Dimensions of Autism Based on the CARS In Different Age and Ability Groups

Background:  Recently factor analytic studies have contributed substantially to the conceptualization of autism.  These have often been based on the ADI-R items (e.g. Snow et al., 2009) or domains (e.g., Georgiades et al., 2007), but occasionally other measures, such as the CARS (e.g., Magyar & Pandolfi, 2007). There is growing consensus that symptom profiles may vary among individuals of different ages and ability/language levels (Ben Itzchak, Lahat, Burgin, & Zachor, 2008; Dyck, Piek, Hay, & Hallmayer, 2007; McGovern & Sigman, 2005).    

The Childhood Autism Rating Scale (CARS; Schopler, Reichler & Renner, 1988; Schopler, Bourgondien, Wellman, & Love, 2010) is a behaviour observation scale rating 15 symptoms of autism. This measure has important clinical and research utility and is widely used in the diagnosis, treatment planning and evaluation of children with ASD.  However, only three studies in the literature report on the underlying factor structure of the CARS (DiLalla & Rogers, 1994; Magyar & Pandolfi, 2007; Stella et al., 1999) with 3 to 5 factors being described.  The inconsistent factor compositions and mixed results may be accounted for by sample characteristics, including: small sample size, differences in ages, developmental levels, and diagnoses.  Our research team reported a pilot study (Levy et al., 2009) which suggested different symptom factors in younger and older individuals.

Objectives:  The underlying factor structure of the CARS was investigated in a large sample (1000+ individuals) with an ASD from several different settings. This analysis was intended to determine whether, according to the CARS, there is symptom coherence across individuals with autism or whether there are distinct and independent groups of symptoms based on age and cognitive level.

Methods:  This is an ongoing file review study involving multiple settings across Ontario, Canada. The CARS was administered to participants within the context of routine clinical assessments by a team of trained clinicians. Further analyses by age, diagnoses, and cognitive level will be conducted with the larger sample.

Results:  Preliminary results suggest that for the older group (N= 130) the 15 items of the CARS load onto four factors (“Repetitive Behaviour”, “Social/Communication”, “Emotion Regulation”, “Cognitive level”), while results from the younger group (N= 828) load onto only two factors (“Autism”, “Emotional Regulation”). Analyses with a larger sample and with subsamples broken down by age group (2-6 years; 7-12; 13-18, over 18) and ability level (severe-profound, mild-moderate, borderline-average) are presently being conducted.

Conclusions:  Results to date suggest that the characteristics of autism, as measured by the CARS, may be different in adolescents and young adults than in younger children and we speculate that analyses will reveal different factor structure in individuals of different cognitive levels as well. In view of the forthcoming transition to the DSM-V and the widespread recognition of a broader group of individuals with ASD, these results may help define symptom constellations that clinicians should look for in different subgroups of people on the autism spectrum.