A Preliminary Analysis of Parent-Child Interaction Therapy Plus Natural Helper Support to Increase Treatment Access and Engagement for Low-Income Families of Color

Abstract

Disparities in care for low-income children of color call for innovative culturally and linguistically responsive solutions to better engage marginalized populations in evidence-based interventions. In partnership with a community organization, the addition of natural helper support as an adjunct to Parent-Child Interaction Therapy (PCIT+NH) was examined as a strategy to increase recruitment, engagement, and retention in PCIT for families historically unreached by a university-based clinic. Natural helpers provided home-based skills practice and support for forty-two families whose parents were more racially and linguistically diverse and had lower income and lower caregiver education than the typical population served by the same program (i.e., program population). Families who received PCIT+NH had comparable or higher rates of engagement and improvements in clinical outcomes (i.e., decreased child externalizing and internalizing behaviors, increased child compliance, decreased caregiver stress, increased caregiver parenting skills) relative to the program population. Furthermore, higher doses of natural helper support were associated with higher rates on most measures of treatment engagement (i.e., treatment completion, completion of the Child Directed Interaction phase of treatment, PCIT sessions, homework in the Parent Directed Interaction phase of treatment), with the exception of homework in the Child Directed Interaction phase of treatment and overall session attendance rate. Next steps for testing the treatment engagement and clinical outcome effects of the PCIT+NH model are discussed.

1. Introduction

Children of color and their caregivers are less likely to access, engage in, and benefit from high-quality, evidence-based mental health services (Alegria et al., 2010; Aratani & Cooper, 2012; Baker et al., 2011; Barnett, Lau et al., 2019; Coker et al., 2009). The unmet need for mental health services appears to be greatest among low-income families of color (Villatoro et al., 2018). A range of individual, provider, and systemic barriers limit the ability of low-income families of color to access mental health services, including mental health stigma, lack of insurance coverage, and under-recognition of mental health problems (Alegria et al., 2010; Kilbourne et al. 2006; Lindsey et al., 2013; McKay & Bannon, 2004; Yeh et al., 2003; Young & Rabiner, 2015). For immigrant families, limited English proficiency, lack of familiarity with healthcare systems, and fear of being reported to authorities for undocumented status can exacerbate challenges to engaging in mental health services (Derr, 2016). Further, families of color have additional vulnerabilities that may increase their need for mental health services, as they are more likely to live in poverty, experience community violence, and experience systemic racism, which are risk factors associated with increased internalizing and externalizing symptoms (Fowler et al., 2009, Hodgkinson et al., 2017, Loyd et al., 2019). Despite evidence to support the increased risk for negative outcomes, families from racially, ethnically, and economically diverse backgrounds also are often faced with limited access to and engagement in culturally and linguistically appropriate care (Barnett et al., 2018b; Huang et al., 2004).

To address these disparities in care, substantial efforts have been directed toward understanding how to enhance caregiver engagement in mental health services broadly, and evidence-based behavioral parenting programs specifically (Becker et al., 2018; Chacko et al., 2016; Gopalan et al., 2010; Haine-Schlagel & Walsh, 2015). One strategy that has received attention is partnering with lay health workers, who have similar backgrounds as the caregivers they are working with, to support caregiver engagement in care (Callejas et al., 2010; Barnett et al., 2018b; Gustafson et al., 2021). The current study investigated the impact of partnering with natural helpers, a lay health worker model, to increase access, engagement, and retention in one evidence-based treatment, Parent-Child Interaction Therapy (PCIT), for low-income, minority families in a large metropolitan area in the Southeast United States.

Behavioral parenting programs, such as PCIT, have been identified as having the strongest evidence base for young children with disruptive behavior disorders (Kaminski & Clausen, 2017). Early research on behavioral parenting programs predominantly included non-Hispanic, White Americans, leading to concerns about the cultural appropriateness and effectiveness of these programs for families of color (Forehand & Kotchick, 1996; Ortiz & Del Vecchio, 2013). However, accumulating evidence suggests that behavioral parenting programs can be effective with African American, Asian American, and Latinx families (Lau, 2006; McCabe & Yeh, 2009; Ortiz & Del Vecchio, 2013; Reid et al., 2001). However, even with comparable clinical outcomes, challenges remain to successfully engage low-income, families of color in behavioral parenting programs (Lavigne et al., 2010; Ortiz & Del Vecchio, 2013). Latinx and African American caregivers have been shown to be less likely to enroll in behavioral parenting programs than non-Hispanic, White caregivers (Baker et al., 2011). Indeed, a review of engagement in behavioral parenting programs, found significant challenges across engagement domains including enrollment, attendance, and retention, especially for low-income families (Chacko et al., 2010; Quetsch et al., 2020). When enrolled, low-income caregivers of color may require more sessions and skill practice to complete treatment successfully (Fernandez et al., 2011; Lau, 2006; Matos et al., 2006; McCabe & Yeh, 2009; Ramos et al., 2018). Finally, attrition in community settings, which are more likely to serve low-income families of color than those found in efficacy trials, can reach 67% (Lanier et al., 2011; Lyon & Budd, 2010; Pearl et al., 2012), further highlighting the critical need to address disparities in mental health care for families at greatest risk.

PCIT has unique benefits and challenges related to treatment engagement. It uses an in vivo feedback model, in which therapists coach caregivers in their skill use for the majority of treatment sessions. Coaching overcomes inherent challenges to didactic and discussion methods used in other behavioral parenting programs, as it necessitates active participation and assesses learning in real time (Barnett et al., 2017). PCIT also requires that caregivers demonstrate a high level of proficiency with the targeted parenting skills before they advance from the relationship-focused phase of treatment to the discipline phase, and then until they graduate from treatment (Eyberg & Funderburk, 2011). Using these criteria guarantees that all caregivers can successfully use the skills upon termination; however, caregivers often drop out before they learn discipline skills, or their length of treatment can far exceed those found in efficacy trials of PCIT (Lanier et al., 2014). Meeting these goal criteria might take longer when the targeted parenting skills are less culturally familiar (Ramos et al., 2018) or caregivers complete less daily home practice (Stokes et al., 2016). Problems with family retention and meeting goal criteria have downstream effects on PCIT implementation in systems of care. It can take three years to meet PCIT certification requirements (i.e., graduating two cases), which can lead to therapist attrition from training and limits the return on service systems’ investments in PCIT (Beveridge et al., 2015, Timmer et al., 2016; Scudder et al., 2016).

In order to improve the accessibility and acceptability of behavioral parenting programs for low-income families of color, it has been recommended that natural helpers (e.g., lay health workers) be incorporated into care (Acevedo-Polakovich et al., 2013; Acevedo-Polakovich et al., 2014; Barnett et al., 2018b; Calzada et al., 2005). Natural helpers are “lay people to whom others naturally turn for advice, emotional support, and tangible aid” (Israel, 1985, p. 68). Social proximity, or closeness to members of the community served, is a core feature of natural helpers, and has been characterized by themes of 1) shared experiences of parenting, 2) familiarity with cultural factors, such as ethnicity and neighborhood characteristics, and 3) personal experiences of life challenges (Gustafson et al., 2018). Given their social proximity to the communities they serve, natural helpers may be uniquely positioned to increase access to behavioral parenting programs and support the engagement of caregivers throughout treatment.

