Are maternal self-reports of social difficulties apparent in interactions with their children?

Abstract

To inform parent-mediated intervention models, this study assessed if family affectedness (i.e., elevated autism symptoms in more than one child) was associated with maternal self-reported social difficulties (as indexed by the Social Responsiveness Scale; SRS-2), and social interactions during play. As part of a prospective study, 71 mothers completed the SRS-2 and a play session. Interactions were coded for a range of prosocial behaviors, including gaze, positive affect, and vocalizations. Overall, mothers with multiple children exhibiting autism symptoms self-reported significantly more social difficulties on the SRS-2, when compared to mothers raising only typically developing children, or one child with autism. However, even with elevated SRS-2 scores, mothers with higher family affectedness demonstrated comparable social exchanges with their children during play.

In families raising children with an autism spectrum disorder (ASD), studies often highlight elements of the broader autism phenotype (BAP) within the family. BAP reflects a wide range of subclinical autism-like features often measured via self- or partner/spouse-report questionnaires, such as the Social Responsiveness Scale (SRS-2; Constantino & Todd, 2005), the Autism Quotient (AQ; Baron-Cohen et al., 2001), or the Broader Autism Phenotype Questionnaire (BAPQ; Hurley et al., 2007). Elevated scores on these questionnaires generally suggest difficulties in social responsivity, evidence of restricted repetitive interests, and/or tendencies towards select personality traits (e.g., aloofness). For children 3 years or younger, standardized developmental assessments and clinical judgements are most commonly used to characterize the BAP (e.g., see Messinger et al., 2013; Ozonoff et al., 2014). For families with a history of ASD, previous studies document subclinical autism traits for biological parents and other children in the home (see reviews Gerdts & Bernier, 2011; Pisula & Ziegart-Sadowska, 2015; Rubenstein & Chawla, 2018). However, as summarized in Rubenstein and Chawla (2018), current sampling and measurement variability contribute to wide-ranging estimates of parental BAP, ranging anywhere from ~2.6 – 80%. Collectively, these reviews support the presence of and developmental importance of sub-clinical autism features in family members. Recognizing current clinical best practice guidelines for young children at elevated risk or diagnosed with ASD recommend naturalistic parent-mediated interventions (PMI), understanding diverse family dynamics, such as levels of household affectedness, may be especially salient to optimize treatment effects and targeted family support.

Parent Mediated Interventions (PMIs)

Prospective monitoring studies¹ have provided a wealth of information regarding ASD recurrence within families and early ASD risk markers over the past decade (e.g., McDonald et al., 2020; Ozonoff et al., 2011). Recognizing best practice concerns, prospective, or pre-emptive early interventions have emerged to support families before and as diagnoses develop (see reviews, Bradshaw et al., 2015; Green & Garg, 2018; Landa, 2018). Elevated risk-oriented interventions for children under 3 years are primarily social in nature and target autism symptoms, developmental competencies, and/or dyadic social skills. To date, prospective infant sibling PMIs that incorporate observational social/play paradigms to index treatment progress document subtle yet significant treatment effects, with the majority of improvements documented in parent-oriented behaviors (e.g., parental synchrony- Green et al., 2017; parental responsiveness- Kasari et al., 2015; Watson et al., 2017). Recognizing prospective interventions require training parents as interventionists, examining the relation between parental self-reported social communication and observational interactions with their children, may be especially salient in identifying families that may benefit from additional support/scaffolding within PMI settings.

Behaviors in Context

Within the context of autism-risk, the majority of studies to date have assessed parent-child social exchanges at the microanalytic level to identify social differences/difficulties within the first few years of life (e.g., infants decreased gaze towards mothers; see review, Wan et al., 2019). Recognizing the important role of caregiver engagement, a handful of prospective monitoring studies have assessed caregiver-specific play behaviors and generally report either comparable social bids or increases in prosocial play behaviors. Specifically, increased/comparable gaze to face, positive affect expressions, vocalizations, and contingent responsiveness are documented in previous research (Choi et al., 2020; Cohen et al., 2013; Ozonoff et al., 2010; Schwichtenberg et al., 2019; Talbott et al., 2015, 2016). Although, it’s important to note that observational-based metrics in prospective monitoring studies are not without limitations. Diverse approaches to assessing group membership and social construct quantification limits our capacity to compare across studies. Group membership has traditionally included risk status (elevated/high vs. low-risk) and/or the enrolled child’s developmental outcome classifications (ASD vs. Non-ASD developmental concerns vs. typical development). However as prospective studies have collectively identified early risk markers over the past decade, more nuanced approaches to group distinctions have emerged to refine our predictions of later developmental outcomes. This includes assessing early markers as a function of combined risk and outcome- high-risk ASD vs. low-risk ASD, or by family ASD incidence - simplex vs. multiplex families. In addition, with respect to contingent-derived dyadic constructs, such as responsivity, both dyadic partners contribute to the occurrence of a given social exchange, such as when an infant looks towards her mother’s face and mother responds with a smile. Thus, a dyad with a “low maternal responsiveness score” infers that either the mother is missing social cues from her infant, and/or minimal opportunities for response are initiated by her infant. Regardless, identifying the source of the aforementioned mismatch is crucial to provide concrete recommendations for ongoing prospective developmental monitoring and ultimately, PMI efforts.

