Parenting Stress, Child Social Functioning, and Part C Early Intervention in Predominantly Low-Income Families of Children With or at High Risk for Developmental Delay From Minoritized Racial and Ethnic Groups

Abstract

In this study, the relationships between child social functioning, parenting stress, and Part C Early Intervention (EI) enrollment were examined in 227 ethnically and racially diverse, low-income families of 15- to 27-month-old children. All toddlers in the sample were identified with or at high risk for developmental delay via universal screening in primary care; 41.4% were enrolled in EI at the time of study data collection and 83.3% ultimately enrolled in EI before turning 3 years old. Generalized linear mixed model results indicated no direct relationship between EI enrollment and parenting stress, but a significant inverse association was found between child social functioning and parenting stress (β = −0.61, p = .005) that was moderated by EI enrollment (p < .001). This suggests potential benefits of timely referral and access to EI for families of children at risk of developmental delay in historically underserved communities. Furthermore, exploratory bivariate analyses indicated that clinically elevated parenting stress related to higher worry about the child’s development and higher M-CHAT-R/F scores; EI enrollment was related to parents not working outside the home, older age of both parent and child, and lower child social functioning level.

Part C Early Intervention (EI), as defined in the Individuals with Disabilities Education Act (IDEA Part C, 2004), is the primary public source of services for children under 3 years old with or at risk for developmental delays in the United States. It is implemented on a broad scale across all 50 states, with approximately 3.5% of all children under 3 years old receiving services (U.S. Department of Education, 2020). In addition to providing direct services targeting children’s individualized developmental goals, current best practices for Part C EI services emphasize parent consultation and/or coaching for the purpose of improving parental capacity to support children’s development.

Part C EI is a widespread and seemingly important means of support for children with developmental delays and their families. However, a major challenge to rigorously studying the impact of Part C EI services on children and families is that random assignment to these services would be unethical; states are mandated to offer these services to all eligible families. Thus, in lieu of causal testing via the random assignment, researchers have assessed the effects of Part C EI by examining longitudinal changes among those receiving Part C EI services without a comparison group. Now that universal developmental and autism-specific screening has become a prevalent best practice in Pediatric Primary Care for toddlers (Lipkin et al., 2020), there is an opportunity to conduct innovative natural experiments that compare child and family outcomes among those with identified or suspected developmental delays who have versus have not already been engaged in the Part C EI service system at the time of universal screening. In this study, we employed this novel approach with a focus on associations with family-level outcomes, specifically in the area of parenting stress.

Family-Level Outcomes

Measurement and Evaluation of Family-Level Outcomes.

Prior studies have examined family-level outcomes mainly through surveying large samples of parents who participated in Part C EI, with a general focus on the five family outcomes recommended for state reporting by the Early Childhood Outcomes (ECO) Center. These include

(1) families understand their child’s strengths, abilities, and special needs, (2) families know their rights and advocate effectively, (3) families help their child develop and learn, (4) families have support systems, and (5) families access desired services, programs, and activities in their community. (Bailey et al., 2006; Early Childhood Technical Assistance [ECTA], 2021)

These five ECO Center-recommended family-level outcomes are thought to contribute to the broader overarching goal of supporting family quality of life (Bailey et al., 2006; Epley et al., 2010; Wainer et al., 2017).

Through the seminal National Early Intervention Longitudinal Study (NEILS), Hebbeler et al. (2007) found that 59% of parents reported that their families were “much better off as a result of the help and information provided” through Part C EI. Using this same dataset, Bailey et al. (2005) found that parents reported feeling competent in caring, advocating, and gaining access to services for their children upon completion of Part C EI services. Raspa et al. (2010)’s analysis of large-scale Family Outcomes Survey data revealed that race and income as well as time spent in Part C EI predicted parents feeling confident and competent in parenting a child with special needs and that race and income, not Part C EI services, predicted families reporting access to informal supports and formal services. Using longitudinal data from the Early Intervention Collaborative Study, Crossman et al. (2018) found no relationship between maternal competence and home visit intensity, but a significant association between maternal competence and the perceived helpfulness of Part C EI home visits. Finally, Epley et al. (2010) studied all five ECO Center-recommended family outcomes as well as family quality of life and found that ECO Center-recommended family outcomes were predicted by parent report on how well Part C EI services had met their perceived needs. The family outcomes in turn predicted increased family quality of life, defined as family interaction, parenting, emotional well-being, and physical/material well-being.

Overall, these studies support positive family outcomes according to parent report within the population of families involved with Part C EI with the major limitation of predominantly representing middle-income White families. For example, Epley et al. (2010) included participants who were 92.2% White and only 11.7% below the federal poverty level. Importantly, when these studies included race, ethnicity, and income factors in their analyses, they found poorer outcomes for families from minoritized racial and ethnic groups, as well as among families with lower income levels (Bailey et al., 2005, 2007; Raspa et al., 2010).

Parenting Stress and Child Functioning.

