Telehealth Treatment of Behavior Problems in Young Children With Developmental Delay

This randomized clinical trial investigates the effects of an internet-delivered parent-child interaction therapy for children aged 3 years with developmental delay.

Key Points

Question

What are the effects of a telehealth format of parent-child interaction therapy (PCIT), ie, internet-delivered PCIT (iPCIT), for children aged 3 years with developmental delay?

Findings

In this randomized clinical trial that included 150 children with developmental delay and their primary caregiver, iPCIT led to improvements in child behavior that were maintained across follow-up visits.

Meaning

Against a backdrop of considerable barriers to mental health services, these findings show the promise of telehealth formats for expanding the reach and scope of care for underserved children with developmental delay and their families.

Abstract

Importance

Early behavior problems in children with developmental delay (DD) are prevalent and impairing, but service barriers persist. Controlled studies examining telehealth approaches are limited, particularly for children with DD.

Objective

To evaluate the efficacy of a telehealth parenting intervention for behavior problems in young children with DD.

Design, Setting, and Participants

A randomized clinical trial was conducted from March 17, 2016, to December 15, 2020, in which children with DD and externalizing behavior problems were recruited from early intervention and randomly assigned to a telehealth parenting intervention or control group and evaluated through a 12-month follow-up. Most children were from ethnic or racial minoritized backgrounds. Over one-half of children were in extreme poverty or low income-need ratio categories.

Interventions

Internet-delivered parent-child interaction therapy (iPCIT), which leverages videoconferencing to provide live coaching of home-based caregiver-child interactions. Families received 20 weeks of iPCIT (provided in English or in Spanish) or referrals as usual (RAU).

Main Outcomes and Measures

Observational and caregiver-report measures of child and caregiver behaviors and caregiving stress were examined at preintervention, midtreatment, and postintervention and at 6- and 12-month follow-ups.

Results

The sample included a total of 150 children (mean [SD] age, 36.2 [1.0] months; 111 male children [74%]) and their caregivers with 75 each randomly assigned to iPCIT or RAU groups. Children receiving iPCIT relative to RAU displayed significantly lower levels of externalizing problems (postintervention Cohen d = 0.48; 6-month Cohen d = 0.49; 12-month Cohen d = 0.50) and significantly higher levels of compliance to caregiver direction after treatment. Of those children with data at postintervention, greater clinically significant change was observed at postintervention for children in the iPCIT group (50 [74%]) than for those in the RAU group (30 [42%]), which was maintained at the 6-month but not the 12-month follow-up. iPCIT did not outperform RAU in reducing caregiving stress, but caregivers receiving iPCIT, relative to RAU, showed steeper increases in proportion of observed positive parenting skills (postintervention odds ratio [OR], 1.10; 95% CI, 0.53-2.21; 6-month OR, 1.31; 95% CI, 0.61-2.55; 12-month OR, 1.64; 95% CI, 0.70-3.07) and sharper decreases in proportion of observed controlling/critical behaviors (postintervention OR, 1.40; 95% CI, 0.61-1.52; 6-month OR, 1.72; 95% CI, 0.58-1.46; 12-month OR, 2.23; 95% CI, 0.53-1.37). After treatment, iPCIT caregivers also self-reported steeper decreases in harsh and inconsistent discipline than did than RAU caregivers (postintervention Cohen d = 0.24; 6-month Cohen d = 0.26; 12-month Cohen d = 0.27).

Conclusions and Relevance

Results of this randomized clinical trial provide evidence that a telehealth-delivered parenting intervention with real-time therapist coaching led to significant and maintained improvements for young children with DD and their caregivers. Findings underscore the promise of telehealth formats for expanding scope and reach of care for underserved families.