Recognizing process-oriented behavioral coding approaches are well suited to capture intricate social processes between infants and caregivers, exploring concurrent associations between self-reported and observed constructs of social responsivity may reveal family-level social communication distinctions, as a function of autism-risk. In our previous research, elevated autism features, as indexed by the SRS-2 maternal self-reports, and differences in maternal core social competencies during play (i.e., gaze; positive affect; vocalizations) were not evident between traditional high- and low-risk family groups (Kellerman et al., 2020). However, in a subset of studies addressing parental BAP characteristics and pre-emptive PMI efforts, modest support for social communication differences were reported. For example, Flippin and Watson (2018) reported significant associations between maternal BAP and play behaviors, such that mothers with higher aloofness scores on the BAPQ used fewer verbal responses with their children. And in elevated/high-risk studies distinguishing between simplex and multiplex status, Gerdts and colleagues (2013) reported more social difficulties in the domains of social interests and conversational skills for both mothers and fathers from multiplex families, as indexed by the Broader Phenotype Autism Symptom Scale (BPASS- Dawson et al., 2007). Building on these discrepant findings, the present study leverages a prospective monitoring study design to further evaluate how self-reports of social responsiveness may relate to observed social exchanges with their children. To move beyond diagnosis-based classifications (i.e., multiplex v. simplex; elevated/high- v. low-risk) this study recognizes that diagnostic status may not be the only meaningful index of social communication difficulties and considers a continuum of high, moderate, and low family affectedness. Wherein, affected is defined by the presence of ASD or elements of the BAP in families raising multiple children exhibiting autism symptoms. And admittedly, this decision is a functional one based on our sample distribution (described further below).

Current Study

To inform ongoing prospective PMIs, the present study aimed to assess if maternal social communication difficulties, as indexed by maternal self-report and observed play behaviors, vary as a function of family affectedness. For Aim 1, we assessed whether mothers raising multiple children exhibiting autism symptoms (henceforth referred to as the high-family affectedness group, high-FA) self-endorsed more social difficulties when compared to simplex ASD (i.e., moderate-FA group) families, and families with no familial history of ASD (i.e., low-FA group). For Aim 2, we examined whether self-reported social responsiveness was associated with concurrent observed maternal play behaviors.

Method

Procedure

The present study was conducted under the approval of Purdue University’s Institutional Review Board. Ninety-nine families enrolled in this longitudinal study when their infants were between 6 to 18 months of age. Given the nature of the prospective design, visit attendance (though strongly encouraged) was not required for all 8 possible laboratory visits from 6 to 36 months of age. The present study includes questionnaire and observational data from 71 enrolled families’ final laboratory visit (occurring between 24 and 36 months of age). For details regarding the 28 families excluded from the current study see Figure 1.

Figure 1. Flowchart of Family Affectedness (FA) Groupings.

Note. ¹Twenty-eight families were excluded from the present study for one or more of the following reasons: ²For adoptive/foster parent-child relationships (n = 3) and multiple enrolled siblings within the same family (n = 4). The first enrolled infant with complete SRS-2 and play data was selected for the current study. ³For missing/incomplete SRS-2 forms (n = 6). ⁴For families that did not complete a final outcome visit (n = 7) and initial low-risk enrolls that received an ASD or non-TYP outcome (n = 13). ⁵High Family Affectedness, includes elevated-risk families with two or more children exhibiting autism symptoms (i.e., enrolled child and at least one older sibling; or at least two older siblings). ⁶Moderate Family Affectedness, includes elevated-risk infants with a family history of one sibling diagnosed with ASD. ⁷Low Family Affectedness, includes low-risk infants who did not exhibit any ASD symptoms or non-TYP concerns.

Participants

As part of the larger prospective study tracking infants’ development from 6 to 36 months, 71 mother-infant dyads from families raising at least one child with ASD (elevated-risk group, n = 34), or families with no history of ASD in 1^st, 2^nd, or 3^rd degree relatives (low-risk group, n = 37) participated in the current study. Mothers completed the SRS-2 adult self-report questionnaire to index their social responsiveness during their child’s final laboratory visit at 24 months (n = 4), 30 months (n = 38), or 36 months of age (n = 29). Of the 71 dyads with completed maternal SRS-2 questionnaires, the video-recorded free-play task was completed by 63 dyads at their final laboratory visit. Missing dyadic data at one of the three qualifying final laboratory visits for these eight dyads included only the father or grandmother accompanying child at the visit (n = 5), paradigm not administered due to family needing to leave early (n = 1), or recording/camera angle errors (n = 2). In addition, none of the biological mothers included in this study self-reported an ASD diagnosis.

