Abstract
Social communicative precursors to autism spectrum disorder may influence how infants who are later diagnosed with autism spectrum disorder interact with their social partners and the responses they receive, thus bidirectionally influencing early social experience. This systematic review aimed to identify a developmental timeline for parent–infant interaction in the first 2 years of life in at-risk infants and in emergent autism spectrum disorder, and to examine any parent–infant interaction associations with later social-communicative outcomes. In total, 15 studies were identified investigating parent–infant interaction in infants at familial autism risk (i.e. with an older sibling with autism spectrum disorder). Starting from the latter part of the first year, infants at risk of autism spectrum disorder (and particularly infants with eventual autism spectrum disorder) showed parent–infant interaction differences from those with no eventual autism spectrum disorder, most notably in infant gesture use and dyadic qualities. While parental interactions did not differ by subsequent child autism spectrum disorder outcome, at-risk infants may receive different ‘compensatory’ socio-communicative inputs, and further work is needed to clarify their effects. Preliminary evidence links aspects of parent–infant interaction with later language outcomes. We discuss the potential role of parent–infant interaction in early parent-mediated intervention.
Keywords
Delays in social communication are one of the most striking early markers of autism spectrum disorder (ASD; see Szatmari et al., 2016). Such precursors to ASD have been distinguished in a number of studies using the prospective ‘at-risk’ approach of studying the infant siblings of children already diagnosed, as they are at increased ASD risk (Ozonoff et al., 2011). However, much less is known about whether these emergent differences in social communication (as measured using standardised tests in interaction with a researcher) impact on how they interact with their regular social partners, which may in turn alter the interactive behaviour of their partners. Such perturbations are seen in the parent-infant interactions (PII) of other groups, such as young children with Down syndrome (Blacher et al., 2013; Slonims and McConachie, 2006), and over the long-term, logically affect the child’s social experience and opportunities.
As social and language development requires and is facilitated by social-communicative input and scaffolding that is consistent with sensitive and mutual interactions (Landry et al., 1997; National Institute of Child Health and Human Development (NICHD) Early Child Care Research Network, 2001; Page et al., 2010), early alterations in PII (such as reduced interaction from parents) are likely to have a cascade of effects ultimately affecting child social and communicative development (Sameroff, 2009). Conversely, this same pathway may open up opportunities for early intervention via PII ‘enrichment’ in this group (Bradshaw et al., 2015; Szatmari et al., 2016).
Longitudinal ‘at risk’ studies have successfully identified a range of early behavioural markers that delineate those high-risk (HR) infants later diagnosed with ASD from those who are not, as well as from ‘low risk’ (LR) infants with no family history of ASD (see Szatmari et al., 2016). However, while these studies have mostly used standardised and experimental tests, consideration of the child’s naturalistic environment is likely to be pertinent to our developmental understanding of early ASD emergence, and perhaps especially so if there is social neurodevelopmental vulnerability. Methodologically, social competence may be more accurately or sensitively reflected within PII as a more familiar and natural social context relative to experimental measures. Studying the infant’s behaviour within PII allows us to see in operation their social, relational and communicative tendencies (as opposed to skills competence per se) in dynamic transaction with the opportunities or constraints provided by primary caregivers.
In this systematic review, we synthesise at-risk studies that have measured PII at-risk or in emergent ASD in an attempt to identify a developmental timeline for PII in the first 2 years of life, in particular distinguishing (1) HR infants from LR infants and (2) HR infants later diagnosed with ASD from those who were not. Second, we reviewed from these studies any evidence for an association between PII and later social and communicative developmental outcomes, consistent with possible cascading effects of PII on subsequent development in the HR group.