There has been a rapid increase in research examining the role of lay health workers, such as natural helpers, in mental health interventions since 2010; however, it is still not clear what roles are most appropriate for the providers in regards to delivering evidence-based treatments (Barnett et al. 2018a; van Ginneken et al., 2013). Lay health workers may serve in a range of roles from conducting community outreach to increase referrals to traditional mental health systems of care to serving as the primary providers of care (Barnett et al., 2018b). To date, the majority of research has investigated the effectiveness of task-shifting treatment delivery to lay health workers within the context of low- and middle-income countries (Barnett et al., 2018a). Even though increasing evidence suggests that non-professional providers can effectively provide evidence-based treatments, a number of barriers limit the implementation of this workforce model within the United States. For example, professional degrees and mental health licenses are often required to provide and bill for treatments. Instead, it has been suggested that natural helpers serve in auxiliary roles to the delivery of evidence-based treatments, in which they connect families to services, address barriers to care, and promote treatment engagement and adherence (Ayala et al., 2010; Barnett et al., 2018b; Rodriquez et al., 2011). However, limited research has been conducted to establish how these auxiliary services impact engagement and clinical effectiveness of evidence-based treatments in general, and behavioral parenting programs specifically (Barnett et al., 2018a; Hoagwood et al., 2010).

1.1. Study Aims and Hypotheses

The current study describes the implementation and initial support for the effectiveness of a lay health worker auxiliary model to increase access, engagement, skill acquisition, and retention for low-income families of color in PCIT through the additional support of a community-based natural helper (PCIT+NH). This PCIT+NH model was developed out of a collaboration between a university-based behavioral health service provider and a community-based organization with a well-established natural helpers program serving a large county where approximately 70% of the population identifies as Hispanic/Latinx and 17.7% identify as Black or African American (U.S. Census, 2019; see Barnett et al. (2016) for more details on model development and natural helper training). The purpose of this collaboration was to reach high-need families who were previously underserved by this university-based PCIT program and explore the potential impact of the PCIT+NH model on access to and engagement with services for low-income families of color. To evaluate the success of these efforts, the current study first examined demographic characteristics (i.e., child age, gender, race, and ethnicity; parent race, ethnicity, education and income; language of service delivery) and treatment engagement (e.g., attendance, homework completion) and retention (e.g., completion of first phase of treatment and completion of treatment) for families served with the PCIT+NH model as compared to families typically served by the same university-based PCIT clinic (i.e., program population). This first comparison allows us to answer the following question: “Does the PCIT+NH model help us reach diverse and historically underserved families we would not otherwise reach in our clinic?” We hypothesized that the PCIT+NH model would allow us to reach more diverse families, in terms of race, ethnicity, income, and education, than our typical program population. Second, outcomes related to clinical effectiveness (i.e., caregiver skill acquisition and stress, child externalizing and internalizing behaviors, and child compliance) were examined for families receiving PCIT+NH support and compared to outcomes for the overall program population served. This second comparison allows us to answer the following question: “Does the PCIT+NH model help these new, historically underserved families just as much, or more than, the PCIT treatment we have historically provided in our clinic to our typical population of families?” We anticipated that the PCIT+NH sample would have at least comparable levels of engagement, retention, and clinical outcomes as those of families typically seen in our clinic given the robust effects of PCIT for diverse families (e.g., McCabe & Yeh, 2009) and the additional opportunities for caregivers to practice skills with their child, as skills practice is known to support parent skill acquisition and decreased child externalizing behavior (Kaminski, 2008). Third, among families receiving the PCIT+NH model, the study examined the relationship between the amount of natural helper support a family receives (i.e., dose) and variables related to treatment engagement (e.g., attendance, homework completion), retention (e.g., completion of first phase of treatment and completion of treatment), and clinical outcomes. This third comparison helped us answer the following question: “Do families get even better the more they are exposed to the NH?” Furthermore, we anticipated a positive relationship between dose of natural helper support and engagement, retention, and clinical outcome variables.

2. Method

2.1. Participants

Participants included forty-two children ages 2–8 years and their primary caregiver, who presented for treatment at three university-based PCIT clinics embedded within three racially, ethnically, and economically diverse communities in a large metropolitan area in the Southeast United States. The PCIT clinics were established in 2011 through a county-funded initiative to increase the reach of parenting interventions to underserved communities. The program provides PCIT services free of charge to children aged 2 to 8 years and their families who live in the large metropolitan county where the program is based. Table 1 presents detailed demographic information for the sample. Child participants had a mean age of 4.96 years (SD = 1.73 years) and were predominantly male (76.2%). Approximately 90.5% of children were identified as a member of a racial/ethnic minority group by the primary caregiver. Primary caregivers were 92.86% biological mothers, had a mean age of 35.98 years (SD = 8.0 years), and education level ranged from did not complete high school to graduate degree, with 50.0% of primary caregivers reporting they completed high school or less, and 31.58% of families were single parent families. Regarding income, of the 40.48% of the PCIT+NH sample that chose to report income, 52.94 % of the sample reported an annual household income less than $35,000. Services were delivered in English (44% of families), Spanish (50% of families) and a mixture of Creole and English (5.26% of families).

Table 1.

Comparing Demographic Characteristics in PCIT+NH Sample and Program Population

Variable	PCIT + NH Sample M (SD) or %	Program Population (SD) or %	t(df) or χ2(df)	p	d
Demographic Characteristics
Child Age	4.96 (1.73)	4.74 (1.64)	0.83 (41)	.41	.13
Child Gender	76.19% Male	72.98% Male	0.12 (1)	.73	.09
Child Ethnicity	66.67% Latinx	67.97% Latinx	0.16 (2)	.92	.02
	4.76% Haitian	3.62% Haitian
	28.57% Non-Hispanic or Haitian	28.41% Non-Hispanic or Haitian
Child Race	61.90% White	71.59% White	6.77 (4)	.15	−.17
	21.43% Black	11.70% Black
	7.14% Multiracial	11.70% Multiracial
	7.14% Other	4.17% Other
	2.38% Native American	0.84% Native American
Parent Gender	92.86% Female	89.97% Female	None, Used	.76	.20
	7.14% Male	10.03% Male	Binomial Test
Parent Ethnicity	66.67% Latinx	66.57% Latinx	0.44 (2)	.80	−.02
	4.76% Haitian	3.06% Haitian
	28.57% Non-Hispanic or Haitian	30.36% Non-Hispanic or Haitian
Parent Race	59.52% White	73.82% White	16.41 (4)	<.01 *	.37
	23.81% Black	13.09% Black
	4.76% Multiracial	7.52% Multiracial
	9.52% Other	3.62% Other
	2.38% Native American	0.28% Native American
Parent Education	19.05% Some Grade School	2.72% Some Grade School	64.68 (5)	<.01 *	−.88
	30.95% HS Diploma/GED	10.90% HS Diploma/GED
	11.90% Some College	14.99% Some College
	2.38% Associates Degree	8.72% Associates Degree
	21.43% Bachelor’s Degree	27.25% Bachelor’s Degree
	14.29% Graduate Degree	35.42% Graduate Degree
Parent Income	23.53% $0–$19,999	16.25% $0–$19,999	2.77 (16)	.01 *	−.60
	29.41% $20,000–$34999	13.75% $20,000–$34999
	17.65% $35,000–$49,999	10.00% $35,000–$49,999
	11.76% $50,000–$74,999	13.13% $50,000–$74,999
	5.88% $75,000–$99,999	13.13% $75,000–$99,999
	11.76% >$100,000	33.75% >$100,000
Language Services Delivered	44% English	76.65% English	24.17 (2)	.01 *	.78
	50% Spanish	22.26% Spanish
	5.26% English & Creole	1.10% Creole