Outcome classifications for infant siblings were assigned at their final laboratory visit, utilizing the Mullen Scales of Early Learning (MSEL; Mullen, 1995) and Autism Diagnostic Observation Schedule (ADOS-2; Lord et al., 2012). Based on criteria developed by the Baby Siblings Research Consortium (Ozonoff et al., 2014), infants were assigned to one of three defined outcome classifications: typical development (i.e., TYP), non-typical development (i.e., Non-TYP), or ASD. Non-TYP outcomes included children with language delays, clinically relevant behavioral challenges, and/or global developmental delay. In addition, elevated-risk infants classified into the Non-TYP group exhibiting subclinical autism features were assigned a distinct sub-outcome classification of BAP based on MSEL T-scores, ADOS-2 severity score, and clinical best estimates. Additional outcome classification details are summarized in Supplemental Material (S.1).

Family Affectedness (FA) Groupings

The final participant groupings for the current study included distinctions based on infants’ initial enrolled risk status and subsequent outcome classifications. The high-FA group (n = 22) included families with 2 or more children exhibiting autism symptoms. Scenario 1 for qualifying as high-FA required the enrolled elevated-risk child to have an older sibling with a confirmed ASD diagnosis and for the enrolled child to either have received either an ASD (n = 5) or BAP (n = 17) outcome classification. Scenario 2 for high-FA group membership required the enrolled elevated-risk child to have received a Non-TYP or TYP outcome and have at least two older siblings with confirmed ASD diagnoses (n = 0). Do note five high-FA children who received BAP outcomes in the current study were from families with at least two older siblings with an ASD diagnosis. The moderate-FA group (n = 12) included enrolled elevated-risk children who received either Non-TYP or TYP outcome classification and had only one older sibling diagnosed with ASD. The low-FA group (n = 37) included low-risk children who received a TYP outcome classification and did not have any siblings diagnosed with ASD. Participant descriptive statistics are provided in Table 1. See Figure 1 for a flowchart illustrating the progression of group membership from enrolled risk status to FA group assignment.

Table 1.

Sample Demographic Information for Enrolled Infant and Mother Stratified by Family Affectedness (FA)

	High-FA	Mod-FA	Low-FA
N	22	12	37
Infant sex, n(%)
Male	15(68%)	8(67%)	19(51%)
Infant race, n(%)
Caucasian	21(96%)	12(100%)	33(89%)
Multiracial	1(4%)		2(5%)
Asian			1(3%)
Unreported			1(3%)
Infant ethnicity, n(%)
Hispanic/Latino	1(4%)	1(8%)
Unreported/Unknown		1(8%)	1(3%)
Outcome Characteristics
Age in months, M(SD)	32.63(4.0)	33.03(4.5)	31.33(3.6)
Outcome Classification, n(%)
ASD	5(23%)
BAP1	17(77%)
Non-TYP		0(0%)2
TYP		12(100%)	37(100%)
Maternal Characteristics
Maternal age in years, M(SD)	31.50(4.3)	31.65(3.1)	31.82(4.3)
Maternal education, n(%)
High school or GED	1(~4%)		2(5%)
Trade or vocational	1(~4%)
Associates or 2-year degree	4(18%)	1(8%)	1(3%)
Some college	5(23%)	2(17%)	2(5%)
College degree	9(41%)	8(67%)	22(60%)
Master’s degree	1(~4%)	1(8%)	7(19%)
Professional degree	1(~4%)		3(8%)

Note. High-FA = elevated-risk group, family history of 2 or more children exhibiting autism symptoms. Mod-FA = elevated-risk group, family history of 1 child diagnosed with ASD. Low-FA = low-risk group, no endorsed ASD symptoms in target child. GED = General Educational Development.

¹BAP criteria established in Ozonoff et al. 2014. Five children exhibiting BAP also had at least 2 siblings with ASD.

²No children in the Mod-FA group met the Non-TYP (excluding BAP) outcome criteria.

Primary Measures

Social Responsiveness Scale (SRS-2).

The SRS-2 is a 65-item informant-report questionnaire which assesses social interactions, relationships, and communication skills (Constantino, 2002; Constantino & Todd, 2005; Constantino et al., 2009). SRS-2 Total T-Scores fall within a range to index levels of social difficulty: Normal (T-scores ≤ 59), Mild to Moderate (T-scores 60–75), Severe (T-scores ≥ 76). SRS-2 standardized Total T-Scores and subsequent domain scores were used to index mothers reported social responsivity.

