30253
Application of the Brief Observation of Social and Communication Change (BOSCC) to ADOS Assessments in the Preschool Autism Communication Trial
Commonly used outcome measures in autism intervention trials are often insensitive to subtle, yet meaningful, changes in the child’s social and communication behaviours. The BOSCC has been recently developed with the aim of providing a more sensitive tool. This study is among the first to retrospectively apply the adapted BOSCC-for-ADOS scheme to ADOS assessments from a previously conducted trial. The Preschool Autism Communication Trial (PACT) is the largest trial the BOSCC has yet been reported on.
Objectives:
To explore the psychometric properties of the BOSCC as applied to ADOS assessments in the Preschool Autism Communication Trial.
Methods:
The study included 152 children (aged 2-4 years 11 months) with a diagnosis of core autism. 117 children (77.0%) were ADOS module 1 and 35 children (23.0%) were module 2. The BOSCC was scored from recordings of the ADOS assessments from the PACT study. The BOSCC module 1 scheme was applied to all children.
Results:
Item distributions
Most social and communication items had a good spread of scores with some skew observed on items related to vocalisations, most likely on account of the characteristics necessary for eligibility. Restricted and repetitive behaviours (RRB), mannerisms and unusual sensory interests were skewed items, in line with previous research. The spread and overall pattern of distribution did not change when module 2 children were removed.
Inter-rater reliability
Four coders scored 15 videos. Intra-class correlations demonstrated high reliability across total scores: total score 0.93 (95% CI: 0.85-0.97), social and communication subscale 0.93 (0.84-0.97) and RRB subscales 0.82 (0.66-0.92).
Correlations – cross-sectional (baseline)
The BOSCC social communication subscale score strongly positively correlated with the ADOS communication and social score (r = 0.72, p <.001). The BOSCC RRB subscale was more moderately positively correlated with the ADOS RRB subscale (r = 0.46, p < .001). The BOSCC total score was weakly correlated with the ADOS CSS score (r = 0.33, p <.001).
Factor analysis
We conducted a confirmatory factor analysis using all Core items (Items 1-13) to test the two-factor model (SC and RRB) identified in previous literature. Our model fit was poor with root mean square error approximation (RMSEA) of 0.130 and comparative fit index (CFI) of 0.85. In line with previous literature, RRB items had lower factor loadings.
Correlations – change (TAU group only)
BOSCC total score change weakly correlated with ADOS CSS change (r = 0.35, p<.004). This is in line with previous research.
Sensitivity analyses
Analyses with module 2 participants removed did not change the interpretation of results.
Conclusions:
The BOSCC-for-ADOS when applied to a large group of children with core autism with minimal language, provides a good spread of scores across items, reliable coding and correlates in the expected directions with the ADOS in line with previous research. Potential reasons for the poor fitting factor analysis will be discussed.