for trauma-informed therapy. Despite the challenges associated with service delivery and program completion (e.g., attrition), children in this study experienced overall improvements with impressive effect sizes. Importantly, the mean scores on symptom measures (ECBS and PEDS) fell generally within normal ranges based on cutoff scores for clinical significance at both posttest and follow-up. This indicates that after receiving services, the frequency of these challenging behaviors or symptoms of anxiety and fear are comparable to a general population of same-aged children.

Limitations and Directions for Future Research

The study used a convenience sample of clients referred to the clinic for mental health services. However, a strength of using the normal referral procedure to recruit participants is that they are more likely to reflect typical cases (Jensen et al. 2014). Another methodological weakness is that the therapists both provided the therapy services and administered the measures. Thus, there is a risk of researcher allegiance bias as therapists were not blind to the treatment condition. As this was a preliminary study with a new treatment program, there were multiple factors that are yet unknown, including the actual minimum number of treatment sessions needed to produce change. There were two cases out of the original sample who were reported by therapists to have completed the treatment “successfully,” but were omitted from the research study because they had only completed four treatment sessions. While these two cases were certainly outliers (as families in the study completed treatment with an average of 10 sessions), it does lead to the question of treatment dosage. Barkham et al. (2006) proposed a good enough level model of dose-effect relations, suggesting that “in routine practice, level of improvement and treatment duration are mutually regulated so that treatments tend to end when clients, on average, have improved to a degree or level that is good enough” (p. 161). This model of treatment dosage also encompasses the idea of “therapist responsiveness” in which the length of treatment is determined by the psychosocial context of the therapeutic environment (Barkham et al. 2006; Stiles et al. 1998). Future research studies may identify the minimum necessary number of treatment sessions needed to produce reliable change in a population of very young children who have experienced trauma. In addition to determining the most efficient and efficacious dose, future research may also identify which specific components of the therapy program are most beneficial to children and caregivers and best predict successful treatment outcomes.

Family attrition remains a consistent challenge for professionals who choose to work with young children in poverty with significant mental health problems. These families frequently experience barriers that often limit their regular participation in treatment over a prolonged period. Recently, Gresl et al. (2016) developed a Treatment Barriers Scale for families with young children in poverty. This study found that the number of barriers endorsed by a family predicted treatment success.

Clinical Implications

There are multiple clinical implications resulting from this study. From an ethical perspective, poverty cannot remain a reason that families in greatest need of support do not receive adequate mental health services. There are several strategies that mental health providers may employ to help counteract the “logistical, attitudinal, and systemic barriers” to receiving services (Santiago et al. 2013, p. 117). These strategies may include: establishing frequent phone contact prior to beginning services, providing services in the home or community, offering flexible scheduling, providing culturally congruent services, offering bilingual services, and when relevant, acknowledging clients’ experiences of oppression or racism (Santiago et al. 2013). The American Association of Pediatrics (AAP) has suggested that the community may be the most effective means of reaching vulnerable children, and recommended that empirically validated, community- and home-based interventions be replicated on a larger scale (Garner et al. 2012). Therefore, it would be valuable to continue training mental health service providers in EP. Finally, premature termination of treatment remains an ongoing challenge for this population. While there is no simple solution to this problem, our research has found that we can identify families by the third treatment session who are likely to succeed. For those who are not showing expected progress for their children, we have a frank discussion about the caregivers’ level of engagement in the treatment process including motivational interviewing strategies. For some families, this helps them become motivated and more engaged. Other families acknowledge that current barriers they are facing need to be addressed before they can fully commit to the treatment program for their children. Clearly, more research on young children with mental health problems who live in poverty is needed. Unfortunately, this line of research is only in its infancy.