The Use of New ADOS Diagnostic Algorithms in Young Children with Williams Syndrome

Background: Recent research has explored the overlap between behaviors characteristic of autism spectrum disorders (ASD) and those seen in young children with Williams syndrome. Studies using the Autism Diagnostic Observation Schedule (ADOS; Lord et al, 2000) have found that while the percentage of children with Williams syndrome (WS) who meet criteria for a comorbid autism diagnosis is relatively low, a significant proportion exhibit behavior patterns similar to those seen in children with ASDs (Klein-Tasman et al, 2009; Lincoln et al, 2007). Newly developed ADOS algorithms have been able to more effectively identify PDD in populations with developmental delays (Gotham et al, 2007). Given recent findings of ASD symptomatology in some children with WS, the new algorithm’s increased ability to detect behaviors consistent with ASD in delayed populations is of interest in a WS sample.
Objectives: The purpose of this study is to examine classification of children with WS on the ADOS according to the newly published ADOS algorithms.

Methods: 29 children with WS (mean age=41.59 months, SD=8.95 months) were administered Module 1 of the ADOS and the Mullen Scales of Early Learning (MSEL). The ADOS administrations were scored according to the original and the newly published algorithms.
Results: Originally, 14/29 (48.3%) of the children with WS met or exceeded the cutoff score for an ASD (Klein-Tasman et al., 2008). With the use of the new algorithm, 16/29 (55.2%) children met or exceeded the cutoff score. In total, the classification of 10 children in the sample changed when the new algorithm was used:  4 of these children were originally classified as ASD and moved to nonspectrum, while 6 were originally nonspectrum and moved to ASD. Although a significant relationship was found between verbal ability and classification using the old algorithm cutoff scores, this relationship was not found when the new algorithm was used. 

Conclusions: The use of new algorithm cutoff points identified a slightly higher percentage of children with WS who exhibit behaviors consistent with ASDs. One third of the participants with Williams syndrome changed classification with the use of the new algorithm. The new algorithm appears to be more effective at identifying social difficulties independent of language. This is especially important in WS, as studies have found that children in this population often have language difficulties early in development (Mervis & Robinson, 2000). Additional implications will be discussed.