Video-Referenced Ratings of Reciprocal Social Behavior in Toddlers: A Twin Study

Background:  Reciprocal social behavior (RSB) is a developmental prerequisite for social competency, and severe deficits in RSB constitute a core feature of autism spectrum disorder (ASD). Although several clinical measures screen for ASD-related impairment in young children, feasibly quantifying RSB across its full range of variation in early childhood, when ASD first arises, remains an important priority, especially for large-scale developmental studies and for measuring incremental response to early intervention.  Since caregivers of preverbal children might have difficulty differentiating nuances of social communication, we developed a video-referenced system for rating early RSB (vr-RSB) in which a rater first views a 3-minute video clip depicting highly competent social behavior in a toddler, and then compares the subject he/she is rating with the child in the video clip, who serves as a scoring anchor for 13 quantitative video-referenced items on the scale.

Objectives:  To examine the psychometric properties of vr-RSB measurements at ages 18-24 months, their relationship to scores on screeners for ASD and language delay, and twin-twin correlations as a function of zygosity. 

Methods:  Parents of 162 epidemiologically-ascertained twins (MZ=20 pairs, DZ=61 pairs) rated their children using the vr-RSB and two developmental screeners, the Modified Checklist for Autism in Toddlers (M-CHAT), and the MacArthur Communicative Development Inventory Short Form (MCDI-s). At the time of this writing, test-retest data for the vr-RSB were available for 37 subjects.

Results:  In contrast to the screeners, scores on the vr-RSB were fully continuously distributed, with excellent 3-month test-retest reliability (ICC=0.76, p=0.000).  Monozygotic twins displayed markedly greater trait concordance than dizygotic twins, (MZ ICC=0.886, p=0.000, DZ ICC =0.197, p=0.063), suggesting substantial heritability of this measurement. Scores on the vr-RSB distinguished subjects with passing versus failing scores on the M-CHAT (t=-5.293, df=19.882, p=.000), and were inversely correlated with number of words produced on the MCDI-s (r=-0.272, p=0.001), showing an association between decreased deficits in RSB and increased expressive vocabulary.

Conclusions:  vr-RSB scores in toddlers exhibit a continuous, quantitative distribution, as has been observed for RSB throughout the lifespan. Rapid video-referenced ratings of RSB appear highly heritable, similar to ASD, and can distinguish toddlers at elevated risk of ASD.  The correlation of vr-RSB scores and language scores, demonstrated here in a normative population, suggests that further investigation of the relationship between the early capacity for RSB and language function is relevant for understanding behavioral development in both children with ASD and typically developing children. These results support the potential utility of video-referenced ratings for quantifying the capacity for early RSB in public health settings and monitoring incremental response to early intervention. These findings also have implications not only for promoting earlier identification of ASD, but also for elucidating the genetic-environmental structure of early RSB in typical and atypical development.