Mother Child Play Interaction.

Mothers were asked to play with their children for five minutes with a standardized set of age-appropriate toys that included a ball, shape sorter, baby-doll, set of toy telephones, car, and baby blanket. Play interactions were video-recorded and later coded in Mangold INTERACT’s behavioral coding program. Child and mother gaze, positive affect, and vocalizations were independently coded using a discrete micro-analytic approach following coding criteria established in Ozonoff et al. (2010). Research assistants were trained to capture when a pre-defined play behavior began (on) and ended (off). For each on and off code, a frequency count of 1 was generated. Coder training included reading the manual, co-coding with a master coder (1 to 3 interactions), completion of a practice set followed by feedback, and independent completion of a reliability set. Coders were unaware of risk or outcome group status. All coders achieved our a priori intra-class correlation coefficients (ICC) threshold of .70 (M = .93; Range: .77 to .99) and received periodic unannounced reliability checks. Periodic checks of reliability consistently revealed ICCs above .70 (M = .94; Range: .73 to .99).

The coded interactions were post-processed using Python script (Version 2.7) to create frequency totals for observed instances of (1) core competency maternal play behaviors (i.e., look face, positive affect, and vocalizations); (2) maternal multimodal play bids; and (3) maternal contingent responsiveness. Multimodal composites were generated sequentially such that an overlap in the first target behavior (e.g., look face) and second target behavior (e.g., positive affect) were required to count as a multimodal bid/response as established in Schwichtenberg et al. (2019). To index contingent maternal responsiveness, theory driven composites were adapted from Feldman (2003, 2007) to assess dyads’ temporal relationships as continuous variables as established in Kellerman et al. (2019, 2020). All qualifying, sequentially ordered play behaviors are summarized in Supplemental Material (S.1).

Results

All data were checked for normality prior to analyses and cleaned using IBM Statistical Package for the Social Sciences (SPSS) Version 26. For variables with skewness and/or kurtosis of 1.5 or greater, a series of transformations were conducted. To account for multiple comparisons, the Benjamini-Hochberg Procedure (1995) was applied to control for false discovery rates at 0.20 for each aim, respectively. Pairwise comparisons (with Bonferroni adjustments) were conducted within SPSS for significant omnibus tests. Significant differences were interpreted if p < .05.

Aim 1

A series of general linear models, with maternal education as a covariate, revealed significant group differences across the SRS-2 Total T-Score and the subsequent six domains. Further inspection revealed that mothers in the high-FA group reported significantly more social difficulties compared to both moderate-FA and low-FA groups, across Total T-Scores, as well as the Social Communication and Restricted, Repetitive Behavior subscales. Results are summarized in Table 2. See Figure 2 for distribution of SRS-2 T scores across all subscales.

Table 2.

Maternal SRS-2 Scores Stratified by Family Affectedness Groups

	High-FA1 (n = 22)	Mod-FA2 (n = 12)	Low-FA3 (n = 37)	F	p	ηp2
Mean(SE)
Total T-Score	51.35(1.3)	45.64(1.7)	43.93(1.0)	10.12**	<.001	.23
Domain Descriptives, Mean(SE)
Social Awareness	49.84(1.2)	50.98(1.5)	46.27(0.9)	4.76*	.01	.12
Social Cognition	48.78(1.4)	44.91(1.8)	43.30(1.1)	4.77*	.01	.13
Social Communication	50.71(1.3)	44.90(1.7)	43.88(1.0)	8.15*	.001	.20
Social Motivation	53.44(1.8)	47.84(2.4)	46.04(1.4)	5.02*	.01	.13
Social Communication Index	50.80(1.3)	46.05(1.8)	44.05(1.0)	7.74*	.001	.19
Restricted and Repetitive Behaviors	52.80(1.2)	44.96(1.6)	43.94(0.9)	17.23**	<.001	.34
	Mean Difference (SE)				p	95% CI
Estimated Marginal Mean 4
Total T-Score
High-FA x Mod-FA	5.71(2.1)*				0.03	0.51, 10.92
High-FA x Low-FA	7.42(1.7)**				< .001	3.36, 11.51
Mod-FA x Low-FA	1.71(2.0)				1.00	−3.14, 6.55
Social Awareness
High-FA x Mod-FA	−1.14(1.9)				1.00	−5.84, 3.56
High-FA x Low-FA	3.57(1.5)				0.06	−0.12, 7.27
Mod-FA x Low-FA	4.72(1.8)*				0.03	0.34, 9.09
Social Cognition 5
High-FA x Mod-FA	3.87(2.3)				0.36	−0.04, 0.19
High-FA x Low-FA	5.49(1.8)*				0.01	0.02, 0.20
Mod-FA x Low-FA	1.61(2.1)				1.00	−0.07, 0.15
Social Communication 5
High-FA x Mod-FA	5.81(2.2)*				0.03	0.02, 0.78
High-FA x Low-FA	6.82(1.7)*				.001	0.18, 0.77
Mod-FA x Low-FA	1.01(2.0)				1.00	−0.28, 0.43
Social Motivation
High-FA x Mod-FA	5.60(3.0)				0.20	−1.76, 12.96
High-FA x Low-FA	7.40(2.4)*				0.01	1.61, 13.19
Mod-FA x Low-FA	1.80(2.79)				1.00	−5.05, 8.66
Social Communication Index
High-FA x Mod-FA	4.74(2.2)				0.10	−0.63, 10.12
High-FA x Low-FA	6.74(1.7)*				0.001	2.52, 10.97
Mod-FA x Low-FA	2.00(2.0)				0.99	−3.00, 7.01
Restricted and Repetitive Interests
High-FA x Mod-FA	7.84(2.0)*				0.001	2.98, 12.70
High-FA x Low-FA	8.86(1.6)**				< .001	5.04, 12.68
Mod-FA x Low-FA	1.02(1.8)				1.00	−3.50, 5.54

