Exclusion Bias in ASD fMRI Studies: The Effect of Participant Anxiety on Scan Motion Artifact

Background:  Functional MRI (fMRI) is used to understand the neurobiological basis of neurodevelopmental disorders, such as Autism Spectrum Disorder (ASD). While data from an fMRI scan can be highly informative, it is susceptible to artifacts caused by participant motion. Previous literature suggests that 30% of data is lost from scans conducted with ASD participants due to scan motion artifacts (Yerys et al., 2009). Researchers can effectuate methods to reduce these motion artifacts with use of mock scanner training protocols (de Bie et al., 2010; Raschle et al., 2012; Nordahl et al., 2016). Nevertheless, there are some variables that may increase motion but are difficult to control, such as participant trait anxiety. Given the high co-morbidity of anxiety disorders in individuals with ASD, it is important to consider whether anxiety affects fMRI data quality and leads to exclusion biases. No previous study has examined the relation between anxiety-like symptoms and scan motion in a childhood ASD sample.

Objectives:  We investigated the relation between child-reported and parent-reported anxiety and average scan motion (i.e., frame displacement) during a social interaction fMRI paradigm in children with ASD.

Methods:  To date, participants include 16 children (15 males), 11.59 ± 2.26 years, FSIQ 110.94 ± 21.45, diagnosed with ASD. The Screen for Child Anxiety Related Emotional Disorders Scale (SCARED; Birmaher et al., 1999) was administered to participants and to one of their parents. The SCARED is a validated questionnaire used as a continuous measure of childhood anxiety symptoms and includes the following factors: somatic/panic, general anxiety, separation anxiety, social phobia, and school phobia. Before entering the fMRI scanner, children received practice in a mock scanner to decrease scan-related anxiety. During the scan, participants completed an innovative social interaction paradigm, in which they believed they were chatting with a peer in real time (functional results from this experiment are discussed in another submission). Average frame displacement (FD) was calculated for each run. We conducted partial correlations between average FD and SCARED factor scores, including age and FSIQ as covariates.

Results: Results demonstrated a significant positive correlation between average FD and the parent-reported generalized anxiety factor, r(12) = .540, p = .046. In addition, there was a significant positive correlation between average FD and the parent-reported somatic/panic factor, r(12) = .743, p = .002 (see Figure 1). There were no significant correlations between child-reported anxiety factors and average FD.

Conclusions: Preliminary results suggest that parent-reported anxiety on the generalized and somatic/panic factors relate to fMRI scan motion in children with ASD. Scan motion can cause artifacts and spatial misalignment and thus the data is often excluded from research analyses. Researchers should be aware of these moderate to strong correlations between participant anxiety and scan motion, as excluding these participants from their dataset may result in a sample that is not fully representative of the ASD population. These findings are particularly relevant for paradigms examining social interactions, which may exacerbate social anxiety in participants with ASD.