^*p<.05

2.2. Historical Treatment Population

The comparison was selected from a database of families served by our clinic before establishing the natural helper model (between the years of 2011 and 2019). This historical sample is defined in the current manuscript by the term “program population.” We were able to make these comparisons because we obtain participating families’ consent to keep administrative medical records on every family we have seen in our clinic. The assessment process for the program population did not differ from the PCIT+NH sample (see Study Design and Procedure for more detail), with the exception of the mid assessment timepoint, which was not completed for the program population. Institutional Review Board (IRB) approval was obtained from the university and all participants who agreed to be in the study signed an informed consent. All study procedures were conducted in accordance with the ethical standards of the IRB and with the 1964 Helsinki declaration and its later amendments.

2.3. Measures

2.3.1. Eyberg Child Behavior Inventory (ECBI)

The ECBI (Eyberg & Pincus, 1999) is a 36-item caregiver-report measure of disruptive behavior in children 2–16 years. The ECBI has demonstrated acceptable levels of reliability and validity in racially/ethnically diverse populations (α = 0.94, test-retest = 0.75; Gross et al., 2007) as well as stability over time and sensitivity to treatment change. The ECBI consists of two scales. The Intensity Scale assesses the frequency with which the caregiver observed a behavior, rated on a scale ranging from 1 (never) to 7 (always). The Problem Scale assesses whether or not the caregiver perceives the behavior to be problematic, rated as either “yes” or “no.” Raw scores ≥ 131 (T score ≥ 60) on the Intensity Scale are considered clinically significant. The ECBI was used in this study to assess eligibility as well as to measure changes in child disruptive behavior over the course of treatment. The ECBI was administered before treatment began (i.e., “baseline”) and immediately after treatment ended (i.e., “post-treatment”).

2.3.2. Behavior Assessment System for Children, Third Edition, Parent Rating Scale (BASC-3 PRS)

The BASC-3 PRS (Reynolds and Kamphaus, 2015) is a caregiver-report measure of emotional, behavioral, and adaptive functioning of children aged 2–21 years. The BASC-3 has well-established validity and reliability (α = 0.83–0.96, test–retest = 0.87–0.92; Reynolds & Kamphaus, 2015). Caregivers rate the frequency with which they observe behaviors, ranging from (1) never to (4) almost always. T-scores ≥ 70 are clinically significant and T-scores 60–69 are considered at-risk. The BASC-3 Externalizing and Internalizing Problems composite scores were used as outcome measures of child externalizing and internalizing behaviors. The BASC-3 PRS was measured before treatment began and immediately after treatment ended.

2.3.3. Dyadic Parent-Child Interaction Coding System, Fourth Edition (DPICS-IV)

Caregiver-child interactions were coded using the DPICS-IV (Eyberg et al., 2013), a behavioral observational tool that assesses caregiver behaviors and the quality of caregiver-child interactions. The DPICS-IV demonstrates acceptable reliability and validity of the coding categories (Eyberg et al., 2013). It is comprised of three 5-minute observations: child-led play (CLP), caregiver-led play (PLP), and clean-up (CU) situations. During CLP, the clinician conducted live coding of the frequency of positive parenting skills (e.g., labeled praises, reflections, behavior descriptions) and negative parenting skills (e.g., questions, commands, criticisms). During PLP and CU, caregiver verbalizations were coded for the rate of caregiver effective commands, the rate of caregiver correct follow-through on effective commands (i.e., praises for child compliance or time out warnings following noncompliance), as well as the rate of child compliance to effective commands. Coding was conducted live by PCIT therapists who were trained to 80% coding reliability by a certified PCIT International Trainer. The DPICS-IV was used as an outcome measure of caregiver skill acquisition, and was conducted both before treatment began and immediately after treatment ended. Natural helpers were also trained to use the DPICS-IV to code parent-child interactions during natural helper visits; however, natural helper coding data were not used in the present study.

2.3.4. Parenting Stress Index, Fourth Edition: Short Form (PSI-SF-4)

The PSI-SF-4 (Abidin, 2012) is a well-validated, 36-item self-report questionnaire designed to assess stress and caregiver-child relationship difficulties for caregivers of children from birth through 12 years of age and has been shown to have good internal consistency (α = 0.96; Abidin, 2012). The PSI-SF-4 items are rated on a 5-point scale from strongly agree to strongly disagree and yield a Total Stress score as well as scores for 3 subscales: Parental Distress (PD), Parent-Child Dysfunctional Interaction (PCDI), and Difficult Child (DC). Scores include T-scores and percentiles, with 85^th-89^th percentiles in the high range and the 90^th or higher percentiles in the clinically significant range. The PSI-SF-4 Total Stress percentile scores were used as an outcome measure of caregiver stress and were measured both before treatment began and immediately after treatment ended.

2.3.5. Treatment Engagement

2.3.5.1. Treatment Attendance.

Treatment attendance was calculated as the total number of coaching sessions attended for both CDI and PDI phases of treatment, as well as the rate of session attendance.

2.3.5.2. Natural Helper Session Attendance.

For the PCIT+NH group, Natural Helper session attendance was calculated as the total number of Natural Helper sessions attended by a family.

2.3.5.3. Homework Completion.

As per the PCIT protocol (Eyberg & Funderburk, 2011), caregivers were instructed to complete daily skills practice. CDI and PDI homework completion were defined as the number of days in which the caregivers reported to have engaged in practice of CDI and/or PDI skills outside of sessions divided by the number of days available to practice between sessions. For further description of homework record-keeping and calculation of practice rate variables, see Stokes and colleagues (2016).

2.3.6. Treatment Retention

2.3.6.1. CDI Completion.

Completion of the CDI phase was defined as caregivers meeting CDI skill criteria (i.e., 10 behavioral descriptions, 10 reflections, 10 labeled praises, and less than 3 questions, commands, and criticisms during a 5-min child-led play observation). At CDI completion, families were scheduled to move to the PDI phase of treatment.