Trial Registration

ClinicalTrials.gov Identifier: NCT03260816

Introduction

Behavior problems in young children with developmental delay (DD) represent a significant public health concern. Roughly 13% of young children have a DD,¹ and over one-half of these children have at least 1 mental health disorder.² Behavioral challenges are particularly common, as children with DD show 3 times the risk of clinically significant externalizing problems relative to typically developing children.³ Externalizing problems in children with DD are associated with a wide range of functional impairments⁴ and result in significant economic costs when left untreated.⁵ Caregivers of children with DD experience high levels of caregiving stress⁶ associated with the severity of child behavior problems.⁷ Accordingly, effective treatment of behavior problems in young children with DD is critical.

In the US, early intervention (EI), via Part C of the Individuals With Disabilities Education Act, provides services to children aged 0 to 3 years with DD. However, EI focuses on child delay(s)⁸ rather than associated behavior problems and caregiving stress that predict functional outcomes.⁹ Caregivers completing EI report less confidence managing their child’s behavior than meeting their child’s developmental needs.¹⁰ Research demonstrates efficacy of clinic-based parenting interventions for treating behavior problems in children with DD.^11,12,13,14 For example, parent-child interaction therapy (PCIT)¹⁵ draws on attachment and social learning theories, with efficacy supported across multiple studies of youth with DD and autism spectrum disorder.^{11,12,16,17,18,19} A distinguishing feature of clinic-based PCIT is its use of in-session caregiver coaching via a 1-way mirror and wireless caregiver-worn earpiece.

Despite support for PCIT and other clinic-based parenting interventions, considerable barriers interfere with accessibility and acceptability of children’s mental health services, including transportation obstacles, regional clinician shortages, limited availability of non-English language services, and stigma-related concerns about visiting a clinic.^20,21 Youth from marginalized, low-income, and/or remote communities are particularly underserved.²² Among children with DD who received clinic-based treatment for behavior, those living in poverty and from ethnic or racial minoritized backgrounds have lower session attendance and higher dropout rates.²³ In EI, Spanish-speaking families and families living in poverty experience delayed or inadequate services,²⁴ and EI services abruptly end and transition to school-based services after the child’s third birthday.^25,26

Rapidly developing technology, broadening internet availability, and increasingly sophisticated capabilities for live home-based broadcasting have transformed behavioral health delivery and can overcome issues of geography, transportation, and stigma.²⁷ Telehealth strategies are not new, although when COVID-19 stay-at-home guidelines put traditional clinic-based mental health care further out of reach for families,²⁸ telehealth practices entered the clinical mainstream.^29,30 Despite increasing reliance on telehealth strategies in mental health care, randomized clinical trials (RCTs) evaluating telehealth strategies with families of young children are limited.

Research shows initial support for internet-delivered PCIT (iPCIT) in broadening reach for treating typically developing youth.³⁰ Given the importance of in-session caregiver coaching,³¹ iPCIT therapists use videoconferencing to watch family interactions and provide in-the-moment feedback to caregivers via a wireless caregiver-worn earpiece. In an initial RCT³² comparing iPCIT vs clinic-based PCIT for young typically developing children with disruptive behavior, outcomes were comparable, except that caregivers receiving iPCIT reported fewer barriers to care, and more children receiving iPCIT were rated by masked evaluators as having an excellent response compared with those receiving clinic-based PCIT. Findings suggest iPCIT can reduce barriers and may offer incremental benefits over clinic-based PCIT. However, this preliminary iPCIT evaluation excluded children with DD and caregivers who did not speak English, studied a sample of mostly non-Hispanic or non-Latino White families, and was not powered to test broad clinical utility.

We conducted, to our knowledge, the first RCT of a telehealth-delivered parenting intervention for young children with DD and present the main outcomes herein. Participants were from predominantly economically disadvantaged and underrepresented ethnic or racial minoritized backgrounds. In a 2-armed RCT, families were randomly assigned to receive iPCIT or referrals as usual (RAU) for 20 weeks after ending EI services at age 3 years and evaluated across a 12-month follow-up period. For primary outcomes, we hypothesized iPCIT, relative to RAU, would be associated with significantly reduced child externalizing problems, higher levels of observed child compliance with caregiver direction, and a higher rate of children showing clinically significant change. For secondary outcomes, we expected caregivers receiving iPCIT, relative to RAU, would show increases in positive parenting skills and decreases in controlling/critical behaviors during play with their child, as well as decreases in harsh and inconsistent discipline and caregiving stress.