Methods
Search strategy and inclusion criteria
In our systematic analysis of studies on PII in infants at familial risk of ASD, an electronic search of PsychINFO and PubMED databases included English-language peer-reviewed journals up to 10 April 2017 using the following keyword (or ‘title and/or abstract’) search terms: (autism OR autistic) AND infants AND risk AND (interaction OR synchrony OR respons* OR sensitivity OR play). The search yielded 498 articles (Figure 1). The abstracts of 351 articles (after duplicate removal) were manually screened independently by two reviewers (J.S., M.W.W.) for their potential in meeting the following inclusion criteria: (1) ASD risk design – A study of infants who have an older sibling with ASD, which includes at least one comparison group in the analysis (HR infants compared with LR controls, or HR infants with subsequent ASD (HR-ASD) compared with those with no ASD (HR-no ASD); (2) Measurement – An empirical observational study of relatively naturalistic interaction (i.e. in free play or face-to-face play, with or without toys) between infant and a primary caregiver; (3) Infant age < 24 months or, if unspecified, no older than the ‘18-month visit’.
PRISMA flowchart.
The reference lists of the 25 articles were further searched, providing three additional papers which were screened to potentially meet criteria for inclusion. Regarding search reliability, reviewer agreement was 96%. The one disagreement was whether Talbott et al.’s (2016) study fulfilled the measurement eligibility criterion of free play; consensus agreement was to exclude as the paper combined data from two set-ups, of which book sharing was considered to be prespecified rather than free play interaction.
Of the 28 papers assessed, 16 met full inclusion criteria. Wan et al.’s (2012, 2013) papers involved largely the same sample and PII variables, so were treated as one study, resulting in 15 studies.
Coding procedures
The following information was extracted from the final 16 studies: publication source, sample characteristics (N in each group, age, N and age at diagnosis where relevant, ASD outcome criteria), PII characteristics (set-up/format, location, duration), PII measure characteristics (coding scheme, main variables measured and how), and results pertaining to PII effects by autism risk/outcome (HR vs LR, and/or HR-ASD vs HR-noASD) and statistical associations between a PII variable and later developmental outcomes in the full sample or by subgroup.
Results
Across the final 15 studies (Table 1), PII was observed in a total of 426 infants at familial risk of ASD (we include only the largest sample in this calculation if a different cohort was involved at each time point within a study). Most studies were cross-sectional. Of the six studies that repeated PII measurement, three captured individual longitudinal change. In total, 11 studies followed their HR sample up to full ASD or provisional diagnosis, of which 8 reported results by ASD outcome (i.e. HR-ASD, HR-noASD and LR). Samples sizes were a mean of 27.7 HR infants, with an infant age range of 5–18 months. PII set-ups varied between studies in location (home: six studies; lab: nine studies) and structure, including the kind of toys provided or allowed, positioning (e.g. floor play, face-to-face), and given instructions (though in most cases, caregivers were asked to play with their child as they would usually do at home). Interaction length varied (1–45 min), most commonly taking 5 min clips (six studies), and two studies analysed audio files only. Studies reported on infant (13 studies), parent (8 studies) and/or dyadic (5 studies) aspects of PII, and used microanalytic measures (12 studies), global rating measures (4 studies) or automatic acoustic analysis (1 study); 2 studies used more than 1 type. Five developmental areas were identified, the key findings of which are summarised below.
Key methods and results of reviewed studies.
ASD: autism spectrum disorder; PII: parent–infant interaction; HR: high risk; LR: low risk; DSM-IV-TR: Diagnostic and Statistical Manual of Mental Disorders, 4th Edition, Text Revision; G: global measure type; NICHD ECCN: National Institute of Child Health and Human Development Early Child Care Network scales; M: microanalytic measure type; Noldus: Noldus Observer Pro; LD: language delay.