2.3.6.2. Treatment Completion.

As per the PCIT protocol (Eyberg & Funderburk, 2011), treatment completion was defined as caregivers meeting CDI (see above) and PDI skill criteria (i.e., 75% effective command rate, 75% correct follow-through after child compliance/non-compliance), rating their child’s behavior within normal limits on the ECBI (Raw scores ≤ 114), and reporting they feel confident managing their child’s behavior on their own.

2.4. Study Design and Procedure

2.4.1. Recruitment and Screening

Participants in the PCIT+NH sample were recruited through a community-based organization, ConnectFamilias, which has an academic-community partnership with the PCIT program and serves as one of the sites for the university-based PCIT clinics (Barnett et al. 2016) provides free services, including care coordination wraparound services and parenting classes, in English and/or Spanish, to the predominantly immigrant Latinx population in the surrounding neighborhood. ConnectFamilias’s natural helpers are recognized and trusted individuals who serve as a bridge between families and informal and formal community resources and supports. Natural helpers participate in community health fairs and events, conduct community outreach, and facilitate parent meetings and workshops in the community to inform families and link them to community resources. Given their relationship with families in the community and access to ConnectFamilias’s cross-sector provider network and systems of care partners (i.e., schools, child protective services, etc.), natural helpers were in a position to provide insight and accessibility to historically underserved families who have traditionally shied away from therapeutic partnering services and who were previously unreached by our PCIT program.

Through this ConnectFamilias university partnership, screening criteria were identified and cross learning was utilized (e.g., community-based organization provided insight related to community stigma, fear, and norms) to co-develop strategies to: identify families who would be appropriate for PCIT, connect with diverse families and introduce PCIT in a manner that is culturally and linguistically responsive, assist with completion of initial screening and eligibility forms, and provide a warm hand-off for families throughout the PCIT intake process. Once eligibility was established and families were enrolled in PCIT+NH and natural helpers provided the family with PCIT-related support throughout treatment. See Barnett et al. (2016) for detailed information regarding demographics, as well as recruitment and training of the natural helpers. Participants in the program population sample were recruited primarily through self-referral and community provider referral and were screened for eligibility via a phone screening, which consisted of providing verbal consent and completing a background form and behavioral questionnaires in English or Spanish. Eligibility criteria were the same for the PCIT+NH sample and the program population. Specifically, if the primary caregiver a) rated the child’s behavior as clinically elevated on the Eyberg Child Behavior Inventory Intensity Scale (ECBI; Intensity Raw Score ≥131; Eyberg & Pincus, 1999) or the Externalizing Problems subscales or composite of the Behavior Assessment System for Children, Third Edition (BASC-3 T-Score ≥ 60; Reynolds & Kamphaus, 2015), or b) reported a history of child protective services involvement, the family was eligible to receive free PCIT services through their preferred clinic. Families who did not meet these inclusion criteria were referred to other interventions for appropriate services when needed. Families were excluded from receiving services if they resided in a different county (due to county-specific funding) or if they required services in a language other than English, Spanish, or Creole (due to lack of clinician availability). Comorbid child disorders did not exclude eligibility.

2.4.2. Assessments

If inclusion criteria were met, families were contacted by a therapist to confirm eligibility, assess interest in the program, and schedule the first appointment (i.e., baseline assessment). Families completed the baseline assessment in clinic, which included written consent, completion of questionnaires about their child’s functioning, a 20-minute behavioral observation of parenting skills and child behavior, and completion of a clinical interview, which contained questions related to the presenting concern, developmental milestones, medical history, educational history, previous assessment and treatment history, and use of discipline strategies. Families involved in the current study also participated in a smaller pilot study examining the use of the Cultural Formulation Interview (CFI; Lewis-Fernández, et al, 2014); wherein families were randomized to receive a standard clinical interview or the CFI in addition to their standard clinical interview. Families from both groups were included in the current study sample. CFI group status was controlled for in all appropriate analyses in the current study.

Following the baseline assessment, assessments also were conducted at post-treatment (i.e., following graduation) and consisted of primary caregivers completing questionnaires about their child’s functioning and a 20-minute behavioral observation of parenting skills and child behavior. Assessments were completed in either English or Spanish, according to caregiver preference. In addition, the ECBI was administered as a weekly measure of disruptive behavior during treatment.

2.4.3. PCIT +NH Model

PCIT services were provided at three university-based PCIT clinics embedded within three different communities throughout the county, according to caregiver preference and/or therapist availability. Masters and doctoral level therapists with backgrounds in clinical psychology and mental health counseling delivered the standard PCIT treatment protocol (Eyberg & Funderburk, 2011) and received PCIT training and weekly supervision with a Certified PCIT Trainer. See Barnett et al.(2016) for detailed information regarding recruitment and training of the natural helpers. During the intake process, the natural helper attended one PCIT session to complete the warm hand-off process and to establish both the natural helper and the PCIT therapist as members of the same treatment team working together to support the family.

PCIT consisted of one-hour sessions conducted weekly where therapists taught caregivers parenting skills over the course of two phases of treatment: Child-Directed Interaction (CDI) and Parent-Directed Interaction (PDI). Parenting skills were taught to caregivers during two didactic sessions before starting coaching in each phase. During CDI coaching, the caregivers learned to use positive and differential social attention to improve child behavior and the parent-child relationship, whereas in PDI coaching, caregivers learned to set limits and consistently use discipline to increase child compliance to commands. Therapists coached caregivers on their parenting skills in vivo from behind a one-way mirror via a wireless headset. The length of PCIT varied as a function of caregiver skill acquisition and proficiency and child behavior change (i.e., child’s behavior falls to within normal limits [raw score ≤ 114], as measured by the ECBI), as caregivers were required to meet skill criteria to proceed from CDI to PDI and to complete treatment.

Natural helper sessions were scheduled in the family’s natural setting (e.g., home, community) for approximately 30 minutes each week after the CDI didactic session. During these sessions, natural helpers worked with families to practice parenting skills taught in PCIT and to identify and troubleshoot barriers to PCIT homework completion (e.g., toys being used, space in home, time of day) and treatment attendance. Specifically, during the CDI phase of treatment, the natural helpers observed and coded 5-minute caregiver-child interactions, provided brief feedback regarding the caregiver’s “most” and “least” used CDI skills, identified a target skill to practice and improve, and provided recommended strategies for improving that skill. Natural helpers subsequently engaged in role plays with the caregivers to practice the target skill. During the PDI phase of treatment, natural helpers continued to provide support around CDI skill acquisition as well as PDI homework support, including assistance with selection of an appropriate time out chair and time out room. Notably, however, Natural Helpers did not actively observe or offer feedback on parent PDI skills, simply because doing so might have placed Natural Helpers in a difficult position wherein they would have had to help parents physically move their children to the time out chair. To protect against liability and ensure treatment fidelity, we decided as a team such situations were better navigated in regular PCIT sessions with the licensed mental health provider.

