Autism Diagnostic Observation Schedule- Generic (ADOS-G) with Typical Children: Evaluating Diagnostic Validity

Background: The Autism Diagnostic Observation Schedule- Generic (ADOS-G) (Lord et al., 2000) is considered the current gold standard diagnostic tool for autism. There is sufficient evidence to suggest that the ADOS-G has good sensitivity and specificity and found to be better at discriminating between individuals with ASD and non-ASD than between autism and PDD NOS. However, an inherent problem with the majority of the studies examining the diagnostic validity of the ADOS-G is that few include a typical control group to demonstrate that the ADOS-G does not misdiagnose typically developing (TD) children as having ASD (Bishop & Norbury, 2002). Many of the behaviours measured within the ADOS-G have been found to discriminate between young children with ASD and TD. However, due to large variation in TD between 18 and 24 months, some behaviors measured are present in TD toddlers and may be less discriminatory.

Objectives: The objective of the current study was to examine the performance patterns of TD children on the ADOS-G and to explore the relationship of these patterns and cognitive and adaptive functioning.

Methods: Participants included 25 TD toddlers between the ages of 18 and 26 months (M= 22.76). Participants completed the Mullen Scale of Early Learning (MSEL), the Vineland Adaptive Behaviour Scale (VABS) and the ADOS- G. The ADOS-G item means and domain means of the original and revised algorithms, domain total distributions and item distributions, were analyzed to determine the range of normal scores. Correlations between ADOS-G scores and summary scores of the MSEL and VABS were calculated.

Results: A large proportion of TD toddlers demonstrated abnormalities on some of the items measured on the ADOS-G. Thirty six percent of the participants obtained a score above ‘ASD cut off’ on the Communication domain and 16% obtained a score above ‘ASD cut off’ on the Reciprocal Social Interaction domain. Twelve percent obtained a score above ‘ASD cut off’ on the Social Affect domain and 8% obtained a score above ‘ASD cut off’ on the Social Affect and Restrictive and Repetitive Behavior total domain. None of the toddlers obtained a score in the ASD or autism range on the original algorithm and two toddlers obtained a score consistent with ASD on the revised algorithms. Correlational analyses revealed significant relationships between Receptive Language (MSEL) (-.35, p<.05), Expressive Language (MSEL) (-.39, p<.05) and ADOS-G scores. Significant relationships were also found between Communication skills (VABS) (-.73, p<.01; -.38, p<0.05), Daily Living skills (VABS) (-.45, p<.05; -.40, p<0.05) and ADOS-G scores (original and revised).

Conclusions: The results of the study provide researchers and clinicians with a better understanding of how typical children perform on the ADOS-G and insight into which areas the performance of typical toddlers overlaps with individuals diagnosed with ASD.