Parent-Child Informant Discrepancies of Social Anxiety in ASD Relate to ASD Symptoms and Adaptive Functioning

Background: Social anxiety is highly comorbid with autism spectrum disorder (ASD), and impairs individuals’ adaptive responses in daily life. Parents and youth often endorse discrepant levels of youth anxiety, which may be exacerbated by the socio-communicative impairments inherent to ASD. However, despite best-practices calling for gathering information from multiple informants, little research has comprehensively evaluated correlates of parent-child informant discrepancies (PCIDs) for social anxiety in youth with and without autism spectrum disorder (ASD).

Objectives: This study examined PCIDs for social anxiety symptoms in youth with ASD and an age-matched typically-developing comparison sample (COM), and factors that moderate the extent of disagreement.

Methods: Participants included 223 verbally-fluent (i.e., verbal IQ>70) children and adolescents (8 to 16 years old, 110 ASD, 113 COM) and their parents, who completed the Social Anxiety Scale for Children, Revised (SASC-R; La Greca & Stone, 1993). Parents also completed the Social Communication Questionnaire (SCQ; Berument et al. 1999) assessing the child’s autism symptoms and the Behavior Assessment System for Children (BASC-2; Reynolds, 2010) to measure adaptive functioning.

Results: A two (informant: parent, youth) by two (group: ASD, COM) repeated-measures ANCOVA on SASC scores revealed that participants with ASD (M_{ad j}=49.90, SE=1.00) were rated as having higher social anxiety than COM participants [M_adj=35.57, SE=0.99; F(1, 219)=98.06, p<.001]. Informant discrepancies were lower, but more variable in the ASD group, relative to COM participants, F(1, 219)=7.67, p=.006, Levene’s Test of Error Variances: F(1, 221)=12.06, p=.001 (Figure 1). We then assessed whether the PCIDs of social anxiety were related to demographic and diagnostic characteristics using polynomial regression analyses (Laird & Weems, 2013). In this model, patterns of agreement or discrepancy are characterized using an interaction between parent and youth informants’ reports, controlling for main effects (linear and quadratic) of each informant. Informant agreement was not related to demographic factors of age, or verbal IQ for either group (Table 1). In the ASD group, youth autism symptoms (SCQ) and adaptive functioning (BASC) were related to PCIDs, such that profiles that agreed on youth anxiety symptoms had lower lifetime autism symptoms and better adaptive skills. A different pattern emerged in the COM group, where youth that agreed with their parents on high anxiety symptoms had the lowest adaptive skills.

Conclusions: This was the first study to assess PCIDs of social anxiety using polynomial regression analyses, which allow examination of different profiles of agreement and discrepancy. Given the group differences in the degree and variability of PCIDs, it appears that youth with ASD do not reliably differ from their parents on reports of social anxiety, highlighting the importance of characterizing correlates of PCIDs in this group. PCIDs of social anxiety were not related to verbal IQ or age for either group, indicating that demographic factors likely do not influence the accuracy of informant reports of anxiety. Associations between PCIDs and outcomes indicate that for individuals with ASD, parent-child agreement on level of youth anxiety, be it high or low, was related to better outcomes. Implications for clinical and research practices are discussed.