Note.

¹High-FA = elevated-risk group, family history of 2 or more children exhibiting autism symptoms.

²Mod-FA = elevated-risk group, family history of 1 child diagnosed with ASD.

³Low-FA = low-risk group, no endorsed ASD symptoms in target child. Table includes SRS-2 subscale T-scores. Maternal education in years was included as a covariate. No significant covariates were observed.

⁴Estimated marginal means for significant group differences are reported, with Bonferroni adjustments.

⁵Transformed for analyses. Interpretation: Higher SRS-2 scores are associated with more social responsive difficulties. Grey shading provides a visualization of where significant group differences were present across the three groups.

^*p < .05,

^**p < .001

Figure 2. SRS-2 Total Score Distribution Stratified by Family Affectedness Groups.

Note. SRS-2 Total T Score ranges: Normal (≤ 59), Mild to Moderate (60–75), Severe (≥ 76).

* p < .05, ** p < .001

Aim 2

To assess if self-reported indices of social responsiveness on the SRS-2 were associated with concurrent maternal play behaviors during a free play task, a series of general linear models were conducted with terms for family affectedness, infant sex, and maternal education in years. The models were specified with all available data across the three levels of play behaviors: base prosocial behaviors, multimodal bids, and contingent responsiveness. Overall, mothers self-reported SRS-2 total or domain scores were not associated with any of our observed maternal play behaviors between infants’ 24 to 36 month laboratory visit (p > .05). Thus, maternal SRS-2 scores were not a significant predictor. Results are summarized in Table 3.

Table 3.

Concurrent Maternal Play Behavior Associations with Self-reported Social Responsiveness

Overall (N = 631)	SRS Total T-Score		Social Awareness		Social Cognition		Social Communication		Social Motivation		Social Communication Index		Restricted Repetitive Behavior
	F	ηp2	F	ηp2	F	ηp2	F	ηp2	F	ηp2	F	ηp2	F	ηp2
Base Prosocial Behaviors
Look Face	.011	.000	.070	.001	.003	.000	.079	.001	.562	.010	.038	.001	.001	.000
Positive Affect	340	.006	.186	.003	.002	.000	.905	.016	.224	.004	.451	.008	.007	.000
Vocalizations	.093	.002	.003	.000	.003	.000	.013	.000	.520	.009	.122	.002	.009	.000
Multimodal (MM) bids
Total MM Composite	.072	.001	.037	.001	.008	.000	.388	.007	.000	.000	.054	.001	.000	.000
Look Face-led MM Composite	.067	.001	.090	.002	.070	.001	.008	.000	.231	.004	.092	.002	.050	.001
Positive Affect-Led MM Composite	.000	.000	.012	.000	.601	.010	.121	.002	.001	.000	.000	.000	.033	.001
Vocalization-Led MM Composite	.454	.008	.038	.001	.120	.002	.570	.010	.429	.007	.416	.007	.104	.002
Contingent Response
Responsiveness Composite	.437	.008	2.099	.036	.245	.004	.289	.005	.315	.006	.546	.009	.024	.000

Note.

¹Eight dyads did not have parent-child data at one of the qualifying visits. Reasons for missed data are included in the task description in the Methods section. No significant covariates were observed.