1. Caregiver and dyadic/interactional qualities
Qualities of interaction in the dyad start to diverge possibly as early as 5 months as Yirmiya et al. (2006) reported that although synchrony among HR dyads was not more mother led than in LR dyads, coherence of synchrony was lower and tended to be below mid-range within interactions led by HR infants (Yirmiya et al., 2006). Of the two studies focused on both parent and infant PII variables in relation to later ASD, dyadic variables were found to predict ASD outcome, namely infant social reciprocity at 11–12 months (Campbell et al., 2015) and mutuality at 14 months (Wan et al., 2013). In Campbell et al. (2015), infant social reciprocity among the HR group, and not specific social behaviours as seen within interaction, predicted later severity of ASD markers (Campbell et al., 2015). In Wan et al. (2013), dyadic mutuality and two infant variables (attentiveness to parent, negative affect) predicted 3 year ASD status.
Two studies found that parents of HR infants are more directive towards their infants than parents of LR infants at 7 and 14 months (Wan et al., 2012, 2013) and at 9 months (Harker et al., 2016). Parent directiveness predicted slower growth in parent-directed smiles within PII from 9 to 18 months, while parental responsiveness predicted higher rates of concurrent parent-directed smiles (Harker et al., 2016). By contrast, parental responsiveness – both behavioural and emotional – remained remarkably unaffected among the HR group (Baker et al., 2010; Campbell et al., 2015; Wan et al., 2013). However, neither parent directiveness nor sensitive responsiveness predict ASD outcome (Wan et al., 2013).
2. Social communication
Delays in social communication become apparent towards the end of the first year. While parent-directed smiles usually decline sharply from 9 months, HR infants showed an increase (after controlling for maternal directiveness; Harker et al., 2016). This may reflect a HR group delay in the developmental shift from dyadic to triadic play interaction rather than increased positive affect or communication.
From 11 months, delays in the use of give and show gestures are seen among HR-ASD infants’ interactions (Campbell et al., 2015) and, from around 1 year, vocal delays are apparent in PII, as HR infants’ vocalisations tend to be less speech-like (Leezenbaum et al., 2014), less reminiscent of true babbling (Paul et al., 2011), and less often accompanied by gestures (Parladé and Iverson, 2015) than those of LR infants.
The amount that infants vocalised in the first year and that parents spoke, however, did not differ by risk status (Campbell et al., 2015; Northrup and Iverson, 2015; Parladé and Iverson, 2015; Rozga et al., 2011; Talbott et al., 2015). Furthermore, Talbott et al. (2015) reported no 12-month ASD outcome group differences in the proportions of gesture types used. Harder to explain is that Talbott et al. (2015) found that HR-ASD infants showed more different meanings conveyed through a gesture than did HR-noASD and LR infants at 12 months.
By the first half of the second year, HR delays in gesture frequency and sophistication are consistently found (Leezenbaum et al., 2014; Parladé and Iverson, 2015; Talbott et al., 2015; Winder et al., 2013) and is most marked in those with an eventual diagnosis of ASD (Parladé and Iverson, 2015; Talbott et al., 2015). As HR infants head towards 18 months, longitudinal studies reveal that they were slower to adopt combined vocal-gesture communication (Parladé and Iverson, 2015) and to decrease production of non-word vocalisations (Leezenbaum et al., 2014).
HR delays in the integration of infant gesture use did not necessarily adversely affect parental interactions (Campbell et al., 2015; Leezenbaum et al., 2014). Preliminary evidence raises the suggestion that parents may even use compensatory strategies in response to (knowledge of) their infant’s risk and/or emerging language difficulties by increasing their verbal labelling of infant give/request gesture referents at 13 and 18 months (Leezenbaum et al., 2014). Furthermore, Talbott et al. (2015) reported that mothers of HR-noASD infants produced significantly more gestures, and significantly more different meanings conveyed through gesture when their infant was 12 months old than mothers of HR-ASD or LR infants.
3. Emotional expression
HR infants (and HR-ASD infants) may become increasingly less positive in affect in their interactions over time. A non-significant trend towards more negativity was found at 6 months in HR infants (Cassel et al., 2007), while HR-ASD infants specifically were more negative at 14 months but not at 7 months (Wan et al., 2013).