2.5. Analytic Plan

Our analyses proceeded in a series of iterative steps in accordance with best practices (Kline, 2011; McDonald, 2014). First, we investigated whether our PCIT+NH model was truly reaching and engaging families not typically served by our clinic. To do so, we compared our PCIT+NH sample’s demographic characteristics (i.e., child age, gender, race, and ethnicity; parent race, ethnicity, education and income; language of service delivery) and treatment engagement (e.g., attendance, CDI and PDI homework completion) and retention (e.g., completion of phase of treatment and completion of treatment) to families typically and historically seen in the clinic. For continuous dependent variable comparisons, we conducted one-sample t-tests (to examine whether the PCIT+NH mean differed significantly from the mean for the overall population seen at our clinic; McDonald, 2014). Similarly, for dichotomous dependent variable comparisons, we conducted one-sample binomial tests, and for categorical dependent variable comparisons, we utilized chi-square goodness of fit tests (if expected values in each cell were five or higher) or Fisher’s exact test (if expected values in at least 20% of cells were < 5; McDonald, 2014). Additionally, we investigated whether the clinical outcomes (i.e., caregiver skill acquisition and stress reduction, child decreased externalizing and internalizing behaviors and increased compliance) experienced by families in our PCIT+NH sample differed from the clinical outcomes observed in our historic program population. We did so by conducting one-sample t-tests comparing mean outcomes in our PCIT+NH sample with historic averages we have seen in the population of families we serve (McDonald, 2014). As mentioned above, our goal with these analyses was to determine if a) the Natural Helper intervention allowed us to reach a new population of historically underserved families, b) whether the Natural Helper intervention helped this new population of families just as much as our typical in-clinic PCIT has helped our typical population of families historically. Consequently, comparing PCIT+NH families to those our clinic historically served was especially appropriate for answering these questions.

Second, we investigated whether participants served with the PCIT+NH model improved in clinical outcomes over time. We did so by conducting paired samples t-tests that examined whether changes from baseline to post-treatment on these measures were statistically significant (McDonald, 2014). In examining the initial effectiveness of PCIT+NH in this way, we methodologically align our work with that conducted by other PCIT investigators who have tested special PCIT programs in the past in open trial formats (e.g., Bagner et al., 2013; Graziano et al., 2015; Ros & Graziano, 2019).

Third and finally, we investigated whether the dose of natural helper support a family received (i.e., the number of natural helper sessions they attended) was related to the family’s treatment engagement, retention, or clinical outcomes. We utilized regression-based path analyses in MPlus to investigate this question (Kline, 2011). In these analyses, the treatment engagement, retention, and outcome variables were each predicted by the following independent variables: a) dosage of natural helper support (our main predictor of interest), b) child age and sex (covariates), and c) whether the family participated in the CFI intervention mentioned above in Assessments (to control for those intervention effects). Examining the question in this way ensured that effects of natural helper dosage emerged above and beyond baseline characteristics and behavior and aligns with best practice (De Los Reyes, 2017). Additionally, following best practice, full-information maximum likelihood estimators were invoked to handle any missing data (Kline, 2011).

2.5.1. Intention-to-Treat Analytic Framework

To determine how to best address missing data, treatment completers and non-completers were compared across several treatment engagement and clinical outcomes. Families who did not meet treatment completion criteria (43% of sample) did not differ from treatment completers (57% of sample) in rates of CDI homework completed, reduction of negative parenting skills (i.e., “Don’t” skills), effective command rate, correct follow-through rate, child compliance rate, caregiver stress, or child BASC-3-Externalizing or BASC-3-Internalizing scores. However, families who did not meet treatment completion criteria completed significantly less PDI homework (t(40) = −6.70, p < .01; M = 14.72% vs. M = 71.78%), demonstrated fewer positive parenting skills in their last session (t(25) = −2.51, p = .02; M = 28.5 vs. M = 38.76) and had higher child ECBI scores in their last session (t(38) = 3.88, p < .01; M = 124.6 vs. M = 73.46) compared to families that completed treatment. In other words, families who did not complete treatment still derived clear treatment gains that did not significantly differ from treatment completers on some measures of parenting skills, caregiver stress, and child externalizing and internalizing symptoms; however, non-completers had worse PDI phase engagement, exhibited fewer positive parenting skills, and experienced greater child disruptive behavior. To ensure that we accounted for these differences, we utilized the intention-to-treat analytic framework (Rothenberg et al., 2020) which included all available data from all subjects (regardless of treatment completion status) in study analyses. This ensured our results were not biased by only including treatment completers.

3. Results

3.1. PCIT+NH Sample versus Standard PCIT Sample

3.1.1. Access to PCIT

For these and all other findings, the reader is referred to the tables for detailed results, which will be summarized in the prose to avoid redundancy with the tables. First, we compared whether our PCIT+NH model was truly reaching and engaging families that were not typically served by our clinic. As seen in Table 1, a significantly greater portion of caregivers in our PCIT+NH sample identified as Black (23.81% vs. 13.09%) or an “Other” race (9.52% vs. 3.62%), whereas a significantly greater portion of the caregivers in our typical program population identify as White (59.52% vs. 73.82%). Representative of the population in the city where services were offered, both the PCIT+NH sample and the program population consist of a high proportion of Latinx families (comprising 66.67% of all families served in both clinics), not differing significantly from each other. Caregivers in our PCIT+NH sample were significantly more likely to have a high school diploma or less education (50% of our sample) than families in our program population (13.62% of our typical clinic population), whereas families in our program population were more likely to have a graduate degree. Additionally, parents in our PCIT+NH sample were significantly more likely to make less money, as 52.94% of the sample made less than $35,000 per year compared to 30% of families in our program population. It is important to note that this income data is limited, however, as only 40.48% of the PCIT+NH sample and only 43.60% of the historic comparison sample chose to disclose their income. Finally, compared to our program population, caregivers in our PCIT+NH sample were significantly more likely to prefer services delivered in Spanish (50% vs. 22.26%) or Creole (5.26% vs. 1.10%) as opposed to English (44% vs. 76.65%). No significant differences between PCIT+NH and our program population emerged for parent gender or child age, gender, race (despite significant differences in caregiver race noted above), or child/caregiver ethnicity.

3.1.2. Engagement and Retention in PCIT

Comparison between our historic program population and PCIT+NH sample on engagement and retention outcomes can be found in Table 2. Encouragingly, our PCIT+NH sample attended more total coaching sessions (M = 12.12 vs M = 7.93) and had a higher overall attendance rate (M = 76.76% vs. M = 52.48%) than our program population. Families attended an average of 12.1 treatment sessions (with 76.19% completing at least 8 sessions) and were in treatment for an average of 17.09 weeks. Families completed, on average, 5.64 natural helper sessions, with 76.19% of families completing ≥2 natural helper sessions. Completion of the CDI phase of treatment and CDI homework completion for the PCIT+NH sample were similar to rates for our program population. However, our PCIT+NH sample completed PDI homework at lower rates than our program population (47.07% vs. 72.02%) (Table 2). Rates of treatment completion were similar to rates in our program population. Of the 42 families who started receiving PCIT+NH, 24 (57%) completed all treatment goals, including meeting criteria for achieving expertise in all parenting skills and having their children score within typical normal limits on the ECBI (i.e., score of ≤114). Sixteen families dropped from the program before all these criteria were met.

