Evaluating a Parent-Rated Measure of Anxiety Symptoms in Children with Autism Spectrum Disorders

Background:  Anxiety disorders affect 30-40% of children and adolescents with autism spectrum disorders (ASDs).  Due to the cognitive and language deficits in this population, measuring anxiety symptoms is particularly challenging. A reliable and valid measure of anxiety in children with ASDs, that is sensitive to change with treatment, is a prerequisite for use in pharmacological and psychosocial intervention studies.

Objectives: The purpose of this study was to evaluate the psychometric properties of a 20-item, parent-rated anxiety scale derived from the Child and Adolescent Symptom Inventory (CASI Anxiety) in children with ASDs (Gadow et al., 2002; 1997; Sukhodolsky et al., 2008). Items of this DSM-IV-based scale are rated from 0 to 3 with higher scores reflecting greater severity. We examined the clinical characteristics of children with high anxiety symptoms and explored the symptom patterns in ASD subgroups (autistic disorder, verbal and non-verbal children; children with IQ above and below 70). 

Methods: The 404 participants (343 boys, 61 girls; age 4-17 years) took part in one of four federally-funded, multisite trials: Research Units on Pediatric Psychopharmacology (RUPP) Autism Network: risperidone vs placebo; methylphenidate vs placebo; risperidone only vs risperidone plus parent training) and STAART citalopram vs placebo (RUPP Autism Network, 2002; 2005; King et al., 2009; Aman et al., 2009). Of these 404 participants, 325 were diagnosed with Autistic Disorder, 59 with PDD-NOS and 20 with Asperger’s disorder. Across all diagnostic groups, 222 were classified with IQ < 70 and 92 were classified as non-verbal.

We examined the distribution, calculated the internal consistency and compared the item-mean to total correlations of the 20-item CASI Anxiety scale. Mean scores on the CASI Anxiety scale were compared in verbal vs non-verbal participants and children with an IQ above and below 70. For divergent validity, we examined the correlations with the Aberrant Behavior Checklist (ABC) subscales and the modified Children’s Yale Brown Obsessive Compulsive Scales (CYBOCS-PDD). Exploratory logistic regression also examined the clinical and cognitive characteristics of participants in the highest and the lowest quartiles on the CASI Anxiety scale.

Results: The 20-item CASI-Anxiety scores ranged from 0 to 50 (mean 14.2 + 9.39); with an alpha coefficient of 0.87. Five language-dependent items (including nightmares, expressed worries and physical symptoms) were rarely endorsed. This pattern was similar in verbal and non-verbal children. However, verbal children showed significantly higher mean scores on the CASI Anxiety scale (p<0.05). The high anxiety group (n=104 in the upper quartile; CASI Anxiety score > 19) was associated with the presence of language (OR: 3.77), IQ above 70 (OR: 3.24) and high scores (top 25%) on the ABC Irritability scale (OR: 3.76). Correlations of the CASI Anxiety scale and ABC subscales, CYBOCS-PDD and Vineland scales were modest (range 0.10 to 0.33) suggesting that the CASI Anxiety scale is measuring a separate construct from these other scales. 

Conclusions:  The DSM-IV based, CASI Anxiety scale is a starting place for measuring anxiety in children with ASDs. Additional research is needed to confirm sensitivity to change for use in clinical trials.