The Reliability of Computerized Neurocognitive Assessments in Autism Spectrum Disorder

Background: The administration and implementation of computerized neurocognitive assessments is emerging, however, the reliability of such assessments remains in question (Vrana & Vrana, 2017). The feasibility of computerized assessments has been demonstrated in typically developing populations, although its use is not considered a substitute for clinician-administered assessments (Butcher, Perry, & Atlis, 2000; Gualtieri & Johnson, 2006). The reliability of computerized neurocognitive assessments in comparison to clinician-administered assessments has yet to be studied in individuals with autism spectrum disorder (ASD).

Objectives: This study explores the performance of children with ASD and their unaffected siblings on both computerized and clinician-administered measures of intellectual functioning.

Methods: Thirty-six individuals participated in this study, including 17 individuals with ASD (ages 6-24 years) and 19 unaffected siblings (ages 6-27 years). ASD diagnoses were confirmed via gold-standard tools and expert clinical judgement. Computerized administrations of the Peabody Picture Vocabulary Test-4 (PPVT-4) and a Visual Reasoning (VR) (testmybrain.org) task were completed on a laptop in the clinic. All participants were administered the Wechsler Abbreviated Scale of Intelligence-II (WASI-II). Probands also completed a comprehensive developmentally-appropriate test of cognitive functioning (Wechsler Adult Intelligence Scale-IV, Wechsler Intelligence Scales for Children-5, Stanford Binet-5, Mullen Scales of Early Learning). Developmental quotients were calculated for participants who completed the Mullen.

Results: T-tests were run between clinician-administered and computerized measures. Results revealed that probands’ performance did not differ significantly between assessment modality (i.e., clinician-administered versus computerized) across domains, except for one comparison (WASI-II Perceptual Reasoning Index standard score (SS) vs. VR SS), with significantly stronger performance on the clinician-administered WASI-II (p=.04). In contrast, unaffected siblings performed significantly better on all clinician-administered assessments compared to computerized assessments (p<.01).

Conclusions: This study demonstrated that children with ASD performed similarly on computerized- and clinician-administered assessments, whereas their unaffected siblings performed significantly better on clinician-administered assessments. Results suggest that individuals without ASD may respond better to human social interaction and prompting. Future directions include further examination of the reliability of computerized cognitive assessments in larger and more homogenous subgroups of individuals with ASD.