There is an additional body of research that has examined parenting stress, an important family quality of life indicator, among families of children with developmental disabilities. To some degree, parenting stress, defined as the “aversive psychological reaction to the demands of being a parent” (Deater-Deckard, 1998, p. 315), is considered normal—it “is part of the job” (Crnic & Greenberg, 1990; Deater-Deckard & Scarr, 1996, p. 46). A substantial literature, however, shows the deleterious effects of clinically elevated parenting stress on parents’ well-being and family system functioning (Deater-Deckard, 1998; Karst & Van Hecke, 2012). Parents of children with developmental disabilities, especially autism spectrum disorder, have reported chronic, clinically elevated levels of parenting stress that relates to parents having poorer physical and mental health outcomes (Abbeduto et al., 2004; Baker-Ericzén et al., 2005; Carter et al., 2009; Dabrowska & Pisula, 2010; Hayes & Watson, 2013; Murphy et al., 2007; Zablotsky et al., 2013). Studies of parenting stress in families with school-aged children on the autism spectrum suggest that between 40.4% and 47% of this subgroup of parents have chronically elevated parenting stress levels (Huang et al., 2014; Schutte et al., 2018; Valicenti-McDermott et al., 2015). Across three waves of data collection in a recent study of parents with autism-specific concerns for their young child, 46.2% to 52.2% reported clinically elevated parenting stress levels (DesChamps et al., 2020). In contrast, 10.3% to 16.7% of the parents in this study with other developmental concerns for their young children aside from autism reported clinically elevated stress levels (DesChamps et al., 2020).

Child social, communication, behavioral, adaptive, and cognitive impairments are considered major predictors of elevated parenting stress (Bebko et al., 1987; Davis & Carter, 2008; Hall & Graff, 2011; Herring et al., 2006; Lecavalier et al., 2006; McStay et al., 2014; Valicenti-McDermott et al., 2015). Among the various domains of child functioning, child social functioning in particular has been found to predict increased parenting stress, which may partially explain the difference in reported parenting stress between parents with autism-specific and other developmental concerns for their child (Davis & Carter, 2008; DesChamps et al., 2020). In addition, external factors related to difficulties with service systems (e.g., navigating special education, navigating health care, financial strains) and weak informal social supports also relate to substantial parenting stress and burden for families (Boyd, 2002; Kuhn et al., 2018; Nesbitt, 2000; Sharpe & Baker, 2007; L. E. Smith et al., 2012; Tobing & Glenwick, 2007).

Addressing Elevated Parenting Stress.

Due to these overall elevated parenting stress levels, there are ongoing efforts to understand and alleviate parenting stress in families of children with developmental disabilities (Crnic et al., 2017; Durand et al., 2013; Ginn et al., 2017; Hastings & Beck, 2004; Lindo et al., 2016). “Action-oriented” coping strategies such as planning, acceptance, and positive reframing have been shown to improve family adjustment (Bekhet et al., 2012; Feinberg et al., 2013, 2014; Melnyk et al., 2004, 2006; Whitehead et al., 2015). Furthermore, positive informal supports from family and friends, as well as formal supports through the service system, can promote parents’ resilience to stressors (Barker et al., 2011).

Across the literature that examines associations between Part C EI services specifically and parenting stress, findings have been mixed. Warfield et al. (2000) found no difference in parenting stress based on the intensity, duration, or comprehensiveness (i.e., number of service types) of EI services. Shonkoff et al. (1992) found associations between parenting stress and several factors relating to EI services including hours of services delivered in the home setting and provider type. Furthermore, the findings from these studies should be interpreted with caution in current times because they were completed over two decades ago with families who were predominantly non-Hispanic and White. Of note, 89% of the families in Shonkoff et al. (1992)’s sample was White; and in the Warfield et al. (2000) sample, 92% were White.

Representation of Low-Income Families and Racially and Ethnically Diverse Communities in Research

The limitations noted in the above body of work represent a larger problem. Most families of children with developmental disabilities participating in intervention studies have historically been White, highly educated, and at higher income levels (Barroso et al., 2018; Broder-Fingert et al., 2019). As a result, little is known about parenting stress or intervention outcomes among families of children with developmental delays or disabilities from minoritized racial and ethnic groups or households with lower socioeconomic status. A small amount of evidence indicates that Black parents of Head Start preschoolers do not experience elevated parenting stress relative to a norm-referenced sample (i.e., mean parenting stress at the 45^th percentile; Reitman et al., 2002). In the period following a positive screen for developmental concerns, Hispanic families from the same dataset as the present study reported lower levels of parenting stress and less frequent use of coping strategies than non-Hispanic families who were predominantly (79.4%) Black (Hickey et al., 2021).

Black as well as other minoritized racial and ethnic groups in the United States experience substantial, pervasive disparities in broader physical and mental health outcomes as well as inequitable access to physical and mental health care (National Academies of Sciences, Engineering, and Medicine, 2017). Minoritized racial and ethnic groups as well as low-income families are more likely to experience increased threats to well-being far beyond the stressors that can come with having a child with special needs. Structural inequities in the United States may alter individual family outcomes, use of services, and experiences of parenting stress related to child development. Thus, it is important to advance the field’s understanding of how child functional characteristics, parenting stress, and Part C EI services relate among families from minoritized racial and ethnic groups and low-income households who have been historically underrepresented in research.

Present Study Rationale and Research Questions

The body of prior literature evaluating parent and child outcomes of Part C EI services has focused exclusively on children and families who were already engaged in the Part C EI service system, with poor representation of low-income households and minoritized racial and ethnic groups. To our knowledge, no studies have compared outcomes between families of children enrolled in Part C EI and families of children with similar developmental profiles but who were not receiving Part C EI services. Many families of eligible children with, or at risk of, developmental delays do not enroll in Part C EI (Rosenberg et al., 2013), presumably for a variety of reasons. According to a recent systematic review, these reasons extend beyond family and child-related factors to include state-level policies, funding, referral systems, and administrative system structures (Twardzik et al., 2017).