Methods

Participants and Study Design

The Florida International University institutional review board approved this study and oversaw all procedures (Supplement 1). Families were recruited between March 17, 2016, and August 28, 2019, at 3 Part C EI sites in South Florida. Recruitment occurred during the child’s EI exit evaluation within 3 months of the child’s third birthday. Inclusion criteria among these youth with DD were (1) Child Behavior Checklist (CBCL)³³ externalizing problems T score greater than or equal to 60 and (2) the primary caregiver speaks English or Spanish. Exclusion criteria were as follows: (1) child receiving psychiatric medication for behavior problems; (2) child/caregiver deafness or blindness; (3) severe child social communication deficits (ie, caregiver report on Social Responsiveness Scale, second edition,³⁴ T score >75), although children with moderate social communication deficits (scores between 66-75), including those with autism spectrum disorder, were eligible; and (4) primary caregiver standard score less than 4 on the vocabulary subtest of the Wechsler Abbreviated Scale of Intelligence, second edition³⁵ (for English speakers) or Escala de Inteligencia Wechsler Para Adultos, third edition³⁶ (for Spanish speakers). This study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guidelines.

Families meeting study eligibility were invited to participate. After providing informed written consent, families completed a baseline assessment. Eligible families were randomly assigned 1:1 (stratified by child sex) to up to 20 weeks of either iPCIT or RAU by the masked study statistician. A total of 124 families (62 per condition) was determined sufficient to detect a medium effect with 80% power at α less than .05. Major assessments, consisting of caregiver reports and observational tasks of family interactions, were conducted in the family’s home at baseline and week 20 (posttreatment/RAU) and again at 6- and 12-month follow-ups. An abbreviated caregiver-report assessment was collected at week 8 (ie, midtreatment). Consistent with other studies of parenting interventions^37,38 and ethical guidelines for research participant payment,³⁹ families received $100 for each major assessment ($50 at midtreatment) and a tablet (valued at $50) at study completion. All participants completed the treatment phase before the COVID-19 pandemic. The study flow and retention rates were consistent with other studies conducting home-based assessments.⁴⁰ Participants from the following race and ethnic groups were included in the study: Asian, Black or African American, Hispanic or Latino, non-Hispanic or non-Latino White, and other, which included biracial individuals. Race and ethnicity were identified by caregiver report.

Treatment

iPCIT draws on encrypted videoconferencing technology in which therapists provide live coaching of caregiver-child interactions via webcam and a caregiver-worn earpiece. As in clinic-based PCIT, iPCIT progresses through child-directed interaction and parent-directed interaction phases. During child-directed interaction, caregivers learn to follow their child’s lead in play by using PRIDE skills (ie, praising child behavior, reflecting child statements, imitating child play, describing child actions, and showing enjoyment) and avoiding questions, commands, and criticisms. They learn to use PRIDE skills in response to appropriate child behaviors and ignore undesirable behaviors. During parent-directed interaction, caregivers learn to use effective commands and consistently follow through with timeouts to increase child compliance. Families received a tablet (and data plan for families without a wireless network) and wireless earpiece for treatment. Sessions were conducted weekly by a remote therapist and lasted 1 to 1.5 hours. Approximately one-half of iPCIT-treated families received treatment in Spanish. Therapists were psychology doctoral students or postdoctoral fellows, who completed initial PCIT training before working on this project, and received ongoing supervision and consultation from the first author, a PCIT International Global Trainer.

