Factor Analysis of the Childhood Autism Rating Scale in Two-Year-Olds with ASD

Background: The Childhood Autism Rating Scale (CARS) (Schopler, 1980) is a 15-item observation-based rating scale designed to accurately differentiate children with autism from those with developmental delays without features of autism. The CARS is a commonly used component of ASD diagnosis, yielding a single total score (15 to 60) that is reflective of overall symptom severity. Given that symptom severity varies across domains for children with ASD, the utility of the CARS may be extended by understanding factors that underlie its total score. At present, few factor analyses have been conducted on the CARS, and results are mixed.

Objectives: The current study seeks to extend existing research on the factor structure of the CARS utilizing a well-characterized sample of two-year-old children with a DSM-IV ASD diagnosis.

Methods: Developmental and diagnostic evaluations were conducted at approximately 26 months following positive screening on an autism-specific screener (MCHAT(-R)). The Autism Diagnostic Observation Schedule (ADOS), Mullen Scales of Early Learning (Mullen), Vineland Adaptive Behavior Scales (VABS II), and Childhood Autism Rating Scale (CARS) were completed. Diagnosis was assigned based on the clinical judgment of experienced clinicians. Children were included in the current study if they received a diagnosis of Autistic Disorder (AD) or Pervasive Developmental Disorder, Not Otherwise Specified (PDD-NOS) (N=289). 217 (77.8%) children were male, and the sample was predominantly white (N=204, 73.1%). A factor structure evaluation of the CARS was conducted using principal axis factor analysis with promax rotation. Factors were retained utilizing the Kaiser criterion (1960). Items were considered to load on a factor if their loadings were .40 or above in the pattern matrix.

Results: CARS scores ranged from 20.0 to 48.5 (M=32.26, SD=5.22). Mullen Early Learning Composite (ELC) standard scores ranged from 41 to 120 (M=59.72, SD=12.01), with 71.6% of children demonstrating ELC scores of 70 or below (N=207). Principal axis factor analysis revealed three factors that accounted for 52.39% of the common variance among CARS items. The first factor is labeled Social-Communication and includes Relating to People, Imitation, Listening Response, Verbal Communication, Nonverbal Communication, General Clinician Impressions and Level/Consistency of Intellectual Response. The second factor is labeled Emotion Regulation and includes Emotional Response, Adaptation to Change and Fear/Nervousness. The third factor is labeled Stereotyped Behaviors and Sensory Sensitivities and includes Body Use, Object Use, Taste/Smell/Touch Response and Use, and Activity Level.

Conclusions: Factor analysis of the CARS in a sample of approximately two-year-old children with ASD and a broad range of intellectual abilities revealed three factors: Social-Communication, Emotion Regulation, and Stereotyped Behaviors and Sensory Sensitivities. These factors are conceptually meaningful, and are consistent with our current understanding of ASD as a multidimensional construct. Factors found in the current study are similar to those found in previous investigations conducted with four- year-olds with ASD, with some important differences that may reflect differences in symptom presentation between children of different ages. Utilizing age-appropriate factor scores, in addition to the CARS total score, may extend the utility of the CARS as a measure of ASD symptom severity.