Table 2.

Comparing Engagement, and Retention Outcomes in PCIT+NH Sample and Program Population

Variable	PCIT + NH Sample M (SD) or %	Program Population (SD) or %	t(df) or χ2(df)	p	d
Treatment Completion	57.14% Completed	68.21% Completed	None; Used Binomial Test	.17	−.26
CDI Completion	80.95% Completed	78.32% Completed	None; Used Binomial Test	.85	.09
Total Coach Sessions	12.12 (7.71)	7.93 (6.85)	3.52 (41)	<.01*	.60
Attendance Rate	76.76%	52.48%	6.05 (41)	<.01*	.60
CDI Homework Completion Rate	63.39%	66.18%	−0.67 (41)	.51	−.07
PDI Homework Completion Rate	47.07%	72.02%	−4.07(41)	<.01*	−.59

3.2. PCIT+NH Treatment Outcomes

Second, we investigated whether participants served with the PCIT+NH model showed improvements in clinical outcomes over the course of treatment.

As Table 3 reveals, caregivers improved on their parenting skills, including providing more positive parenting skills (“Do” skills) to their children, providing higher rates of effective commands and correct follow-through on those commands, and providing fewer negative parenting behaviors (“Don’t” skills) from baseline to post-treatment. Caregivers also significantly decreased in their levels of stress according to the PSI-SF-4 from baseline to post-treatment. Additionally, children demonstrated significantly fewer disruptive behaviors (as measured by the ECBI), fewer externalizing and internalizing symptoms (as measured by the BASC-3), and higher rates of compliance at post-treatment compared to baseline. These treatment outcomes appear promising, but one critical consideration is that only 24 families in the PCIT+NH group completed treatment. Therefore, power to detect significant effects in such a small sample should be considered. Fortunately, post-hoc power analyses indicate that the study was well-powered to find outcomes reported in Table 2. Specifically, power = .88 to detect the median effect size of the 9 reported in Table 2 (d = 1.34), meaning that there was an 88% chance that if a difference from baseline to post-treatment in PCIT+NH group scores was significant at p < .05, it would be detected.

Table 3.

Comparing Baseline and Post-Treatment Clinical Outcomes in PCIT+NH sample

Variable	PCIT + NH Sample M (SD)	Typical Clinic Families M (SD)	t(df)	p	d
Do Skills	4.41 (3.95)	36.48 (9.70)	−16.35 (26)	<.01 *	3.18
Don’t Skills	24.39 (16.90)	3.70 (3.68)	6.52 (26)	<.01 *	1.34
Effective Command Rate	40.29% (26.88%)	71.32% (30.90%)	−4.67 (26)	<.01 *	0.95
Correct Follow Through Rate	1.72% (9.28%)	75.04% (30.38%)	−13.07 (22)	<.01 *	2.57
Child Compliance Rate	23.57% (24.96%)	78.17% (19.72%)	−3.06 (22)	<.01 *	2.08
ECBI	160.03 (28.37)	93.90 (47.62)	9.63 (39)	<.01 *	1.52
BASC-Externalizing	60.53 (11.99)	46.65 (10.45)	4.73 (22)	<.01 *	1.12
BASC-Internalizing	57.31 (13.26)	46.83 (9.75)	4.96 (22)	<.01 *	0.94
PSI	60.94 (25.46)	35.08 (28.61)	5.31 (25)	<.01 *	1.12

^*p<.05

Additionally, we examined whether clinical outcomes seen in our PCIT+NH sample differed from those we see in our program population (Table 4). Overall, it appears that our PCIT+NH model delivered clinical outcomes on par with those seen in our typical clinic setting. With regards to post-treatment parenting skills, parents in the PCIT+NH sample demonstrated similarly low levels of negative parenting skills (“Don’t” skills), comparable high levels of effective commands, and similar rates of correct follow-through on effective commands compared to our program population. Moreover, the PCIT+NH sample demonstrated significantly higher positive parenting skills (“Do” skills) at post-treatment compared to our program population. The PCIT+NH sample also experienced similar improvements in parenting stress at post-treatment compared to our program population. With regards to child outcomes, children in the PCIT+NH sample achieved comparably low levels of disruptive behavior (measured by the ECBI) and internalizing behavior as our program population. The average child disruptive behavior score at the end of treatment on the ECBI was 93.9 (well within normal developmental limits, and with 75% of families scoring below the 114 threshold). Though children in our PCIT+NH sample experienced significantly less externalizing behavior post- treatment compared to our program population, they also experienced less externalizing behavior at baseline (Table 4). Thus, it ultimately appears that the two groups experienced similar improvements in externalizing behavior (Table 4). Children in the PCIT+NH sample also complied at significantly lower rates than children in our program population before the start of treatment (23.57% compliance versus 60.70% compliance; Table 4). However, levels of compliance at post-treatment in the PCIT+NH sample were indistinguishable from those in our program population (78.17% vs. 80.77%; Table 4).

Table 4.

Comparing Clinical Outcomes in PCIT+NH Sample and Program Population

Variable	PCIT + NH Sample M (SD)	Typical Clinic Families M (SD)	t(df)	p	d
Baseline Do Skills	4.41 (3.95)	5.34 (5.58)	−1.40 (35)	.17	−.17
Post-Tx Do Skills	36.48 (9.70)	27.99 (12.41)	4.55 (26)	< .01 *	.70
Baseline Don’t Skills	24.39 (16.90)	29.35 (17.35)	−1.76 (35)	.09	−.29
Post-Tx Don’t Skills	3.70 (3.68)	4.88 (6.63)	−1.67 (26)	.11	−.19
Baseline Effective Command Rate	40.29% (26.88%)	20.41% (15.82%)	4.43 (35)	<.01 *	1.15
Post Effective Command Rate	71.32% (30.90%)	59.40% (28.31%)	2.00 (26)	.06	.42
Baseline Correct Follow Through Rate	1.72% (9.28%)	5.60% (20.53%)	−2.25 (28)	.03 *	−.20
Post Correct Follow Through Rate	75.04% (30.38%)	64.82% (34.14%)	1.75 (26)	.09	.30
Baseline Child Compliance Rate	23.57% (24.96%)	60.70% (38.31%)	−6.82 (20)	<.01 *	−.99
Post Child Compliance Rate	78.17% (19.72%)	80.77% (29.26%)	−0.54 (16)	.59	−.09
Baseline ECBI	160.03 (28.37)	147.89 (31.45)	2.71 (39)	<.01 *	.39
Post ECBI	93.90 (47.62)	90.46 (30.91)	0.46 (39)	.65	.10
Baseline BASC Externalizing	60.53 (11.99)	64.77 (12.61)	−2.12 (35)	.04 *	−.34
Post BASC-Externalizing	46.65 (10.45)	52.31 (9.67)	−2.60 (22)	.02 *	−.58
Baseline BASC-Internalizing	57.31 (13.26)	57.12 (12.05)	0.08 (35)	.93	.02
Post BASC-Internalizing	46.83 (9.75)	49.31 (9.32)	−1.22 (22)	.24	−.26
Baseline PSI	60.94 (25.46)	67.94 (21.77)	−1.65 (35)	.11	−.32
Post PSI	35.08 (28.61)	44.80 (26.40)	−1.73 (25)	.10	−.37

