Validation of a Novel Assessment Instrument for Identifying Autism Spectrum Disorder (ASD) in Children with Visual Impairment

Oral Presentation
Friday, May 3, 2019: 2:42 PM
Room: 517C (Palais des congres de Montreal)
N. J. Dale1, E. Sakkalou2, D. H. Skuse3 and A. Salt4, (1)Great Ormond Street Hospital NHS Foundation Trust, London, United Kingdom, (2)UCL Great Ormond Street Institute of Child Health, London, United Kingdom, (3)Behavioural and Brain Sciences Unit, Population Policy and Practice Programme, UCL Great Ormond Street Institute of Child Health, University College London, London, United Kingdom, (4)Great Ormond Street Hospital for Children, London, United Kingdom of Great Britain and Northern Ireland
Background: Children with profound or severe visual impairment (VI) are at high risk of developing difficulties in social communication and interaction (prevalence of ASD ~30% in several studies). The administration or coding of existing diagnostic tools depend on vision-dependent items and are therefore inappropriate for this clinical population. A novel observational schedule (the Visual Impairment Social Communication Observational Schedule - VISCOS), drawing on elements of ADOS with novel items and methods which do not require vision function, has been developed and validated for the first time.

Objectives: To investigate the performance of the VISCOS in reliably differentiating children with VI at high risk of ASD from those at low risk according to independent expert clinician formulation. Additional investigations aimed to establish concurrent validity by correlating the VISCOS scores with parent-report questionnaires that address social responses in children.

Methods: Four-to-seven-years-olds (M=5.4 years, SD=10.5 months; range 4.0 years to 7 years 9 months; 59% males), with profound VI (n=15; light perception at best), to severe VI (n=67, M=.98 logMAR) participated. The children engaged in social interaction and play including symbolic play, conversation, auditory story and non-visual theory of mind materials with a trained psychologist assessor, using standard presses designed specifically for children with very low or no vision (VISCOS). Parents filled out the Social Responsiveness Scale (SRS-2) and completed an interview consisting of the Development And Wellbeing Assessment (DAWBA) tool. Items which were heavily vision dependent were removed and scoring adjusted. Presented are preliminary data from 67 children, who were rated by the assessor using VISCOS coding schedule. A paediatrician, expert in VI and ASD, independently scored children using the video assessments and the parent-reported DAWBA to reach diagnostic formulation (Low Risk of ASD, Borderline, ASD) according to DSM-5 criteria.

Results: The internal consistency of VISCOS-Total items was good (α=.87). Inter-rater reliability was excellent (ICC=.95, p=.0001). Significant positive associations were found between VISCOS-Total and the parent-rated questionnaire SRS-2 suggesting good criterion validity. Clinician formulation showed: Low Risk ASD=65.7%; Borderline=19.4%, and ASD=13.4%. A ROC analysis revealed excellent predictive discriminant validity (AUC=0.92) with a sensitivity/specificity of 0.86 for clinician ratings and identified a VISCOS threshold score for High Risk for ASD (≥13.5) or Low Risk for ASD (<13.5). Correlations revealed the High Risk group had significantly higher reported difficulties than the Low Risk group on SRS-2 subscales (p=.005, η2=.122).

Conclusions: The VISCOS has been designed as the first visual impairment-specific assessment tool for children to support the identification and diagnosis of ASD and social communication difficulties. Results suggest a reliable instrument depicting strong internal coherence, construct validity, inter-rater reliability and predictive discriminant validity. Good criterion/construct validity was demonstrated using the results of the SRS-2 questionnaire. The results of the clinician formulation showed that about 32% of the children were classified as being at high risk of ASD (ASD plus borderline), a percentage similar to that found in other studies. A more accurate diagnostic algorithm is being developed. The profiles of ASD in the context of VI are discussed, providing new insights into non-visual ASD.