^*p < .05,

^**p < .001

Post-hoc Analyses

To aid in our interpretation of the results, a series of post-hoc analyses were conducted. First, recognizing that the SRS-2 is an index of reported social difficulties, observable differences may only be apparent in those endorsing elevated/clinical level SRS-2 scores (rather than the full continuum). To test this, an additional series of general linear models, were conducted with the top quartile vs. the bottom three quartiles in SRS-2 scores for the current sample (N = 71). The top quartile of reported SRS-2 Total T Scores were scores of 50 or higher (n = 16; Range = 50 to 67) in this sample. These models were specified based on all available SRS-2 data (regardless of family affectedness group) across the three core competency behaviors, five multimodal composites, and contingent responsiveness composite, respectively. Of the models specified, only two models reached our interpretation threshold of p < .05. Overall, mothers endorsing significantly more SRS-2 social communication difficulties were less likely to exhibit multimodal-led bids, t(15) = −2.85, p = .015, η_p² = .404, 95% CI [−13.15, −1.75]. Further inspection revealed this association to be primarily driven by fewer instances of vocalization-led multimodal bids, t(15) = −2.21, p = .047, η_p² = .289, 95% CI [−3.90, −.03].

Second, to provide a more cohesive representation of family dynamics relevant to reported SRS-2 indices of responsivity, a series of exploratory descriptive statistics were examined. Given our unique group classifications for the high-affectedness group, it’s conceivable that ‘true’ multiplex families (i.e., two or more children with ASD) may have driven our reported findings. Such that mothers from ‘true’ multiplex families may self-endorse higher rates of social difficulties on the SRS-2 compared to our working definition of high-FA. However as summarized in Table 4, the eight traditionally defined multiplex mothers from the high-FA group reported comparable social responsiveness scores to the high-FA group.

Table 4.

Mothers’ SRS-2 T-scores Stratified by Family Affectedness^a and Multiplex^b

	Low-FA1 (n = 37)	Mod-FA2 (n = 12)	High-FA3 (n = 14)	Multiplex4 (n = 8)
Total T-Score5, M(SD)	43.86(4.0)	45.67(4.1)	51.86(9.7)	50.75(6.6)
Domains, M(SD)
Social Awareness	46.22(4.1)	49.91(7.0)	52.64(6.4)	45.13(5.5)
Social Cognition	43.27(5.0)	48.82(8.7)	49.00(10.0)	48.50(6.3)
Social Communication	43.84(3.9)	50.77(9.1)	51.42(9.6)	49.63(8.6)
Social Motivation	45.84(6.8)	53.73(11.3)	53.64(12.5)	53.88(9.7)
Social Communication Index	43.97(4.3)	50.91(8.8)	51.64(9.7)	49.63(7.5)
Restricted and Repetitive Behaviors	44.05(4.1)	52.64(7.8)	51.86(9.4)	54.00(3.5)

Note.

^afamily affectedness reflects a dimensional /continuous approach to ASD symptoms in the family,

^bmultiplex reflects the number of individuals diagnosed with ASD in the family (i.e., more than one person diagnosed),

¹Low-FA = low-risk group, no family ASD history.

²Mod-FA = elevated-risk group with a family history of 1 child diagnosed with ASD.

³High-FA = elevated-risk group with a family history of 2 or more children exhibiting autism symptoms- now excluding the eight families with 2 or more children with ASD-specific diagnoses.

⁴Multiplex= traditional multiplex group of the eight high-risk families with 2 or more children with ASD-specific diagnoses.

⁵SRS-2 Total T Score ranges: Normal (≤ 59), Mild to Moderate (60–75), Severe (≥ 76).

And finally, to ground our findings within existing familial risk play behavior research and to inform the null results from Aim 2, we assessed maternal approaches to play to determine whether distinct play differences exist by household incidence. First, we ran a series of general linear models, with terms for maternal education and infant sex, by traditional elevated/high v. low-risk infant sibling status. Overall, no significant high v. low-risk status differences were evident across the base prosocial behaviors, multimodal bids, and contingent responsiveness. Second, we ran a series of general linear models with terms for maternal education and infant sex by family affectedness (high-FA, moderate-FA, and low-FA). In keeping with the risk status results, mothers did not exhibit significantly different use of prosocial, multimodal, or contingent play behaviors as a function of FA-type. Null results are summarized in Table 5.

Table 5.