It should be acknowledged from the outset that this natural experiment approach, in which outcomes are compared between families enrolled in Part C EI to their presumably eligible counterparts, cannot result in causal findings and carries an inherent limitation that various child-, family-, and system-level factors outside the control of researchers are what determine Part C EI enrollment. The present study investigated associations between child social functioning, parenting stress, and Part C EI enrollment in a sample of families from predominantly low-income households and minoritized racial and ethnic groups. All participating families had a toddler recently identified at high risk for a developmental disorder through primary-care-based universal autism screening across three urban medical centers in the United States. Prior to this positive autism screen, only a portion of children in our sample had previously been identified with a need for and enrolled in Part C EI services. We sought to replicate prior study findings that have demonstrated a relationship between child social functioning and parenting stress levels in this sample. We then examined whether receiving Part C EI service receipt was directly associated with lower parenting stress levels, and whether the relationship between child social functioning and parenting stress is different for families of children who have versus have not been receiving Part C EI. Our two primary research questions were as follows:

Research Question 1: How do Part C EI enrollment (enrolled versus not enrolled) and child social functioning relate to parenting stress?

Research Question 2: Does Part C EI enrollment (enrolled versus not enrolled) moderate the hypothesized relationship between child social functioning and parenting stress?

This study also included the following two exploratory research questions:

Research Question 3: Are there differences in sociodemographic or child functional characteristics between children who were already enrolled versus not yet enrolled in Part C EI at the time of a primary care-based positive developmental screen?

Research Question 4: Does the parental report of clinically elevated parenting stress differ based on sociodemographic factors and child functional characteristics?

Method

Larger Study Context and Participants

In this study, we analyzed a subset of data collected for a larger randomized controlled clinical trial (for a full report on the main trial, see Feinberg et al., 2021. The primary aim of the larger study was to test the effects of Family Navigation, a multi-component care management strategy, in reducing the time to receive a diagnostic developmental/behavioral evaluation after primary care referral. The larger study was intentionally conducted with families from low-income households and minoritized racial/ethnic backgrounds to represent those who have historically experienced disparities in access to developmental-behavioral specialty care (Liptak et al., 2008; Magaña et al., 2013; Mandell et al., 2002; K. A. Smith et al., 2020). The institutional review board at [Blinded] approved this study and informed consent was obtained from all participants.

Children from 11 pediatric primary care clinics across three large, urban integrated care networks in the northeastern United States were referred for a developmental-behavioral pediatrics evaluation and to the larger study after their primary care pediatrician identified them as at risk of developmental delay through developmental surveillance and universal autism-specific screening using the Modified Checklist for Autism in Toddlers, Revised with Follow-Up (M-CHAT-R/F; Robins et al., 2014).

A total of 854 families were referred to the larger study by primary care physicians, of whom 514 were excluded due to ineligibility after referral (n = 279), the study team’s inability to contact them (n = 135), or the family declining participation (n = 100). Study personnel re-administered the M-CHAT-R/F (Robins et al., 2014) following standard procedures to all potentially eligible families to ensure its standard administration and invited those with positive screens to participate. The M-CHAT-R/F was selected for use in this study because of its universal use at the participating primary care sites and the focus of the larger study on reducing disparities in access to developmental/behavioral diagnostic and treatment services. Although the M-CHAT-R/F is an autism-specific screener, its validation studies suggest that children who screen positive on it are at very high risk for developmental delay more broadly as well as at an elevated risk for autism. Specifically, the original MCHAT-R/F validation study found that a positive screen relates to a 47.5% risk of autism and a much higher (94.6%) risk of any developmental disorder (Robins et al., 2014). A recent urban population-based M-CHAT/F accuracy study demonstrated that upon diagnostic evaluation, only 14.6% of those who screened positive were found to have autism, whereas 72.4% were found to have any developmental delay or concern including autism. For these reasons, we describe the sample of children in our study as being at high risk for any developmental delay as opposed to at risk of autism specifically. Of note, by the time that the larger trial concluded, 59.7% of this sample of children had received a diagnosis of autism spectrum disorder and the other 40.3% received a different developmental/behavioral clinical diagnosis.

Of those families who were ineligible after referral, the most common reasons for larger study ineligibility were a negative M-CHAT-R/F screen with the study personnel (n = 141) and the child falling outside of the study age range of 15 to 27 months (n = 98). Less common reasons for ineligibility were issues such as families moving (n = 3) or the referred child having a prior clinical autism diagnosis that rendered the referral for a diagnostic evaluation unnecessary (n = 13). There were no exclusions based on the language spoken; interpreters were used as needed. Ultimately, dyads with one parent/guardian (henceforth referred to as parent) and one child from 339 families were enrolled in the larger trial.

Present Study Participants

Of the 339 larger study participant dyads, we restricted the sample for the present analysis to the 227 dyads who completed all key measures for Research Questions 1 and 2 with reference to the child involved in the larger study (i.e., index child)—specifically, the Parenting Stress Index-Short Form (PSI-SF), the Adaptive Behavior Assessment System, 2^nd Edition social subscale (ABAS-II), and a single question regarding Part C EI enrollment status. Of these 339 total participants, 103 dyads were excluded because the parent completed the PSI-SF with reference to a sibling rather than the index child, as they may choose to report on any of their children per the standard instructions on this measure (“Please focus on the child you are most concerned about. . . Answer all questions about the same child”). Among these remaining 236 dyads, three did not have a PSI-SF score due to missing responses and six dyads did not complete the ABAS-II. No dyads were excluded for missing C EI enrollment status data.

