Differentiating Profiles of Anxiety and Executive Function in ASD without ID

Saturday, May 14, 2016: 10:55 AM
Room 310 (Baltimore Convention Center)
C. E. Pugliese1, L. G. Anthony1, X. You2, A. C. Armour3, Y. Granader1 and L. Kenworthy4, (1)Children's National Medical Center, Rockville, MD, (2)Children's National Health System, Washington, DC, (3)Children's National Medical Center, Arlington, VA, (4)Children's Research Institute, Children's National Medical Center, Rockville, MD
Background:   Youth with ASD experience high rates of co-morbid anxiety, which has been linked to rumination and inflexibility in typically developing (TD) youth and those with developmental disorders.  Inflexibility is a near ubiquitous problem in ASD, which is also associated with a range of other EF problems. Given the fact that not all youth with ASD develop anxiety, it is important to explore whether anxiety in ASD is driven by inflexibility and/or other specific EF deficits. A further question is whether the heterogeneity of ASD anxiety symptomatology can be parsed with specific profiles of EF. 

Objectives:   To examine whether (1) specific domains of parent-reported EF, particularly flexibility, was predictive of greater parent-reported anxiety symptoms, and (2) specific profiles of EF and anxiety cluster together in subgroups of youth with ASD.  

Methods:  220 youth (44 females) with a DSM diagnosis of ASD between the ages of 8 and 13 (M=10.46, SD=1.69) were evaluated on the Behavior Rating Scale of Executive Function (BRIEF) and Child Behavior Checklist (CBCL). Participants possessed average IQ (M=106.13, SD=19.17) and met CPEA criteria for ‘broad ASD’ on the ADI-R and/or ADOS. To determine whether specific EF domains were predictive of anxiety, we regressed CBCL Anxiety Problems DSM-oriented scale scores onto BRIEF subscales (predictors), age, and IQ (covariates). We also used a data driven, model free community-detection approach to parse heterogeneity in EF and anxiety profiles in ASD. This approach is an optimization clustering method, which identifies subgroups that share similar behavioral phenotypic features. It reveals profiles-based dimensions to detect how EF and anxiety cluster together in terms of relative strengths and weaknesses.

Results:  When all predictors and covariates were entered into the regression model, the only significant predictor was the BRIEF shift scale (t=3.56, p<.001) with the overall model explaining 22.6% of the variance in Anxiety scores (F=8.43, p<.001). The community detection analyses revealed three subgroups: 1) a high anxiety group (n=88) clustered with high shifting and emotional control issues, 2) a low anxiety group (n=60) clustered with high disinhibition and working memory problems but fewer issues with flexibility and emotional control, and 3) a medium anxiety group (n=77) clustered with metacognitive impairments and fewer inhibition, shifting, and emotional control problems. See attached image for group profiles. Approximately 64%, 17%, and 53% of participants in each group, respectively, received Anxiety scores above the borderline cut-off on the CBCL. 

Conclusions:  Both regression and cluster analyses revealed positive associations between anxiety and inflexibility. In addition, the cluster analysis revealed that two other EF profiles were associated with anxiety: group 1 was characterized by relatively high anxiety, inflexibility, and emotion dysregulation; group 2 was associated by relatively high inhibition problems and relatively low anxiety. As such, clinical treatment may differ for each of these profiles; the highest anxiety group may need behavioral intervention for cognitive flexibility and emotion regulation along with effective, direct treatments for anxiety.