Maternal Play Behaviors Stratified by Risk Status and Family Affectedness

	Elevated-Risk (n = 29)		Low-Risk (n = 34)	F	ηp2
Base Prosocial Behaviors, Mean(SE)
Look Face	5.28(0.6)		5.62(0.5)	0.174	.003
Positive Affect	1.71(0.4)		2.24(0.3)	1.100	.018
Vocalizations	10.06(0.8)		11.33(0.7)	1.275	.021
Multimodal (MM) bids, Mean(SE)
Total MM Composite	18.49(1.7)		19.11(1.5)	0.073	.001
Look Face-led MM Composite	4.73(0.5)		5.02(0.5)	0.157	.003
Positive Affect-Led MM Composite	1.68(0.3)		1.83(0.3)	0.126	.002
Vocalization-Led MM Composite	6.80(0.6)		7.08(0.6)	0.112	.002
Contingent Response, Mean(SE)
Responsiveness Composite	8.17(0.5)		9.00(0.5)	1.256	.016
	High-FA1 (n = 18)	Mod-FA2 (n = 11)	Low-FA3 (n = 34)	F	ηp2
Base Prosocial Behaviors, Mean(SE)
Look Face	5.34(0.7)	5.20(0.9)	5.62(0.5)	0.091	.003
Positive Affect	1.78(0.5)	1.60(0.6)	2.24(0.3)	0.571	.019
Vocalizations	9.51(1.0)	10.94(1.3)	11.33(0.7)	1.019	.034
Multimodal (MM) bids, Mean(SE)
Total MM Composite	18.05(2.1)	19.19(2.7)	19.12(1.5)	0.093	.003
Look Face-led MM Composite	4.80(0.7)	4.62(0.8)	5.02(0.5)	0.092	.003
Positive Affect-Led MM Composite	1.73(0.4)	1.60(0.5)	1.83(0.3)	0.086	.003
Vocalization-Led MM Composite	6.68(0.8)	7.00(1.0)	7.08(0.6)	0.087	.003
Contingent Response, Mean(SE)
Responsiveness Composite	7.56(0.8)	9.29(1.0)	8.96(0.6)	1.259	.042

Note.

¹High-FA = elevated-risk group, family history of 2 or more children exhibiting autism symptoms.

²Mod-FA = elevated-risk group, family history of 1 child diagnosed with ASD.

³Low-FA = low-risk group, no family ASD history. No significant covariates for maternal education or infant sex were observed

^*p < .05,

^**p < .001

Discussion

In this study, maternal self-reported social difficulties were elevated for mothers raising multiple children with autism symptoms, wherein affected is defined by the presence of ASD, or elements of the BAP. However, even with these elevated scores, the majority of these scores still fell within the normative range. In addition, the present study did not find distinct differences in mothers’ play behaviors by levels of family affectedness or robust associations between observed and self-reported indices of social responsiveness.

Elevated SRS-2 scores in the present study are distinct from previous prospective studies of family incidence, which generally report elevated scores for multiplex families (see review, Rubenstein & Chawla, 2018). However, the distribution of ASD traits in families often do not follow clear diagnostic lines and optimal support may be offered by taking a more dimensional approach to ASD features rather than a dichotomous diagnostic approach. In addition, traditional multiplex/simplex models may miss opportunities to support some families (e.g., simplex families may have members who do not meet full diagnostic criteria). In the present study, 77% of our high-FA sample could have been classified as simplex if elements of the BAP were not included in the definition of family affectedness. Importantly, the eight families raising two or more children with ASD diagnoses did not appear to ‘drive’ the reported relationship between family affectedness and maternal social communication scores.

Even with the significant SRS-2 score differences in the present study, it is important to note the majority of maternal total scores still fell within the normative range (≤ 59) regardless of group membership. Previous studies of family affectedness do not commonly report elevated SRS scores in mothers (Rubenstein & Chawla, 2018; Schwichtenberg et al., 2010) and it is not our intention to pathologize normative variations in social communicative competence. However, this variability (even when it is well below the clinical range) may still inform who might benefit from additional support in a PMI context. For example, Parr and colleagues (2015) noted that mothers with more BAP traits did not exhibit the same progress as mothers with fewer BAP traits within their PMI. However in Parr et al. (2015), BAP traits were indexed using a family history interview; therefore, direct comparisons to the current study are not possible.

Our second research question addressed how maternal self-reported social responsiveness scores relate to observed play behaviors. In the present study, we expected social communitive behaviors indexed via a questionnaire and observation would be related; however, no significant associations were initially observed. This may reflect a disconnect between the questionnaire content and our observed social constructs. For example, on the SRS-2 indices of social responsiveness often refer to situational statements with qualitative descriptors (e.g., “Her behavior is socially awkward, even when trying to be polite.” “People think she is interested in too few topics, or that she gets too carried away with those topics.”). Whereas in the present study, our observed indices of social responsiveness focused on the occurrence rates of core competency behaviors, use of multimodal bids, and contingent responsiveness. It’s also conceivable that associations between maternal play behaviors and self-reported SRS-2 scores are only apparent in individuals scoring within the higher/more impacted range for social difficulties on the SRS-2. The predominantly normative range of SRS-2 scores in our sample may have contributed to an overall pattern of no/minimal relationship between SRS-2 scores and observed play behaviors. In other words, a full range of social difficulties were not reported in this sample, thus we may not have been able to ascertain whether more impacted individuals exhibit fewer social bids/responses in context. Our post-hoc quartile-based analyses support this notion, such that even with the restricted range of our SRS-2 scores, some dyadic differences emerged for multimodal bids in the top quartile of highest SRS-2 Total T Scores. Specifically, mothers in the top 25% who endorsed significantly more SRS-2 social communication difficulties were less likely to exhibit multimodal-led bids, especially vocalization-led multimodal bids. Although, the overall lack of robust model significance may also reflect relatively low power present in this study.

