Measuring Parent-Child Transactions for Early Identification of Autism Spectrum Disorder

Background: Several early behaviors of ASD can be identified when a child is engaging with their parent (Baranek, 1999), and Wan et al., (2013) suggest that using the context of parent-child engagement to assess ASD risk may be a more sensitive reflection of early atypical behavior compared to structured measures given in unfamiliar contexts. Transactions are a specific aspect of parent-child engagement that are promising in assessing for signs of ASD. Transactions are coordinated turn-taking behaviors where the behavior of one partner is influenced by the previous behavior of the other partner in a reciprocal manner over time. Features of transactions, such as the mean length (i.e., mean number of turns) or miscues (i.e., partner passively observing without providing a physical or communicative indication to the other partner), may offer a more comprehensive view into the atypical interaction behavior associated with ASD, and thus aid in earlier identification of the disorder.

Objectives: The study objectives are to determine if (1) the mean length of transaction at one-year of age is predictive of later ASD symptomatology at preschool age, and (2) the proportion of parent responses followed by child miscues at one-year of age is predictive of later ASD symptomatology at preschool age.

Methods: This study is using extant data from a study of young children with elevated symptoms of ASD screened at 12 months (Watson et al., 2017) with the First Year Inventory 2.0 screener (FYI; Baranek et al., 2003). Thirty parent-child free play videos from when the children were 13-16 months old will be coded for transactions using the Transactional Engagement Coding-Extended (TEC-E), adapted from Yoder et al., 2010. Videos will be coded on ProcoderDV™ (Tapp, 2003) software using continuous timed event coding. Reliability will be determined by using point-by-point agreement with a criteria level of .75 for ICC. Measures from when the children were 13-16 months old (TEC-E) and preschool age (ADOS classification and severity) will be used in ANOVA and regression analyses.

Results: Previous preliminary ordered probit regression analyses using data from 30 parent-child videos indicated that a high occurrence of miscues within transactions were associated with a later ASD diagnosis on the ADOS at preschool age (p=0.041). For this study, additional planned ANOVA and regression analyses will examine the association of mean length of transactions and proportion of miscues with the classification and severity of ASD at preschool age. The length of transactions and proportion of miscues are theorized to predict later ASD symptomatology.

Conclusions: Preliminary analyses show promising evidence that measuring transactions can aid in identifying ASD risk, and that miscues specifically may be an early indicator of ASD. Full results of the study will further add to evidence supporting the practice of assessing ASD risk in the context of parent-child engagement.