Among the 227 parent–child dyads included in the present study, 88.5% of parents/guardians were biological mothers; the remaining were biological fathers, other biological relatives, or nonbiological guardians. On average, parents were 30.7 years old (SD = 7.2). Approximately half (50.2%) of parents identified as Black or African American and non-Hispanic, 32.6% identified as Hispanic, 7.0% identified as White, non-Hispanic, and 10.1% identified as non-Hispanic and another race. A substantial percentage of parents (41%) were born outside the United States and 19.4% spoke a language other than English as their primary language. The sample was largely low-income, with 80.6% using public insurance and 64.3% receiving support from the Women, Infants, and Children (WIC) program. Most parents were high school graduates (79.7%); slightly more than half reported working outside the home, and 55.8% reported that they were married or living with a partner (versus being single parents). Children in the study were 70% male and on average 22.2 (SD = 3.4) months old. In response to the question “How worried are you about your child’s development?” on a 10-point Likert-type scale (10 = extremely, 0 = not at all), the average level of reported concern was 6.5 (SD = 3.1). See Table 1 for complete descriptive statistics from known variables in this sample.

Table 1.

Characteristics of Participating Parent-Child Dyads.

Parent and Child Variables	Overall (n = 227)
	n (%)
Child sex (male)	159 (70.0)
Preterm birth (child gestational age <37 weeks) Parent Race/Ethnicity	34 (15.0)
Black, non-Hispanic	114 (50.2)
Hispanic, Any Race	74 (32.6)
Other, non-Hispanic	23 (10.1)
White, non-Hispanic	16 (7.0)
Parent born in the United States	134 (59.0)
Referral language
English	183 (80.6)
Spanish	33 (14.5)
Other	11 (4.8)
Parent employment status (working outside home)	121 (53.3)
Parent education level (high school graduate)	181 (79.7)
Parent marital status (married or living with partner)	126 (55.8)
Child insurance type (public)	183 (80.6)
Receiving WIC Support	146 (64.3)
Receiving income assistance	41 (18.1)
	M (SD)
Child age (months)	22.2 (3.4)
Parent age (years)	30.7 (7.2)
Number of children in family	2.0 (1.2)
Parent worry about child’s development (0-10 scale)	6.5 (3.1)
M-CHAT-R/F total score	8.3 (3.2)
ABAS Social scaled score	5.2 (2.6)
PSI-SF Total T-score	53.3 (10.5)

Note. Due to missing data, descriptive statistics for the following variables were calculated with lower sample sizes: married or living with partner (n = 226), parent age (n = 224). WIC = women, infants, and children; M-CHAT-RF = modified checklist for autism in toddlers, revised with follow-up; ABAS = adaptive behavior assessment system; PSI-SF = parenting stress index-short form.

Measures

All data were collected via phone interview during the baseline phase of the larger randomized controlled trial prior to any intervention delivery.

Child Social Functioning.

The Adaptive Behavior Assessment System, 2nd Edition (ABAS-II), is a parent-report assessment of a child adaptive functioning across multiple domains (Harrison & Oakland, 2003). Multiple studies have validated the ABAS-II for capturing adaptive functioning impairments with good test–retest reliability and convergent validity (Farmer et al., 2018; Harrison & Oakland, 2003; Kenworthy et al., 2010; Milne et al., 2013). On the ABAS-II, parents rate the extent to which their child can perform each queried activity or behavior without help on a 4-point Likert-type scale from 0 (is not able) to 3 (always when needed). ABAS-II scores are reported in the form of norm-referenced scaled scores, which have a mean of 10 and a standard deviation of 3. The ABAS-II Social domain scaled score was used in present analyses and tested as the child-related predictor of parenting stress. According to the ABAS-II Manual, the Social domain is defined as “skills needed to interact socially and get along with other people, including having friends, showing and recognizing emotions, assisting others, and using manners” (Harrison & Oakland, 2003). Children in our sample had a mean ABAS-II Social scaled score of 5.2 (SD = 2.6), which corresponds to the 5^th percentile relative to children their age in the normative ABAS-II sample.

Part C EI Enrollment.

Enrollment in Part C EI (hereinafter EI enrollment) at the time of the larger study’s baseline period was assessed with parent responses to a single question asking if their child was currently receiving Part C EI services. Parents were considered enrolled in Part C EI if they answered “yes” to this question and not enrolled if they answered “no.” Families who reported that they had begun the referral process for Part C EI, but had not yet started services, were counted as a “no” response because no services had been received. In our sample, those enrolled in Part C EI at the baseline timepoint had already been receiving these services for a mean length of 9.3 months (SD = 7.7). The intensity, quality, or types of services received were not considered. Forty-one percent reported “yes” to EI enrollment at the baseline timepoint.

To verify this parent-reported Part C EI enrollment data, we compared parent report to Part C EI service records that were provided with parent permission from relevant government and local agencies. The records of 146 (64.3%) children in the analytic sample were available for comparison. We found that parent-reported services and agency service records matched 125 (86%) of the 146 participants whose documented services we were able to access. The remaining data were unavailable due to the governmental agencies’ limited capacities to summarize and share all of the children’s records.