Furthermore, parent-directed or ‘social’ smiles in HR infants did not differ from those of LR infants at 6 months (Rozga et al., 2011), but between 9 and 18 months, such smiles increased in the HR group (Harker et al., 2016). This latter finding may reflect increased dyadic interaction (i.e. gaze to parent) among the HR group rather than increased positive affect per se (see ‘Social communication’ section).
4. Vocal features
Studies found no risk group differences in vocal frequency or coordination characteristics in the first year, including interactive partner’s vocal response lag (Northrup and Iverson, 2015; Paul et al., 2011). However, HR infants produced fewer speech-like vocalisations, which did not increase between 6 and 12 months unlike the distinct growth observed among LR infants (Paul et al. (2011). This HR delay is also reflected in a slower decrease in non-speech production, lower 9-month use of canonical syllable shapes (associated with true babbling), and (particularly in HR-ASD infants) lower use of various consonant types (Table 1).
At 12 and 18 months, LR but not HR dyads showed vocal prosodic matching (Quigley et al., 2016). Furthermore, mean pitch decreased among LR group mothers from 12 to 18 months, which may reflect a move away from infant-directed (or ‘motherese’) speech, while the mothers of HR infants increased their pitch on average over the same period. However, a number of other vocal prosody characteristics measured did not differ by risk group.
5. Motor and sensory behaviours
A small study found that HR infants showed less movement and object manipulation than LR infants at 12 months, but no differences in sensory, play, or other motor behaviours, including stereotypies and aversive responses (Mulligan and White, 2012). Wan et al. (2012) reported less liveliness at 7 months in HR infants compared with LR infants, but this did not persist to 14 months and was unrelated to later ASD outcome (Wan et al., 2013).
PII by ASD outcome
Just over half of the reviewed studies (8/15) analysed PII by subsequent ASD outcome, of which five studies required participants meet full DSM-IV criteria for ASD at 36 months (assessed by clinical judgement combined with standardised measurement) for a positive ASD outcome, totalling 48 infants (Table 1). Of these studies, a number of positive findings were reported but with little overlap in the PII variables measured.
Compared with HR-noASD children, HR-ASD children were less socially reciprocal and showed fewer show/give gestures at 11 months (Campbell et al., 2015), slower growth in communicative combinations at 12 months (Parladé and Iverson, 2015), and more negative affect and less attentiveness to their parent at 14 months (Wan et al., 2013). In Winder et al. (2013), HR-ASD infants (n = 3) showed fewer communication initiations and vocalisations from 13 months. By contrast, no group differences emerged in the infants’ degree of positive affect or parent-directed vocalisations (Campbell et al., 2015; Parladé and Iverson, 2015).
Three studies in which ASD outcome was more provisional in part or all of their sample, reported that HR-ASD infants vocalised fewer of specific types of consonant that varied across the first year (Paul et al., 2011) and conveyed fewer meanings through a single gesture at 12 months (Talbott et al., 2015), but showed no 6-month differences in gaze, smiles or vocalisation to their mother within PII at 6 months (Rozga et al., 2011).
Parents of HR-ASD infants have shown no significant differences in interactive behaviour from HR-noASD infants despite the fairly large number of such variables studied (Baker et al., 2010; Campbell et al., 2015; Talbott et al., 2015; Wan et al., 2013). Talbott et al. (2015) reported that mothers of non-diagnosed HR infants gestured more frequently than mothers of LR infants (but not more frequently than mothers of HR-ASD infants). Mothers of HR-noASD infants also conveyed a higher number of meanings through each gesture.
PII in relation to non-ASD developmental outcomes
Six HR studies examined PII in relation to developmental outcomes other than ASD diagnosis (Table 1). Yirmiya et al. (2006) found no link between 5-month synchrony and 14-month cognition, language and non-verbal communication. However, others found that the number of early consonants infants used at 6 months (Paul et al., 2011) and gesture production at 12 months (Talbott et al., 2015) were positively associated with later infant language outcomes, while amount of 9-month infant speech interrupting or chiming in during maternal vocalisations predicted language delay status (Northrup and Iverson, 2015). Given the early age at which most of these PII measures were taken, it was not possible to control for baseline infant language scores.