^*p<.05, TX = Treatment

3.3. Dose of Natural Helper Support and PCIT Outcomes

Lastly, we also examined whether the dose of natural helper support a family received (i.e., the number of natural helper sessions they attended) was related to their treatment engagement, retention, or clinical outcomes (Table 5). Results indicated that higher doses of natural helper support were associated with increased family engagement and retention but not with significant improvement in clinical outcomes. Specifically, higher doses of natural helper support were also associated with a higher total number of coach sessions attended (but not overall attendance rate), and higher completion of PDI homework (but not CDI homework). Further, higher doses of natural helper support significantly increased the odds that a family completed both the CDI phase of treatment and treatment as a whole. However, higher doses of natural helper support did not significantly improve parenting skills, caregiver stress, or child outcomes.

Table 5.

Effects of NH dosage on engagement, retention, and clinical outcomes in PCIT-t-NH sample.

	Treatment Completion OR (SE)	CDI Completion OR (SE)	Total Coach Sessions β (SE)	% of Sessions Attended β (SE)	CDI Homework Completion β (SE)	PDI Homework Completion β (SE)	Do Skills β (SE)	Don’s Skills β (SE)	Effective Command Rate β (SE)	Correct Follow Through Rate β (SE)	Child Compliance Rate β (SE)	ECB1 β (SE)	BASC Externalizing β (SE)	BASC Internalizing β (SE)	PSI β (SE)
Number of Natural Helper Sessions	1.38 (.16) *	2.45 (1.07) *	.43 (.14) *	.10 (.15)	.18 (.16)	.46 (.14) *	−.18 (.23)	−.07 (−21)	.08 (.23)	.08 (.22)	.35 (.19)	−.07 (.15)	.00 (.17)	−.11 (.19)	.10 (.20)
CF1 Condition	0.93 (.69)	0.94 (1.18)	−.05 (.14)	.28 (.14) *	.10 (.15)	− .09 (.15)	.12 (−19)	−.14 (−17)	.13 (.19)	−.09 (.19)	−.11 (.18)	.30 (.14) *	.28 (.15)	.20 (.17)	.01 (.17)
Child Age	1.01 (.22)	1.17 (0.38)	.24 (.14)	−.14 (.15)	−.04 (.15)	−.13 (.15)	−.17 (.21)	.22 (−17)	.00 (.20)	.13 (.20)	.21 (.19)	−.05 (.15)	.08 (.18)	.17 (.21)	.18 (.17)
Child Gender	1.17(1.02)	1.38(1.72)	.11 (.14)	−.34 (.14) *	—.24 (.15)	.09 (.15)	−.07 (−20)	.00 (−18)	.26 (.20)	−.02 (.20)	−.06 (.17)	.22 (.14)	.29 (.15) *	.17 (.18)	.25 (.15)
Score on Variable at Baseline	N/A	N/A	N/A	N/A	N/A	N/A	−.10 (−12)	.48 (.14) *	.19 (.14)	−.27 (.31)	.45 (.12) *	.22 (.09) *	.11 (.13)	.28 (.18)	.44 (.16) *

Note.

^*p<.05, N/A indicates variable does not exist and therefore was not calculated. OR = Odds Ratio, β = Standardized Parameter Estimate. SE = Standard Error.

4. Discussion

In light of the extensive need and limited access to mental health care faced by low-income families of color, the current study sought to respond to the call for innovative culturally and linguistically responsive solutions to better engage marginalized populations (Callejas et al., 2010; Barnett et al., 2018b) by examining the initial effectiveness of a community-based natural helper PCIT model (PCIT+NH) to improve engagement, retention, and child and caregiver outcomes for low-income, Latinx and Black families. This research extends the literature focused on evidence-based strategies (e.g., psychoeducation about services, accessibility-promotion) to enhance different types of engagement among traditionally marginalized caregivers (Becker et al., 2018; Chacko et al., 2016; Haine-Schlagel & Walsh, 2015).

Consistent with our hypothesis, the integration of natural helpers into our PCIT program appears to have had a significant impact on access to and engagement with services for low-income families of color. Specifically, implementation of the PCIT+NH model achieved the goal of reaching caregivers who were more racially diverse, Spanish and Creole speaking, less educated, and lower income than the overall population typically served within our PCIT clinics (i.e., program population). These encouraging preliminary findings further support the role of natural helpers in extending program reach to families who have been historically underserved using traditional recruitment and service delivery means (Barnett et al., 2018b). Additionally, findings suggest families who were served with the PCIT+NH model are at least as engaged in treatment as our program population and may in fact be more engaged in treatment depending on the engagement measure investigated (e.g., attendance rate, total coach sessions). However, contrary to our hypothesis, PDI homework completion rates were lower than our program population. One possible explanation is focus of natural helper sessions. During sessions, natural helpers identify and troubleshoot barriers to CDI and PDI homework completion. However, natural helpers observed and coded parent-child interactions and provided feedback related to CDI skills, not PDI skills. Thus, it is possible that the lack of PDI skill practice during natural helper sessions could contribute to the lower PDI homework completion rates. Natural helper dosage appeared to at least partially contribute to this finding, as higher dosage of natural helper sessions was associated with greater attendance, engagement, and retention, according to multiple measures (Table 5). Although families who completed treatment were more likely to receive more natural helper services, the average number of natural helper sessions was much lower than the average number of PCIT sessions, despite being structured in a similar manner to PCIT sessions (weekly). These findings suggest there may be barriers to natural helper sessions despite flexible appointment times, attempts to reschedule, and in-home sessions, and highlight the need for additional research attention. Further research is needed to identify the appropriate dosage of natural helper sessions to support caregiver engagement without adding too many additional demands on the family’s time, and to identify additional strategies (such as offering telehealth visits) for increasing NH dosage received by families.

