Psychometric Properties, Classification, and Clinical Correlates of the Children's Communication Checklist – 2nd Edition in Autism Spectrum Disorder

Background: The Children’s Communication Checklist-2^ndEdition (CCC-2; Bishop, 2006) is a widely used measure of structural and pragmatic language that has been shown to capture the social communication deficits of autism spectrum disorder (ASD). The CCC-2 is more sensitive to the pragmatic language impairments of ASD than the Test of Pragmatic Language (Volden & Philips, 2010), and distinguishes ASD from other disorders such as Attention Deficit Hyperactivity Disorder (Geurts et al., 2004) and Specific Language Impairment (Bishop, 2006). The CCC-2 consists of 10 subscales (A-J) of 7 questions each. Scales A-D measure structural language, E-H pragmatic language, and I-J behaviors specific to ASD. The General Communication Composite (GCC) and the Social Interaction Difference Index (SIDI) measure overall communication ability and special impairments not due to structural language problems, respectively. The original standardization sample from Bishop (2006) included just 62 participants with ASD, and internal consistency was not reported for any clinical population.

Objectives: Estimate the internal consistency of the CCC-2 scales in a large sample of individuals with and without ASD; provide descriptive data on CCC-2 performance in a subsample of ASD and TDC matched on age and race/ethnicity; determine the utility of the CCC-2 for classifying ASD, TDC, and Other (Rule Out ASD); assess relationships between CCC-2 scores and parent report/clinical observations of social impairment (Social Responsiveness Scale (SRS; Constantino, 2003) and ADOS calibrated severity subscores).

Methods: Parents/caregivers of children aged 6-18 provided CCC-2 data in the context of participation in several specific studies. Research reliable PhD-level clinicians classified children as having a diagnosis of ASD, Other (Rule Out ASD), or typical development (TDC). The analyzed sample consisted of 668 participants with summary scores; 473 with item-level scores, and 429 ASD/TDC matched on age, race, and ethnicity. Cronbach’s coefficient alpha was used to estimate reliability and unidimensionality, logistic regression for classification; Pearson correlations assessed relationships to other parent/clinician ratings.

Results: Subscale reliability ranged from .62-.82 in the ASD group and .67-.83 in the Other group (Table 1). Estimated marginal means for subscale, GCC, and SIDI scores differed significantly between the ASD and TDC groups (Table 2). Logistic regression controlling for age, nonverbal and verbal IQ, sex, race, and ethnicity revealed a significant contribution of GCC to distinguishing ASD from TDC (Beta=-.27, p<.001), increasing sensitivity from 88% to 97% and increasing specificity from 46% to 94%; there was no classification benefit of the GCC for distinguishing ASD from Other. In all three groups, the GCC and SIDI correlated negatively with SRS scores (all ps<.05). The ADOS severity subscore for social affect correlated negatively with GCC in the ASD group and SIDI in Other (ps<.05). Correlations with the ADOS repetitive behaviors/restricted interests severity subscore were not significant.

Conclusions: The CCC-2 is a valuable tool for assessing pragmatic language in ASD, with most subscales and the GCC reaching internal reliabilities of >.7. The GCC distinguishes between ASD and TDC (but not ASD and Other), and correlates with clinician/parent measures of social impairment. 195 additional item-level sets will be included by May 2016.