Early Screening for Autism Spectrum Disorders (ASD) in Two-Year-Old Children with Visual Impairment and Longitudinal Outcomes

Background: Children with congenital profound or severe visual impairment (VI) are at high risk of ASD and social-communication difficulties (prevalence of ASD ~30%). Existing tools used to measure socio-communicative development and ASD are highly vision-dependent and not valid for children with VI. We recently validated a novel visual impairment social-communication observational schedule (VISCOS), drawing on principles of ADOS but specifically designed for 4-7 year olds with VI. We developed initial validation against expert clinician formulation and have identified children who are at high risk of ASD. A proportion of children who participated in the VISCOS at 4-7 years, also participated at 2-years-old in standard behavioural items and social ‘presses’ to elicit social and communicative behaviours and coded using the Social Communication Schedule (SCS-2), which was non vision-dependent. The SCS-2 includes a Social-Communicative ability Scale (SCS) and a Negative Behavioural Screener (NBS) of repetitive and restricted behaviours.

Objectives: To carry out longitudinal analyses between the SCS-2 and clinician and VISCOS outcomes, to establish construct validity between tools. To examine whether the SCS-2 could act as a potential early screener to identify children at-risk of ASD.

Methods: Preliminary data from 39 children at 24 months (M=25.51, SD=2.36) with VI from the longitudinal OPTIMUM project were rated using the SCS (high scores indicated better social communicative abilities) and a negative behaviour screener (NBS – higher scores indicated more negative behaviours), whilst engaging in social and independent play tasks. The same children participated in the VISCOS assessments at 4-7 years (M=64.04, SD=8.12) - DAiSY study.

Results: A Kruskal-Wallis test examined differences in the 3 clinician categories (Non-ASD, Borderline, ASD) and SCS and NBS. A significant difference was found between groups on SCS scores, with a difference present between the Non-ASD group and the ASD group χ²(2)=8.2, p=0.17; children in the ASD group (N = 6; M=16.83 SD=9.13) scored lower SCS scores than children in the Non-ASD group (N= 28; M=26.14 SD=5.75). A Mann-Whitney U-test comparing the combined ASD+Borderline group to the Non-ASD group found the ASD+Borderline group scoring lower (N=11; M=18.63 SD=7.47) on SCS than the Non-ASD group (N=28; M=26.14 SD=5.75), U=63.00 p=.004. No significant findings were found for the NBS. A ROC analysis on VISCOS scores based on the clinician formulation groups had 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). Using this cut-off (at or above 13.5 indicating difficulties) the two groups did not differ significantly on SCS scores (ASD+Borderline N=13; M=23.31, SD = 7.12 vs Non-ASD N=20; M=26.40, SD = 6.01) U= 94.00 p = .184. Despite the nonsignificant p value, a Cohen’s d effect size of 0.475 suggested a medium effect.

Conclusions: The SCS-2 scores at 2 years showed significant differentiation of the clinician diagnostic categories (drawing on the VISCOS tool) at 4-7 years. Findings suggest that this ‘early stage’ tool may provide a potential screener for early signs of ASD in children with VI.