Findings also indicated that rates of completing the first phase of treatment and completing treatment as a whole were similar for families in the PCIT+NH sample as compared to rates for our program population. Further, treatment completion rate (57%) fares well and is higher than other community samples (where treatment completion rates for parenting programs hover between 30% and 40%; Lanier et al., 2011; Lyon & Budd, 2010; Pearl et al., 2012). These findings are especially encouraging given prior PCIT research documenting unique engagement and retention challenges faced by low-income families from ethnically and racially diverse backgrounds (Lanier et al., 2011; Lyon & Budd, 2010). Furthermore, 75% of families reported sub-clinical problems on the ECBI, suggesting clinically meaningful improvements even for families that did not complete the full course of treatment. Consistent with recent research on community implemented PCIT, these findings suggest that even families who “drop out” of treatment might receive a successful course of treatment (Lieneman et al., 2019). In sum, our preliminary findings suggest that involving natural helpers in recruitment and treatment may support PCIT service engagement for diverse and traditionally underserved families, which may relate to their social proximity to caregivers and ability to build trust and help navigate structural barriers to care (Barnett et al., 2018b; Gustafson et al., 2018). In terms of clinical improvements, PCIT+NH participants improved on all clinical outcomes over the course of treatment, and all of these improvements were large in their effect (i.e., all measures of Cohen’s d exceeded the .8 threshold designating a “large” effect; Cohen, 1988). Specifically, families who received the PCIT+NH model experienced clinically significant improvements in children’s externalizing and internalizing behaviors and compliance, as well as improvements in caregivers’ stress and observed parenting skills (i.e., increased Do skills, effective command rate, and correct follow-through rate, and decreased Don’t skills). Moreover, and consistent with our hypothesis, these rates of improvement were comparable to those seen in families typically served by our clinic. Notably, children in the PCIT+NH sample had lower rates of compliance than children in our program population at baseline. However, levels of compliance in the PCIT+NH sample were similar to those in our program population after participating in the intervention, suggesting that children in our PCIT+NH sample “catch-up” to, and show improvement in, their compliance over the course of treatment at rates similar to children typically seen in our clinic. These findings of comparable treatment outcomes are particularly notable given prior research showing mixed findings related to the additional benefit of supplemental treatment components (Kaminski et al, 2008; Thomas et al., 2017). Taken together with our engagement and retention findings, these results tentatively suggest that our PCIT+NH model allowed us to reach, engage, and retain lower income, traditionally underserved families, while also achieving clinical outcomes in those families at rates comparable to families typically seen in our clinic Consequently, the NH intervention shows promise as an equity-building intervention that allows families from traditionally underserved backgrounds to experience just as much benefit from PCIT as more affluent, educated, and predominantly White Latinx families that define our typical clinic population and the affluent, educated, and predominately White families that define the vast majority of families where behavioral parent training programs like PCIT are studied (Ortiz & Del Vecchio, 2013).

Lastly, with regard to natural helper dosage, findings indicated that higher doses of natural helper support were associated with increased family engagement and retention but not with significant improvement in clinical outcomes. Notably, this lack of significant findings on outcomes may be a consequence of the overall success of the intervention. Most families in the intervention achieved high rates of parenting skills, low caregiver stress, and improved child outcomes, so variability (and therefore ability to detect dosage responses) may be limited.

Despite the encouraging initial findings of the current study, we must acknowledge its limitations. First and foremost, the natural helper dosage sample utilized a within-group study design in which all families received both PCIT and natural helper support. Second, confounding variables cannot be ruled out in examining our dosage effects. It may be that families who are more likely to be engaged regardless of intervention format both participated in more natural helper sessions and were more likely to complete treatment. Third, the relatively small sample size limits generalizability of study findings and study findings need replication in a large sample. Fourth, caregiver-report data were only collected from the primary caregiver. Thus, future research should examine the difference across multiple informants. Fifth, the PCIT+NH model was implemented within a university community-based clinic setting. It is possible that findings may not generalize to traditional community settings where the same level of implementation support may not be feasible or sustainable. Thus, future work is needed to examine whether these findings extend to improve engagement in community mental health settings, where low-income families of color are frequently served. Sixth, we did not collect item-level data in the current study (we only had access to measure scale scores), so we were unable to generate internal consistency estimates for measures in the current study, though all measures are well-validated and found to be internally consistent in past studies, have been used regularly in PCIT studies, and are standard measures used as part of the PCIT intervention. Lastly, participants were only followed until the end of treatment; therefore, it is unclear whether gains made during treatment were maintained. Future work should include a longer follow-up period to draw more conclusions regarding the long-term efficacy of PCIT+NH. Despite the above-mentioned limitations, findings contribute to the existing literature on engagement and outcomes in evidence-based parenting programs, as low-income families of color are often understudied in psychosocial intervention research (Becker et al., 2018; La Greca et al., 2009).

4.1. Future Directions

While this study offers a first step to assessing the PCIT+NH model, a randomized controlled trial comparing PCIT versus PCIT+NH would provide further confidence in these findings and allow for in-depth examination of variables related to implementation. Specifically, random assignment to PCIT versus PCIT+NH would allow researchers to test the hypotheses that natural helper support could improve retention rates (reduce attrition), improve homework completion rates (associated with skill acquisition; Stokes et al., 2016), and perhaps reduce the duration of services required to reach skill proficiency, thereby enabling providers to serve more vulnerable families. Future work also should incorporate follow-up assessments, which would allow examination of treatment gains over time. Furthermore, work that examines whether the PCIT+NH is even more beneficial for specific subpopulations of parents or in diverse family constellations (e.g., single parent versus two parent families) would be especially valuable. Moreover, given the low dose of natural helper sessions relative to treatment sessions and the positive impact of dosage on treatment outcomes, strategies for increasing NH dosage during PCIT need to be explored (e.g., offering different formats of visits such as telehealth visits).

In addition to the need to enhance research related to the effectiveness of the PCIT+NH model, it is critical for research and practice to establish strategies to understand how to scale up this model and other lay health worker models. Though lay health workers show potential to enhance engagement and equity, systemic barriers to integrating lay health workers into children’s mental health services have been outlined. These include challenges financing their services, uncertainty about what levels of training and supervision are needed, and the need to develop measurement for what competent service delivery is (Barnett et al., 2021). Some large-scale efforts have begun to address these challenges, which could enhance the feasibility of disseminating PCIT+NH. For example, efforts have been made in New York State to train and establish competencies for family peer advocates and allow for these services to be billed to enhance caregiver engagement (Hoagwood et al., 2018; Rodriguez et al., 2011). While there have been growing efforts to increase large-scale funding for PCIT and improve access to PCIT certification via virtual training and consultation, more work needs to be done to address training-related barriers to large-scale implementation of PCIT for both PCIT Therapists and NHs. Additionally, efforts are needed to investigate if incorporating NHs could enhance implementation efforts with PCIT, as inadequate referrals and high levels of attrition have been associated with challenges for therapist certification and limited sustainability of the model (Barnett et al., 2019). Overall, this study provides some encouraging preliminary support for the use of natural helpers as a supplement to PCIT services. Our findings have important clinical and research implications, as they highlight a potential avenue for effectively recruiting diverse, low-income families historically underserved by evidence-based parenting programs and overcoming well-documented challenges with treatment accessibility. Beyond increasing initial access, the PCIT+NH model was also associated with increased family engagement and retention, and treatment outcomes were comparable to outcomes for more affluent and educated families receiving standard PCIT in the typical clinic setting. The current study represents an important initial step to better serving predominantly low-income families of color by reducing barriers to accessing and successfully completing PCIT services but warrants further examination to document treatment efficacy and sustainability.

Highlights.

• Natural helpers address disparities in access to behavioral parenting interventions

• Natural helpers support recruitment and engagement of low-income families of color

• Comparable clinical outcomes were achieved with natural helper support