In addition, the same question regarding EI enrollment was asked at later timepoints in the larger study. These responses were used to calculate the proportion of the sample who became enrolled in EI at some point in the future, before turning 3 years old. In comparison with the 41.4% who reported “yes” to EI enrollment at the baseline timepoint, 83.3% (n = 189) eventually reported “yes” to EI enrollment at any study timepoint before aging out at 3 years old.

Parenting Stress.

Parenting stress, the primary outcome measure, was measured with the Parenting Stress Index-Short Form, 4^th Edition (PSI-SF; Abidin, 2012), a 36-item parent-report questionnaire, derived from the Parenting Stress Index (PSI) full-length assessment. For each item, parents rate their agreement on a 5-point Likert-type scale from 1 (Strongly Agree) to 5 (Strongly Disagree). The PSI-SF has been shown to have high internal consistency and adequate test–retest reliability (Abidin & Brunner, 1995). Its psychometric properties have been studied and validated primarily with samples of white and Black parents (Abidin, 2012; Haskett et al., 2006; Reitman et al., 2002). However, Lee et al.’s (2016) study of its validity with predominantly Black and Latino caregivers in the United States similarly found good criterion validity and confirmation of its three-factor structure (Lee et al., 2016). The PSI-SF has also been used widely in studies of parents of children with developmental disabilities (e.g., DesChamps et al., 2020; Gupta, 2007; Hayes & Watson, 2013). A higher score indicates greater parenting stress, and a total raw score of 114 or above corresponds to clinically elevated results over the norm-referenced 90^th percentile (Abidin, 2012). To compare differences above and below the clinically elevated parenting stress cut point, this total raw score of 114 was used. For regression and moderation analyses, continuous PSI-SF total raw scores were converted to norm-referenced T-scores using the conversion table published in the PSI-SF 4^th Edition manual (T-score M = 50, SD = 10).

Overall, the mean PSI-SF parenting stress T-score in our sample was 53.3 (SD = 10.5). Raw PSI-SF scores above 114 (90^th percentile) were considered to fall within the clinically elevated range, per the PSI-SF 4^th Edition manual (Abidin, 2012). Given this cut-off, 13.7% (n = 31) of parents’ PSI-SF ratings indicated clinically elevated levels of parenting stress, with no differences in the EI versus non-EI group.

Analyses

All statistical analyses were performed using 2013 SAS^™, version 9.4 Software, with an α level of 0.05. Sociodemographic characteristics including parent race/ethnicity, employment status, insurance type, and child sex were analyzed as categorical variables. Parent and child age, PSI-SF T-scores, and ABAS-II Social scaled scores were analyzed as continuous variables with approximately normal distributions. We answered our first two research questions with a series of generalized linear mixed modeling steps. In an initial unadjusted model that accounted for clustering by site, we estimated the association between level of child social functioning (independent variable) and level of parenting stress (dependent variable). We then added EI enrollment status to this model. Building on this to answer our second research question, we added an interaction term between the child social functioning and EI enrollment status variables to the model. Next, suspected covariates (child sex and gestational age, race/ethnicity, referral language, insurance type, parent employment status, parent education level, parent marital status, parent age, and number of children in the household) were added to the generalized linear mixed model individually. Covariates that resulted in a 10% or greater change in the β-estimate of the ABAS-II Social scaled score X EI enrollment status interaction term were retained for inclusion in the final model; thus, in our final model, we estimated the interaction between ABAS-II Social scaled scores and EI enrollment status including confounding variables, defined as those suspected covariates listed above that resulted in a 10% or greater change in the interaction term β-estimate.

To answer our third research question, differences between participants who were versus were not enrolled in EI (binary yes/no variable) were examined using a series of independent samples t tests for continuous variables and chi-square tests for categorical variables. To examine differences between families with parents reporting parenting stress above and below the clinically elevated PSI-SF cut-point for the fourth research question, these statistical methods were repeated.

Results

Research Question 1

In the initial unadjusted model, a significant inverse association between the level of child social functioning and the level of parenting stress was found (β [SE] = −0.61 [0.21], p = .005). In other words, lower child social functioning was related to higher parenting stress. EI enrollment was then added to the model; the association between ABAS-II Social scaled score and PSI-SF T-score remained significant (β [SE] = −0.66 [0.27], p = .02), and there was not a significant association between EI enrollment and parenting stress (β [SE] = −1.71 [2.13], p = .42).

Research Question 2

When added to the model, the interaction term between ABAS-II Social scaled score and EI enrollment status to the model was statistically significant (β [SE] = 1.06 [0.38], p = .006). EI enrollment significantly moderated the relationship between child social functioning (i.e., ABAS-II Social scaled scores) and parenting stress (i.e., PSI-SF T-scores). The following covariates accounted for >10% change in the interaction β when tested one at a time and were kept in the adjusted models: parent employment status, marital status, education level, and number of children in the household. See Table 2 for the full final model results and Figure 1 for a graphic display of the interaction effect. In Figure 1, the red line represents the adjusted trend between child social functioning and parenting stress for those who were not enrolled in EI, and the blue line represents the same relationship but for those who were enrolled in EI. To further describe this interaction effect, in the adjusted models stratified by EI enrollment, the association between lower child social functioning and higher parenting stress remained significant for those children who were not enrolled in EI (β [SE] = −1.20 [0.33], p < .001) but was not significant for children enrolled in EI (β [SE] = 0.01 [0.32], p = .97).

Table 2.

Interaction Between ABAS Social Scaled Score and Enrollment in EI With Covariates, (n = 227).