Regarding parental interaction, Baker et al. (2010) found that sensitive structuring, a component of maternal sensitivity, at 18 months was related to 2- to 3-year expressive language change in infants who were later diagnosed with ASD. Also reporting on maternal responsiveness, Leezenbaum et al. (2014) found that the proportion of verbal labelling responses that mothers provided to their 13-month-old infant’s point or show gestures was associated with later (18 month) word production in the overall sample; however, such maternal responses to HR infant gestures were uncommon. Furthermore, Talbott et al. (2015) reported that maternal gesture use with their 12-month-old infants was associated with later child language in the LR and HR-no-ASD groups only.
Discussion
At-risk studies have made strides in demarcating how PII is affected in emergent ASD, focused mainly on the first half of the child’s second year. An emerging developmental picture reveals that children with eventual autism (and the broader phenotype) and their primary caregiver interact with each other in some ways that depart from a typical trajectory in the latter months of the child’s first year. The strongest evidence was found for preverbal communication delays, including gesture use, prelinguistic vocalisations and vocal-gesture coordination, mirroring trajectories reported in the wider HR literature based on experimenter-led structured interactions (Landa et al., 2007; Ozonoff et al., 2010; Yoder et al., 2009). A second consistent finding, based on two good quality studies, is that of lowered interactive reciprocity or mutuality, which one study further found predicted later ASD symptom severity better than did social behaviours generally in HR infants. Divergence in both interactive reciprocity and infant preverbal communication are apparent in the few months after the first birthday, although delays in gesture use do not seem to affect parental interactive behaviour.
Weaker evidence based on single studies report that during PII, infants with eventual ASD demonstrate less attentiveness to parent and more negative affect, and HR infants (more generally) exhibit less movement, less object manipulation, and less advanced vocal characteristics than LR infants. These findings are consistent with atypicalities found outside of PII’s at 1 year predictive of ASD outcome (e.g. Clifford et al., 2013; Ozonoff et al.,2010, 2008); though the evidence for motor delay is more mixed (see Jones et al., 2014). Taken together, these PII studies demonstrate that these infants’ early atypicalities are measurable within play interactions with caregivers, and that uniquely PII features (i.e. reciprocity, mutuality) may be affected.
Generally consistent with studies of parents of older children with ASD (e.g. Siller and Sigman, 2002; Van Ijzendoorn et al., 2007), a number of studies reviewed consistently found no interactive differences in parents of infants with eventual ASD from those of infants without eventual ASD. Irrespective of ASD outcome, HR infants may tend to be have different social opportunities or inputs from LR controls, such as reflected in a more directive style and less prosodic matching. Thus, while some HR parents appear to adjust to possible infant atypicalities by decreasing social input, others may use compensatory behaviours in attempting to enrich the infant’s social environment (e.g. retaining infant-directed speech in later infancy, increasing gesture use or verbal labelling of infant gestures); such impacts on infant outcomes and later parent–child interactions require further investigation.
Our secondary aim was concerned with the prediction of later social and communicative outcomes from PII in infants with emergent ASD. Language outcomes in toddlerhood were associated with various communicative delays or differences within PII in the first year. Whether such early differences shape subsequent social experiences that are important for language development, or whether they simply reflect a continuity in neurodevelopmental delay is unclear. Other single PII studies suggest that parental sensitive structuring (or scaffolding) may optimise child expressive language growth in those with eventual ASD, and that parental labelling in response to HR infant object-sharing gestures may promote word production. Thus, PII study findings seem to differentiate the parental provision of a communicative framework that may be quite highly structured (social scaffolding aimed to enhance social opportunities) from parental ‘directive’ behaviours (as defined in PII studies to include intrusive and demanding behaviours), where our review suggests that, at least in infancy, the former enhances, and the latter reduces, socio-communicative response.