Measures

Primary Child Behavior Outcomes

The CBCL for ages 1.5 to 5 years is a 99-item caregiver-rating scale that measures the frequency of children’s behavioral and emotional problems, showing excellent psychometric properties,³³ including in RCTs of interventions for children with DD.^11,14 The externalizing T score was used to measure caregiver-reported child externalizing problems at all assessments (α = .90).

The Dyadic Parent-Child Interaction Coding System, fourth edition (DPICS) is a well-supported observational coding system of caregiver and child behaviors during family interactions.⁴¹ Coders masked to condition were trained to 80% agreement with a criterion tape, and 25% of baseline observations were double-coded for reliability (mean κ = 0.83). Child compliance, defined as the proportion of caregiver commands with which the child complied, was coded during a 5-minute clean-up task at each major assessment. For child compliance scores in DPICS, the range was from 0 to 100 because the scores represent percentage compliance.

Secondary Caregiver Behavior Outcomes

The DPICS was also used to observe caregiver behaviors during family interactions. “Do” skills during play refer to praises, behavior descriptions, and reflections, whereas “don’t” skills refer to questions, commands, and criticisms. The proportion of caregiver do and don’t skills was coded to reflect positive vs controlling/critical parenting behaviors during a 5-minute child-led play task completed at each major assessment. The do and don’t skills in DPICS are proportions, therefore, the range of those scores are from 0 to 100.

The Harsh and Inconsistent Discipline subscale of the Parenting Practices Inventory (PPI)⁴² measured caregiver-reported discipline at each major assessment and showed strong psychometric properties in caregivers of young children (α = .83).

The Negative Feelings Toward Parenting subscale of the Family Impact Questionnaire (FIQ)⁴³ measured caregiving stress and strain at all assessments and has been used extensively with caregivers of children with DD⁴⁴ (α = .70).

Treatment Satisfaction

The Therapy Attitude Inventory (TAI)⁴⁵ and Client Satisfaction Questionnaire 8 (CSQ-8)⁴⁶ assessed caregiver satisfaction with iPCIT at postintervention. Both measures have shown strong psychometric properties^47,48 (TAI α = 0.95; CSQ α = 0.94).

Statistical Analysis

Mixed-effects regression models analyzed the effects of time, treatment condition, and their interaction as predictors of outcome trajectories.⁴⁹ We treated time as a continuous predictor (weeks since baseline) consistent with recommendations to decrease the likelihood of type I error.⁵⁰ For continuous outcomes (CBCL, PPI, FIQ), linear mixed models were used; for proportions (compliance, do/don’t skills), generalized linear (logistic) mixed models were used. For CBCL and PPI models, the natural log of weeks since baseline was used as the time predictor to linearize associations between time and outcomes; this was based on comparing the Akaike information criterion and Bayesian information criterion from this model to the linear time model.⁵¹ Separate mixed-effects models were run for each outcome. Significant condition-by-time interactions were followed up with simple slopes probing to clarify the direction of effects for each group. We evaluated clinical significance by creating a dichotomous variable as to whether the CBCL externalizing T score was below or above the clinical cutoff of 60 at postintervention and follow-up. To compare rates of children moving out of the clinical range of symptoms across conditions, χ² tests were used.

Mixed models provide unbiased estimates when missing values are at least missing at random.⁵² Missing value analysis showed that missingness on variables at later waves was associated with child sex, child age, language of assessment, and BDI score; these variables were included as covariates in all models (child age and BDI were mean centered). All P values were 2-sided, and the cutoff for significance was set at P <.05. Statistical analyses were performed using the lme4 package in R, version 4.2.0 (R Core Team).

Results

Descriptive Findings

The sample included a total of 150 children (mean [SD] age, 36.2 [1.0] months; 111 male children [74%]; 39 female children [26%]) and their caregivers. Children from the following race and ethnicity groups were included: 5 Asian (3.4%), 35 Black or African American (23.5%), 105 Hispanic or Latino (70.5%), 10 non-Hispanic or non-Latino White (6.7%), and 2 other (1.3%). All families completed the baseline assessment and were included in all analyses. Figure 1 presents study flow and retention rates.