Effect	Estimate (SE)	p
Intercept	64.59 (3.15)	.002
ABAS Social scaled score	−1.19 (0.43)	.007
Receiving EI	−9.06 (3.19)	.005
ABAS Social scaled score × Receiving EI	1.25 (0.21)	<.0001
Parent employment status (working outside home)	−3.54 (0.95)	<.0001
Parent education level (high school graduate)	−3.30 (0.61)	<.0001
Parent marital status (married or living with partner)	−3.10 (1.73)	.07
Number of children in family	1.10 (0.75)	.14

Note. Dependent variable = PSI-SF T-score. ABAS = adaptive behavior assessment system; EI = early intervention; SE = standard error; PSI-SF = parenting stress index-short form.

Figure 1.

Moderating Effect of EI on Social Functioning and Parenting Stress.

Note. PSI-SF = parenting stress index-short form; ABAS = adaptive behavior assessment system; EI = early intervention.

Research Question 3

Children enrolled in EI at baseline were more likely than those not enrolled in EI to have been born pre-term (24.5% vs. 8.3%, p < .001), were older (22.8 months vs. 21.8 months, p = .03), and had lower social functioning (M = 4.8 vs. 5.5; p = .03). Parents of children enrolled in EI were significantly older (M = 32.2 years vs. 29.7 years, p = .01) and were less likely to be employed (40.4% vs. 62.4%, p = .001). There were no statistically significant differences between the EI enrollment groups by parent race/ethnicity, country of origin, referral language, parent education level, marital status, insurance type, WIC status, income assistance status, child sex, or the number of children in the household. Parents’ level of worry about their child’s development and M-CHAT-R/F scores did not differ between the EI enrollment groups. See Table 3 for additional characteristics of families who were and were not enrolled in EI.

Table 3.

Exploratory Analysis Comparing EI and non-EI Enrollment Groups, (n = 227).

Parent and Child Variables	Not in EI (n = 133)	In EI (n = 94)	p value
	n (%)	n (%)
Child sex (male)	93 (69.9)	66 (70.2)	.96
Preterm birth (gestational age <37 weeks)	11 (8.3)	23 (24.5)	<.001
Parent race/ethnicity			.47
Black, non-Hispanic	68 (51.1)	46 (48.9)
Hispanic, any race	41 (30.8)	33 (35.1)
Other, non-Hispanic	12 (9.0)	11 (11.7)
White, non-Hispanic	12 (9.0)	4 (4.3)
Parent born in the United States	79 (59.4)	55 (58.5)	.89
Referral language			.65
English	109 (82.0)	74 (78.7)
Spanish	19 (14.3)	14 (14.9)
Other	5 (3.8)	6 (6.4)
Parent employment status (currently employed)	83 (62.4)	38 (40.4)	<.01
Parent education level (high school graduate)	104 (78.2)	77 (81.9)	.49
Parent marital status (married or living with partner)	71 (53.4)	55 (59.1)	.39
Child insurance type (public)	107 (80.5)	76 (80.9)	.94
Receiving WIC support	79 (59.4)	67 (71.3)	.07
Receiving income assistance	24 (18.0)	17 (18.1)	.99
	Mean (SD)	Mean (SD)	p-value
Child age (months)	21.8 (3.4)	22.8 (3.4)	<.05
Parent age (years)	29.7 (6.5)	32.2 (7.9)	<.01
Number of children in family	1.9 (0.9)	2.2 (1.5)	.07
Parent worry about child’s development (0-10 scale)	6.7 (3.1)	6.2 (3.2)	.25
M-CHAT-R/F total score	8.0 (3.0)	8.7 (3.5)	.10
ABAS Social scaled score	5.5 (2.5)	4.8 (2.8)	<.05
PSI-SF T-score	53.8 (9.6)	52.6 (11.6)	.39
Months enrolled in EI	N/A	9.3 (7.7)	N/A

Note. EI = early intervention; WIC = women, infants, and children; M-CHAT-RF = modified checklist for autism in toddlers, revised with follow-up; ABAS = adaptive behavior assessment system; PSI-SF = parenting stress index-short form.

Bold value indicates the significant p values.

Research Question 4

In the exploratory bivariate analyses, parents with clinically elevated parenting stress were more likely to be single (67.7% vs. 41%, p = .005), reported higher levels of worry about their children’s development (M = 7.7 vs. 6.3, p = .02), and rated their children with higher scores on the M-CHAT-R/F (9.7 vs. 8.1, p =.008). No other sociodemographic or child-functional characteristics significantly related to clinically elevated parenting stress (Table 4).

Table 4.

Exploratory Analysis Comparing Groups Above and Below Clinical Cut-off on PSI-SF, (n = 227).

