ADOS Diagnostic Utility in Children with Low Mental Age

Background: Many individuals with autism spectrum disorders (ASDs) have comorbid intellectual impairment, with some children presenting with low mental age (LMA), defined here as cognitive functioning at or below a 12-month level. However, few diagnostic tools are recommended for use in children below age 12 months, with the utility of using common tools in diagnosing ASD in children with concurrent LMA not yet demonstrated.  

Objectives: This study aims to assess the accuracy of a common diagnostic tool, the Autism Diagnostic Observation Schedule (ADOS), in diagnosing ASDs in toddlers with comorbid LMA.

Methods: Participants were drawn from a larger study on the early detection of ASDs, in which they received a developmental evaluation at the approximate age of two years. Cognitive level and autism symptomatology were assessed by the Mullen Scales of Early Learning (MSEL) and ADOS, respectively. Diagnoses were assigned according to DSM-IV-TR criteria, using clinical best estimate judgment of symptoms based on observation, history, and testing. ASD-LMA was given as a research diagnosis to children meeting criteria for Autistic Disorder (AD) or Pervasive Developmental Disorder, Not Otherwise Specified (PDD-NOS) who also demonstrated age-equivalent scores ≤ 12 months in MSEL Visual Reception, Expressive Language, and Receptive Language domains. The current study used Chi square tests to determine level of agreement between ADOS classification (i.e., Autism/AD, Autism Spectrum/PDD-NOS, Non-Spectrum) and clinical best estimate DSM-IV-TR diagnosis (i.e., AD, PDD-NOS) in children with ASD-LMA.

Results: 43 children (33 males; mean age 24.9 ± 4.45 months) were evaluated and diagnosed with ASD-LMA. 43 age-, gender-, and ethnicity-matched children meeting criteria for an ASD without concurrent LMA served as comparison. Independent-samples t-tests revealed significant group differences on all MSEL domains (all p’s < .001), with the ASD-LMA group showing greater impairment as expected. Groups did not significantly differ on ADOS severity (p = .162). Agreement between ADOS classification and DSM-IV-TR diagnosis was slight but non-significant in the ASD-LMA group (Χ²(1) = 2.030, p = .154, Cohen’s kappa = .154). In the ASD-LMA group, the ADOS correctly classified all children as having an ASD but over-estimated severity level in 25.6% of children. 93.0% of children were classified by the ADOS as having AD, resulting in a non-significant Chi square. By comparison, agreement in the ASD group was significant, likely related to a greater spread in ADOS scores (Χ²(2) = 14.353, p = .001, Cohen’s kappa = .372). However, the ADOS misclassified 16.3% of children as not having an ASD. Overall, both groups showed similar percent agreement (ASD-LMA group = 72.1%, ASD group = 67.4%).

Conclusions: This study was designed to determine the utility of a commonly-used assessment tool (i.e., ADOS) in diagnosing ASDs in children with comorbid LMA. Overall, results support the use of the ADOS in children with LMA. However, since it appears to over-estimate symptom severity, the ADOS may not be useful in distinguishing between AD and PDD-NOS in children with LMA, especially when used at age two years. Thus, it is important to use testing data in combination with clinical best estimate judgment.