Clinical Applicability

Prospective infant sibling designs are well suited for exploring indices relevant to various levels of family affectedness. Our investigation of affectedness included families raising multiple children exhibiting autism symptoms, as opposed to traditional family diagnosis-based distinctions. Though, in part, a functional decision based on our sample distribution, this study recognizes that diagnostic status may not be the only meaningful index when supporting individuals with social communication difficulties - an important point for clinicians and researchers to reflect on. In addition, although mothers’ self-reported significant differences in social responsivity on the SRS-2, our post-hoc play behavior analyses did not differentiate levels of family affectedness. The latter finding aligns with a majority of previous research noting comparable ranges of social communicative behaviors in caregivers (see review Wan et al., 2019), which is especially relevant to prospective early interventions training parents as interventionists.

To date, prospective PMIs utilizing dyadic play interactions to measure secondary treatment effects have relied heavily on parent social responsiveness-oriented constructs to promote infants’ social development (e.g., caregiver responsiveness- Davis et al., 2022; parent synchrony- Green et al., 2017; caregiver non-directedness and sensitive responding- Whitehouse et al., 2019). However, family dynamics such as perceived parenting stress, depression, family chaos, adaptability, cohesion, and social responsivity may be especially pertinent considerations prior to teaching caregivers as interventionists. Copious research has identified higher rates of parenting stress and depression in families with children already diagnosed (see review Hayes & Watson, 2013). However, it is probable families may have less access to supports pre-diagnosis and that family factors may play a larger role in how to maximize treatment effects. In fact, few studies have examined associations between the aforementioned family dynamic measures with observed play behaviors in either a developmental monitoring (e.g., parenting stress and synchrony- Steiner et al., 2018) or intervention context (e.g., depression and interactive style- Mahoney & Solomon, 2020). In the present study, the lack of consistent associations between maternal SRS-2 scores and their behaviors in context did not robustly support the use of the SRS-2 to identify mothers who may benefit from additional support implementing PMIs. Collectively, understanding the potential role of diverse family dynamics are not only important for continued specificity of social behaviors in developmental monitoring contexts, but also to clinicians providing early intervention services to children with or at-risk for ASD. Identifying specific sources of clinically relevant concerns and/or family factors prior to parent training, as well as parent perceptions on implementation may meaningfully inform current prospective intervention strategies/supports to ultimately support child developmental potential (as discussed in Amsbary et al., 2020; Greenlee et al., 2018).

Limitations

Prospective infant sibling designs provide unique sampling, measurement, and analytical challenges, as highlighted throughout the discussion. As illustrated in Figure 1, the present study recruited 45 elevated-risk infants but functionally very few infants met criteria for an ASD diagnosis by their outcome visit (n = 5). The majority of developmental concerns were for infants exhibiting subclinical ASD symptoms (n = 17) or some other developmental concern (e.g., language concerns; n = 12). The distribution of developmental outcomes in the present study are not uncommon in prospective monitoring studies, which contributes to a larger field limitation when trying to decipher the clinical implications of a study’s findings. Continued multisite collaborations are needed to robustly explore more nuanced family incidence dynamics. In addition, this study’s overall focus is on maternal behaviors, yet it is conceptually challenging to disentangle infant and parent behaviors within dyadic interactions. One concern, when only studying half of a dyad (mothers) in which the other half experiences social difficulties (e.g., infant with autism symptoms), is that infant influences and/or lack of engagement during play limit the opportunities for parent-led initiations and responses.

Future Direction

Recognizing the increasing demand for prospective PMIs, there is a crucial need to quantify both infant and parent social communication competencies and progress within a dyadic context. To continue supporting families with a history of ASD, future studies may build upon these findings by examining the associations between different types of parental indices of social communication, family dynamics, and other common behavioral targets used in intervention settings (e.g., directedness, initiations) to measure treatment effects. For example, the majority of parent/adult indices of autism traits tend to rely on self- and/or partner-report of social difficulties, such as the SRS-2, which may inherently include informant bias (De la Marche et al., 2015). Including parental measurements with observational and/or structured interview components, such as the BPASS, coupled with comparably measured dyadic behaviors in context, may be better aligned to systematically assess whether additional support is warranted. Ultimately, future prospective monitoring efforts are well suited to build on these findings to optimize child developmental potential, while continuing to minimize unnecessary family burden/distress in patient-centered early intervention efforts.