Parent and Child Variables	PSI Total < 114 (n = 196)	PSI Total ≥ 114 (n = 31)	p value
	n (%)	n (%)
Child sex (male)	135 (68.9)	24 (77.4)	.33
Preterm birth (child gestational age <37 weeks)	32 (16.3)	2 (6.5)	.15
Parent race/ethnicity			.39
Black, non-Hispanic	95 (48.5)	19 (61.3)
Hispanic, any race	67 (34.2)	7 (22.6)
Other, non-Hispanic	19 (9.7)	4 (12.9)
White, non-Hispanic	15 (7.7)	1 (3.2)
Parent born in the United States	113 (57.7)	21 (67.7)	.29
Referral language			.86
English	157 (80.1)	26 (83.9)
Spanish	29 (14.8)	4 (12.9)
Other	10 (5.1)	1 (3.2)
Parent employment status (employed)	109 (55.6)	12 (38.7)	.08
Parent education level (high school graduate)	159 (81.1)	22 (71)	.19
Parent marital status (married or living with partner)	116 (59.5)	10 (32.3)	<.01
Child insurance type (public)	158 (80.6)	25 (80.6)	1.0
Receiving WIC support	128 (65.3)	18 (58.1)	.43
Receiving income assistance	30 (15.3)	11 (35.5)	<.01
	M (SD)	M (SD)	p value
Child age (months)	22.2 (3.5)	22.2 (3.4)	.98
Parent age (years)	30.8 (7.2)	30.2 (7.0)	.66
Number of children in family	2.0 (1.2)	2.2 (1.2)	.36
Parent worry about child’s development (0-10 scale)	6.3 (3.2)	7.7 (2.5)	<.05
M-CHAT-R/F total score	8.1 (3.2)	9.7 (3.1)	<.01
ABAS Social scaled score	5.3 (2.7)	4.8 (2.6)	.30

Note. PSI = parenting stress index-short form; WIC = women, infants, and children; M-CHAT-RF = modified checklist for autism in toddlers, revised with follow-up; ABAS = adaptive behavior assessment system.

Bold value indicates the significant p values.

Discussion

Part C EI is the primary source of publicly available services for children under three with developmental concerns (IDEA Part C, 2004). To our knowledge, this is the first study to specifically examine the associations between Part C EI enrollment, child social functioning, and parenting stress in a sample of young children with or at high risk for developmental delays in the United States. Importantly, this study did so with a sample that has previously been historically underrepresented in research: predominantly low-income families from minoritized racial and ethnic groups.

Research Question 1: Associations Between Child Social Functioning, Part C EI Enrollment, and Parenting Stress

Adding to prior mixed findings on the influence of early childhood intervention on parenting stress and broader family well-being outcomes (e.g., Bailey et al., 2005; Crossman, 2018; Doyle et al., 2017; Epley et al., 2010; Hebbeler et al., 2007; Keen et al., 2010; Raspa et al., 2010), we found no significant association between EI enrollment and parenting stress level. The lack of an association between these variables may reflect the complex and different ways in which parenting stress and parents’ help-seeking behavior can relate. For some parents, higher stress may lead to seeking out support through EI; but for others, higher stress levels may cause parents to feel too overwhelmed to engage. Given that other early childhood interventions have led to reductions in parenting stress (Keen et al., 2010), further research, particularly longitudinal studies that follow families over time as they enter and begin to receive EI, is needed to determine whether EI services can alleviate parenting stress or improve other aspects of family quality of life.

Next, we found that parents whose children had lower social functioning tended to report higher levels of parenting stress. This finding is consistent with prior literature (Davis & Carter, 2008; Hall & Graff, 2011), which supports the notion that social functioning consistently arises as associated with parenting stress in parents of toddlers with or at risk for developmental delays including autism-related concerns.

Research Question 2: Part C EI as Moderator of the Relationship Between Child Social Functioning and Parenting Stress

When Part C EI enrollment was included as a moderator in the model with child social functioning predicting parenting stress, we found that EI enrollment significantly moderated the relationship between these two variables. In other words, we found that lower child social functioning was associated with higher parenting stress only for children not enrolled in Part C EI. For those enrolled in Part C EI, there was no significant relationship between child social functioning and parenting stress. See Figure 1 for a complete graphical display of this finding. To provide an example, an ABAS-2 Social scaled score of 2 had a predicted PSI-SF T-score value of 58.2 in the group that had not been receiving Part C EI, and a predicted PSI T-score value of 51.6 in the group that had been receiving Part C EI. An ABAS-2 Social scaled score of 12 had a predicted PSI-SF T-score value of 46.3 in the group that had not received Part C EI services, and a predicted value of 52.3 in the group that had been receiving Part C EI services.

These findings cannot be interpreted causally due to the study design. Instead, we offer plausible explanations that require testing through future research. One plausible explanation that could be supported through future research is that Part C EI may act to relieve parenting stress that arises along with lower child social functioning levels. Another plausible explanation is confounding by indication (Kyriacou & Lewis, 2016)—families who enroll in Part C EI may be fundamentally different from their presumably eligible counterparts. Families may have differentially enrolled in Part C EI depending on the combination of their child’s functioning level and their own parenting stress, with parenting stress potentially leading to either action or disengagement.

Research Question 3: Exploratory Differences Between Families of Children Who Were Enrolled Versus Who Were Not Enrolled in Part C EI at the Time of Positive M-CHAT-R/F Screen and Referral to Developmental and Behavioral Pediatrics

In contrast to the 41.4% of children in our sample who were enrolled in EI baseline, among the 58.6% who were not enrolled at baseline, 83.3% enrolled in EI at a later timepoint and the remaining 16.7% did not enroll before turning 3 years old. This suggests that the sample largely had developmental delays and a need for EI services. The children enrolled in EI at the baseline time point were older and more likely to be born preterm, which likely reflects the practice of referring to EI in response to prematurity at birth. In addition, as children age, there is simply more time for developmental delays to emerge and for the Part C EI enrollment process to be completed. Although children enrolled in EI at baseline had significantly lower ABAS-scaled scores, children in both groups showed significant impairments greater than one standard deviation below the normative sample for the measure. This is not surprising considering the larger study’s focus on children with identified high risk for developmental delay. Finally, parents of children enrolled in EI at baseline were less likely to be employed, suggesting greater availability to receive these services, which are often delivered in-home with a parent present.

