Using the Anxiety Disorders Interview Schedule to Assess Social Phobia in Autistic Adolescents

Background: Social anxiety is a commonly reported problem among higher functioning adolescents with autism spectrum disorders (ASD) (Kuusikko et al., 2008; Bellini, 2004). There is no ‘best practice' approach for assessing social anxiety, or diagnosing comorbid Social Phobia (SoP), in people with ASD. Accurate clinical diagnosis of social anxiety is essential to effective treatment delivery. The Anxiety Disorders Interview Schedule (ADIS; Silverman & Albano, 1996) is a widely used, empirically supported clinician-rated diagnostic interview. Despite its application in studies of children and teens with ASD (Wood et al., 2008; White et al., 2009; Reaven et al., 2009), little is known about its psychometric properties with this population. This study evaluated the convergent and divergent validity of the ADIS as a tool for assessing SoP in adolescents with ASD. Objectives: The purpose of this study was to evaluate the utility of the ADIS SoP module for assessing social anxiety in ASD. Methods: Twenty-two adolescents (12-17 years; 4 females) with confirmed ASD diagnoses (via ADOS; Lord et al., 2002, and ADI-R; Lord et al., 1994) and comorbid anxiety disorders were administered the ADIS upon entry into a treatment study. Interviewers trained to reliability administered the ADIS jointly to parent and adolescent. SoP diagnoses were based on the clinician's severity rating (CSR) and clinical interview. Of the 22 participants, 18 had SoP, either as a primary or secondary diagnosis. Parents completed measures of anxiety (MASC-P; March, 1997, ASI-4; Gadow & Sprafkin, 1998) and social ability (SRS; Constantino & Gruber, 2005), teachers completed measures of social ability (SRS), and youth completed a self-report measure of anxiety (MASC-C; March, 1997). Results: Intraclass correlation was used to evaluate agreement among the three raters of the ADIS – clinician, parent, and child. Inter-rater ratings were acceptable (ICC=.802) indicating that parent, child, and clinician ratings of SoP on the ADIS were largely consistent. Kendall tau correlations were calculated in order to measure convergent and divergent validity of the ADIS with other measures. In support of divergent validity, SoP scores on the ADIS were not correlated with verbal IQ (tau=-.037), with measures of social functioning (tau=.214 to .323) or global anxiety (tau=.073). No evidence was found to support convergent validity of the ADIS: SoP scores on the ADIS did not correlate with social anxiety ratings (tau=.063 to .139). Conclusions: Although the results suggest that the ADIS measures something distinct from core ASD symptoms, these results yield inconclusive evidence for its convergent validity. Scores on the ADIS are not confounded with ability to report symptoms or lack of social skills. Additionally, scores from parent, child, and clinician ratings are consistent. However, clinician rated ADIS scores did not converge with other commonly used measures of social anxiety. Interpretation is difficult because we do not have a universal ‘gold standard' for measuring social anxiety in ASD. Future studies must continue to develop valid and psychometrically sound measures of social anxiety, as it is uniquely expressed in ASD, to aid clinical decision-making.