Figure 1. Consolidated Standards of Reporting Trials (CONSORT) Flow Diagram.

CBCL indicates Child Behavior Checklist; EIWA, Escala de Inteligencia Wechsler Para Adultos; iPCIT, internet-delivered parent-child interaction therapy; RAU, referrals as usual; SRS, Social Responsiveness Scale; WASI, Wechsler Abbreviated Scale of Intelligence.

Table 1 presents sample baseline characteristics. Most primary caregivers identified themselves (135 of 149 [91%]) and their child (140 of 149 [94%]) as being from an ethnic or racial minoritized group (1 family did not report race or ethnicity data). Approximately one-half of primary caregivers reported speaking multiple languages (60 of 145 [48%]). Based on income-need ratios (household income divided by Federal Poverty Threshold),⁵³ over one-half of the sample was classified in extreme poverty or low-income categories (78 of 137 [57%]). Most caregivers were biological/adoptive mothers (137 [91%]); 12 (8%) were biological/adoptive fathers, and 1 (1%) was a biological grandmother. Mean developmental functioning across the sample (as determined by the Battelle Development Inventory, second edition⁵⁴ administered at the EI exit evaluation) was in the low average range, and mean CBCL externalizing T scores were in the clinical range. No baseline group differences were found.

Table 1. Baseline Demographic Variables Across Conditions.

Characteristic	Treatment condition, No. (%)
	Total (n = 150)	iPCIT group (n = 75)	RAU group (n = 75)
Child sex
Male	111 (74.0)	54 (72.0)	57 (76.0)
Female	39 (26.0)	21 (28.0)	18 (24.0)
Child age, mean (SD), mo	36.20 (1.0)	36.30 (1.3)	36.10 (0.6)
Child ethnicity and racea
Asian	5 (3.4)	1 (1.4)	4 (5.3)
Black/African American	35 (23.5)	14 (18.9)	21 (28.0)
Hispanic or Latino	105 (70.5)	58 (78.4)	47 (62.7)
Non-Hispanic or non-Latino White	10 (6.7)	3 (4.1)	7 (9.3)
Otherb	2 (1.3)	0 (0)	2 (2.7)
Primary caregiver ethnicity and racea
Asian	5 (3.4)	1 (1.4)	4 (5.3)
Black/African American	35 (13.5)	14 (18.9)	21 (28.0)
Hispanic or Latino	99 (66.4)	53 (71.6)	46 (61.3)
Non-Hispanic or non-Latino White	15 (10.1)	7 (9.5)	8 (10.7)
Otherb	4 (2.7)	1 (1.4)	3 (4.0)
Primary caregiver age, mean (SD), y	34.49 (6.3)	34.30 (6.1)	34.68 (6.5)
Primary caregiver education (highest)c
Did not complete high school	14 (9.6)	7 (9.7)	7 (9.5)
Graduated high school	60 (41.1)	27 (37.5)	33 (44.6)
Graduated college	49 (33.6)	25 (34.7)	24 (32.4)
Completed graduate degree	23 (15.4)	13 (18.1)	10 (13.5)
Income-to-needs classificationd
Extreme poverty/poor	42 (30.7)	23 (33.8)	19 (27.5)
Low income	36 (26.3)	21 (30.9)	15 (21.7)
Adequate income	32 (23.4)	12 (17.7)	20 (30.0)
Affluent	27 (19.7)	12 (17.7)	15 (21.7)
Primary caregiver languagee
English and Spanish	58 (40.0)	25 (34.7)	33 (45.2)
English only	41 (28.3)	20 (27.8)	21 (28.8)
Spanish only	35 (24.1)	22 (30.6)	33 (45.2)
English and other (eg, Creole)	11 (7.59)	5 (6.9)	6 (8.2)
CBCL Externalizing T score, mean (SD)	68.52 (6.9)	67.75 (6.6)	69.29 (7.2)
BDI developmental quotient, mean (SD)	74.74 (10.0)	75.23 (9.6)	74.26 (10.4)

Abbreviations: BDI, Battelle Developmental Inventory; CBCL, Child Behavior Checklist; iPCIT, internet-delivered parent-child interaction therapy; RAU, referrals as usual.