Research Question 4: Exploratory Differences Between Families Reporting Above Versus Below the Cut-off of Clinically Elevated Parenting Stress

Our exploratory analyses shed light on differences between parents with parenting stress levels above versus below the clinically elevated cut-point. Related to the child’s development, parents with clinically elevated parenting stress were more likely to endorse a higher M-CHAT-R/F score and expressed more worry about their child’s development. It is unknown whether parental worries about children’s development and responses to the M-CHAT-R/F are related to true differences in children’s functional presentations considering that these measures are reliant on parent reports. This is especially true given the conflicting finding that there were no differences in ABAS-II social subscale scores between these two groups.

Regarding parents’ social support and parenting stress, single parents, who often report less social support than their married peers (Nunes et al., 2021), were more likely to have clinically elevated parenting stress. Experiencing parenting-related demands and responsibilities without the support of a co-parent may evoke increased parenting stress. This finding is consistent with other studies showing that marriage can decrease perceived stress and depressive symptoms, and improve maternal mental health (Beam et al., 2017; Hannighofer et al., 2017).

Within our sample of families who predominantly represented minoritized racial and ethnic groups, we did not detect differences in clinically elevated stress based on sociodemographic factors such as race, ethnicity, referral language, or country of origin. Families within specific subgroups may possess unique protective and risk factors as well as different stress responses to various combinations of adversity and prosperity. Extant literature includes mixed findings regarding the potentially compounding impact of the stressors associated with racial and ethnic identity in the United States and having a low household income level on the stressors associated with having a child with a developmental disability (Barroso et al., 2018). In our sample, 13.7% (n = 31) of parents’ PSI-SF scores fell in the clinically elevated range; our sample included children who went on to be given a variety of developmental and behavioral diagnoses, including autism.

The level of parenting stress found in our study is similar to prior findings of parenting stress levels among parents with developmental concerns aside from autism (DesChamps et al., 2020) and substantially lower than samples of children with autism diagnoses or undiagnosed autism concerns (DesChamps et al., 2020; Huang et al., 2014; Valicenti-McDermott et al., 2015). Parenting stress levels in our sample were substantially higher when compared with a prior study with predominantly African American parents whose preschool-aged children were enrolled in Head Start (raw M = 86.6, SD = 25.3, percentile = 63 versus raw M = 73.44, SD = 25.56, percentile = 45; Reitman et al., 2002). This may signal differences in experienced stress, increased protective factors related to parenting stress, and/or reduced self-reported stress among the low-income and U.S. racially and ethnically diverse communities represented in our sample, combined with general increases in stress among parents of children with developmental delays and disabilities. Along these lines, Valicenti-McDermott et al. (2015) and Hickey et al. (2021) previously found that Hispanic parents of children with developmental disabilities or risk of developmental delays reported significantly lower parenting stress than non-Hispanic parents. Investigators in these prior studies were similarly unable to explain whether this difference was due to true resilience, differences in self-reporting in this domain, or differences in child functioning levels. Further research is needed in this area.

Limitations

First, the generalizability of our study may be limited to urban, low-income populations from minoritized racial and ethnic groups in the United States who were recruited from primary care after a positive screen using a tool similar to the M-CHAT-R/F. Our findings may not extend to rural populations, high-income populations, other racial or ethnic groups, or children at risk of developmental delays that are unlikely to be picked up with the M-CHAT-R/F (e.g., those born prematurely). This study also relied on self-report of parenting stress and parent-report of child’s social functioning. Self-report measures can be limited insofar as they are subject to inherent recall and social desirability bias. As a result, associations between child social functioning and parenting stress may be conflated by having the same single reporter for both the ABAS-II and the PSI-SF. Considering that this was a cross-sectional study examining associations, our conclusions cannot be interpreted as causal. As noted earlier, it is possible that the differences detected between groups actually reflect that parents with certain combinations of characteristics or circumstances differentially enroll in EI. In this study, we assessed the relationship between EI enrollment and only a limited number of child and parent-related factors at one point in time; additional unexplored factors may exist that can better predict which presumably eligible families actually enroll in EI services. Future research is needed to clarify the relationship between these variables. Given that this study examined data from only one baseline time point, we were unable to assess long-term trajectories of parenting stress before and after enrollment in EI.

Conclusion

Our findings suggest that among predominantly low-income families from racially and ethnically diverse communities who have not received EI, children’s social functioning has a strong relationship with parenting stress. In contrast, no relationship between child social functioning and parenting stress was found among those enrolled in EI. While findings cannot be interpreted causally, they suggest that further research to test whether Part C EI may help parents cope with stress related to their children’s social functioning is warranted. When examining the parenting stress levels reported in our sample within the context of extant research findings, it appears that unique combinations of income level, race/ethnicity, child functioning level, services received, and additional unexplored factors may together influence parenting stress and resilience toward raising a child with developmental delays.

Future Research Directions

Future longitudinal research is needed to compare parenting stress before and after Part C EI enrollment and to determine whether Part C enrollment leads to a reduction in parenting stress over time and for whom. Overall, parenting stress in our sample was generally below the clinically elevated range, suggesting either a reluctance to self-report parenting stress or resilience against parenting stress. These findings are an important step in further understanding the potentially critical role that Part C EI services play in promoting the well-being of families of young children with developmental delays.