Methodological limitations
Despite advances in understanding PII in HR infants, it remains unclear whether such differences by subsequent ASD outcome are the result of cascading effects on ASD severity rather than simply reflecting emergent ASD. Drawing clear conclusions is further limited by the heterogeneity of studies, most notably in the age of infant samples and the PII variables measured. Some studies included fathers whose PII’s may tend to differ from those of mothers (e.g. Feldman, 2003). Studying PII in HR infants is resource intensive; this is reflected in the sample sizes of most studies, a third of them involving fewer than 15 HR infants, and five studies following up to 3-year ASD diagnosis. Thus, significant effects may have been difficult to detect, and no studies reported effect sizes. Sample representativeness is unclear from study descriptions and a self-selection bias is likely. A likely publication bias must be considered when evaluating these studies, as positive results are more likely to be published than negative findings.
Clinical and research directions
This first review on PII in infants at autism risk demonstrates how early emerging atypicalities and delays may alter their social experiences and that of their parents’. The findings highlight the usefulness of brief videotaped PII as a context for measuring infant behaviours and vocalisations in a more naturalistic setting than experimental paradigms and offer the advantage of observing the interactive partner’s behaviour, partly affected by the infant, and the possible consequences of any interactive alterations. As yet, however, no HR studies explicitly test bidirectional effects within PII trajectories, and only one study tested PII effects to social-communicative development in eventual ASD (Baker et al., 2010). Measuring PII longitudinally in larger samples is needed to investigate the impact that parental strategies may have, positive or otherwise, on longer term outcomes.
The findings of this review suggest that very early intervention in infancy and early toddlerhood may target PII, which may be beneficial in itself, and for optimising social and communicative outcomes. Although most early interventions are parent-mediated and therefore alter PII, two approaches have been taken, one being primarily relational (works on ‘realigning’ PII to effect socio-communicative outcomes) and the other being a coaching strategy (which ‘directs’ social enrichment – thus altering PII – to effect socio-communicative outcomes).
Taking a primarily relational strategy, Green et al. (2015, 2017) demonstrated in their parent-mediated intervention using non-directive guidance with HR infants (who were selected on familial risk) that prodromal symptoms reduced, with effects on parental aspects of interaction (directiveness reduced and synchrony increased) and infant aspects of interaction (attentiveness to parent and communication initiation increased). Our review supports the benefits of targeting these PII areas. Other recommended targets are infant gesture use, dyadic reciprocity and (with less evidence) infant vocalisation delays.
Most early interventions use a primarily coaching approach with symptomatic infants and (mostly) toddlers and have demonstrated child social engagement and communicative improvements by training parents to apply developmentally appropriate behavioural principles to daily routines and play (e.g. Bradshaw et al., 2017; Kasari et al., 2010; Rogers et al., 2014). Using this approach, the PII dynamic is altered through the parent’s use of coached enrichment strategies. This approach lies somewhat at odds with the current findings, which suggest PII as an important focus for intervention along with preliminary evidence for parental sensitive structuring to support language development in HR-ASD infants. Thus, an important empirical question is raised: Where along the continuum from sensitive structured behaviours to directive behaviours may be most beneficial for enhancing, for example, shared attention and infant gesture use, and subsequent communicative and relational outcomes in HR infants? Parent-mediated interventions can serve as a vehicle to test for such causal mechanisms.
Supplemental Material
AUT777484_Lay_Abstract – Supplemental material for A systematic review of parent–infant interaction in infants at risk of autism
Supplemental material, AUT777484_Lay_Abstract for A systematic review of parent–infant interaction in infants at risk of autism by Ming Wai Wan, Jonathan Green and Jordan Scott in Autism
Footnotes
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