^a Based on 149 primary caregivers who reported ethnic or racial identity data. Percentages add up to more than 100% as categories are not mutually exclusive.

^b Other race or ethnicity indicates biracial.

^c Based on n = 146 of primary caregivers who reported education data.

^d Based on n = 137 who reported income and dependents in household to calculate the income to needs ratio based on Federal Poverty Threshold.

^e Based on n = 145 of primary caregivers who reported language(s) spoken at home.

On average, families assigned to iPCIT completed 9.44 sessions (range, 0-19 sessions) in 20 weeks; 55 (73%) completed at least 5 sessions, and 42 (56%) completed at least 10 sessions. A total of 147 sessions (21%) were randomly selected and coded for fidelity, defined as the percentage of session elements detailed in the PCIT manual to which the therapist adhered. Fidelity was high across sessions (mean = 86%). At posttreatment, iPCIT-treated caregivers reported high satisfaction with treatment on the TAI (mean = 44.45 of 50) and the CSQ-8 (mean = 28.74 of 32). Among families in the RAU group, 44 of 71 (62%) received services for their child between baseline and postassessments, predominately from speech (34 of 71 [48%]) or occupational (27 of 71 [38%]) therapists.

Child Behavior Outcomes

Table 2 presents model-implied means and details of regression models predicting changes in child behavior outcomes. Caregiver report of child externalizing problems on the CBCL had a significant condition-by-time interaction term, indicating the conditions had different effects on child externalizing problems across time. Specifically, caregiver report of child externalizing problems significantly decreased over time for both conditions but decreased more steeply in response to iPCIT (between-conditions postintervention Cohen d = 0.48; 6-month follow-up d = 0.49; 12-month follow-up d = 0.50). These steeper iPCIT declines in externalizing problems unfolded during the acute treatment phase, after which gains were largely maintained across the 12-month follow-up (Figure 2). Significantly more children in the iPCIT-treated group than in the RAU group demonstrated clinically significant change in externalizing problems at postintervention (50 of 68 [74%] vs 30 of 71 [42%]; P < .001) and 6-month follow-up (45 of 62 [73%] vs 31 of 69 [45%]; P = .002), but not at 12-month follow-up (42 of 62 [68%] vs 46 of 71 [65%]; P = .85).

Table 2. Model-Implied Means Across Conditions and Results of Mixed Models^a.

Outcome	Model-implied means
	Baseline	Midintervention	Postintervention	6-mo Follow-up	12-mo Follow-up	Time	Condition	Condition by time interaction
Primary outcomes: child
Externalizing problemsb
iPCIT	61.18	54.43	53.82	53.43	53.15	−0.60c	4.97d	0.23e
RAU	64.05	59.87	59.49	59.25	59.08
Observed proportion compliancef
iPCIT	0.12	NA	0.15	0.19	0.23	0.01c	0.26	−0.01c
RAU	0.16	NA	0.17	0.18	0.20
Secondary outcomes: caregiver
Observed proportion do skillsf
iPCIT	0.11	NA	0.12	0.13	0.14	0.01c	0.08	−0.01c
RAU	0.12	NA	0.11	0.10	0.09
Observed proportion don’t skillsf
iPCIT	0.45	NA	0.40	0.36	0.31	−0.01c	0.11	0.01c
RAU	0.47	NA	0.48	0.49	0.50
Harsh/inconsistent disciplineb
iPCIT	3.05	NA	2.57	2.54	2.52	−0.04c	0.14	0.02d
RAU	2.99	NA	2.77	2.76	2.75
Caregiving stressg
iPCIT	11.10	10.72	10.05	9.09	7.89	−0.04c	0.95	0.02
RAU	12.05	11.80	11.35	10.72	9.92

Abbreviations: BDI, Battelle Developmental Inventory; iPCIT, internet-delivered parent-child interaction therapy; NA, not available; RAU, referrals as usual.

^a All models include child sex, age, language, and BDI score as covariates; BDI and child age were mean centered. Slopes reflect change in outcome per week. Means are calculated at the mean week for each assessment. Mean timing of post = 22.10 weeks after baseline; mean timing of 6-month follow-up = 42.34 weeks after baseline; mean timing of 12-month follow-up = 67.52 weeks after baseline.

^b Log transformation of week model. Slopes reflect change in outcome variable per log (week). Externalizing problems measured via Child Behavior Checklist. Harsh and inconsistent discipline measured via Parenting Practices Inventory.

^c P < .001

^d P < .01

^e P < .05

^f Logistic mixed model. Slopes reflect change in logit of outcome per week. Observed child compliance and observed caregiver do and don’t skills measured via Dyadic Parent-Child Interaction Coding System.

^g Linear mixed model. Slopes reflect change in outcome variable per week. Caregiving stress measured via Family Impact Questionnaire.

Figure 2. Changes in Child Externalizing Problems Across Conditions.

The solid lines represent the growth curve or slope of the change in the Child Behavior Checklist (CBCL) in each group. iPCIT indicates internet-delivered parent-child interaction therapy; RAU, referrals as usual.

Observed child compliance with caregiver commands during the cleanup task also had a significant interaction term (between-conditions postintervention odds ratio [OR], 0.93; 95% CI, 0.51-1.71; 6-month OR, 1.01; 95% CI, 0.57-1.80; 12-month follow-up OR, 1.11; 95% CI, 0.64-1.91). Specifically, the rate of child compliance roughly doubled by 12-month follow-up among children in the iPCIT-treated group but only increased by approximately one-third for children in the RAU group.

Caregiving Outcomes

Table 2 also presents model-implied means and details of regression models predicting changes in caregiving outcomes. Observed do and don’t skills during a structured child-led play task had significant condition-by-time interaction terms, with steeper increases in proportion of observed positive parenting skills (postintervention OR, 1.10; 95% CI, 0.53-2.21; 6-month OR, 1.31; 95% CI, 0.61-2.55; 12-month OR, 1.64; 95% CI, 0.70-3.07) and sharper decreases in proportion of observed controlling/critical behaviors (postintervention OR, 1.40; 95% CI, 0.61-1.52; 6-month OR, 1.72; 95% CI, 0.58-1.46; 12-month OR, 2.23; 95% CI, 0.53-1.37). These results indicated that the conditions had different effects on caregiving behaviors across time. Caregiver do skills significantly increased and don’t skills significantly decreased over time among iPCIT caregivers, with the opposite pattern for RAU caregivers (Figure 3).

Figure 3. Changes in Observed Proportion of Do and Don’t Skills Across Conditions.

The top set of lines are the proportion of don't skills, and the bottom set of lines are the proportion of do skills. These are also growth curves of these measures for both groups. iPCIT indicates internet-delivered parent-child interaction therapy; RAU, referrals as usual.

Caregiver-reported use of harsh and inconsistent discipline also had a significant condition-by-time interaction term. Specifically, harsh and inconsistent discipline significantly decreased across conditions but decreased more steeply among iPCIT families (between-conditions postintervention Cohen d = 0.24; 6-month follow-up Cohen d = 0.26; 12-month follow-up Cohen d = 0.27). Caregiving stress decreased in both conditions and showed no significant condition-by-time interaction.

Discussion

Considerable barriers interfere with accessibility of clinic-based care for families of young children with DD and behavioral challenges. This study offers the first RCT, to our knowledge, that examined a telehealth-delivered parenting intervention for children with DD and found that iPCIT significantly outperformed usual referrals in reducing child behavior problems. Between-group effect sizes based on both caregiver report and structured observations were mostly in the moderate range and maintained across time. Findings are consistent with previous studies examining clinic-based PCIT for children with DD^{11,12,16,17,18,19} and telehealth strategies for typically developing youth.³² Moreover, iPCIT-treated children not only showed reductions in behavior problems, such as aggression, but demonstrated higher rates of following directions, which is especially important for children entering kindergarten.⁵⁵

It is encouraging that caregiver behaviors improved after iPCIT, as such behaviors are associated with the development of adaptive emotion regulation in children with DD,^56,57 which, in turn, predicts subsequent social skills.⁵⁸ It is also promising that iPCIT was able to reduce harsh and inconsistent discipline in this population, as such discipline predicted decreases in surface area and cortical thinning of brain regions involved with sensorimotor and social functioning⁵⁹ and in regions implicated in DD.⁶⁰ However, iPCIT did not outperform RAU in reducing caregiving stress. This result is not completely surprising given mixed findings on the impact of parenting interventions on stress in caregivers of children with DD.^11,12,17 It is worth noting that, unlike these previous studies, most families in this sample were from ethnic/racial minoritized and/or immigrant backgrounds. Given that acculturation and enculturation can differentially worsen or buffer stress in minoritized caregivers of children with DD,⁷ it may be important to consider how parenting interventions can incorporate cultural factors into care.^61,62

The present findings have important clinical implications. The study adds to a small but growing literature supporting the effects of telehealth approaches to children’s mental health care.^32,63 Telehealth strategies can overcome geographic disparities in care availability, logistical obstacles (eg, transportation, parking), and stigma-related concerns about visiting a mental health facility. Telehealth strategies can even improve the ecological validity of care by treating families in natural settings where problems occur most.⁶⁴ Integration of iPCIT into existing systems of care, such as EI, may have broad public health impact. For example, there are notable gaps in the transition from Part C to Part B services, especially among children living in poverty and from underserved ethnic/racial minoritized backgrounds.²⁶ When children with DD complete Part C services, iPCIT can provide a critical resource as children enter the school system.

Limitations

Several limitations in this study warrant comment. First, the present analysis focused on primary and secondary outcomes, and examination of other factors was beyond the scope of this report. Second, RAU reflected standard care for families exiting EI, but future comparison of iPCIT with an active control group (eg, clinic-based PCIT) would help clarify the differential impact of clinic-based vs telehealth delivery, control for potential dosage effects, and examine whether quantity and/or quality of services received impacts outcome. Third, although three-quarters of iPCIT-treated children reached normative levels of behavioral problems by posttreatment, by 1-year follow-up, these clinically significant gains were not better than the clinically significant gains in RAU children, suggesting that effects may wane over time. Fourth, future research should follow iPCIT effects beyond 1 year, as children with DD enter elementary school. Fifth, equipment was provided to families, therefore, differences in treatment response could not be attributed to differences in household technology. Some families may not own the needed equipment (eg, tablet), which reduces study generalizability. Accelerating growth in household internet connectivity is encouraging, but home-based telehealth may not be feasible for some resource-insecure households. Sixth, the study was a single-site study in 1 metropolitan area, and a relatively high number of participating caregivers completed college, thus limiting study generalizability.

Conclusions

Despite limitations, this was the first RCT, to our knowledge, to demonstrate positive effects of a telehealth-delivered parenting intervention for children with DD. The sample included mostly economically disadvantaged and ethnic/racial minoritized families, for whom DD and behavior problems are prevalent and treatment resistant²³ and are underserved by clinic-based care. Overall, findings underscore the value of leveraging technological innovations to overcome barriers to effective care for underserved families of young children with DD.

Supplement 1.

Trial Protocol

Supplement 2.

Data Sharing Statement