Fenson

The ADOS is a clinician-administered, standardized observation designed to elicit social communication and restricted, repetitive behaviors related to ASD (Lord et al., 2000 (link)). Four original modules are each tailored to an individual’s language level and age to control for the effects of language level on social communication and play behaviors. The second edition of the ADOS (ADOS-2; Lord et al., 2012a ; Lord et al., 2012d ) adds a Toddler Module for children age 12 to 30 months with language skills ranging from no verbal language to single words and simple phrases. Toddlers must be walking independently, and a nonverbal mental age of at least 12 months is recommended. The Toddler Module follows the structure of Module 1, which is designed for language levels ranging from nonverbal to single words and simple phrases. Module 1 activities, child behavioral descriptions, and scoring criteria were modified based on developmental expectations for toddlers.
The Toddler Module algorithm contains separate domain categories of Social Affect and Restricted, Repetitive Behaviors and a single total score to determine classification. Separate algorithms are provided based on age and language level: all children age 12 to 20 months, and children age 21–30 months who produce fewer than five words during the ADOS-2, receive the 12–20/Nonverbal 21–30 algorithm, and children age 21–30 months who produce five or more words during the ADOS-2 receive the Some Words 21–30 months algorithm. Clinical cut-off scores are grouped within levels of concern for ASD, acknowledging the diagnostic uncertainty inherent in very young children due to significant developmental variability or confounding conditions (e.g., intellectual disability, language impairment). Research classifications with cut-points for ASD and nonspectrum also are available (Luyster et al., 2009 (link)).
We examined the sensitivity of Toddler Module research classifications and concern ranges for our samples, and results were similar to those reported in the original validation study (Luyster et al., 2009 (link)). Using the research cutoffs of a total score of 12 for 12–20/Nonverbal and 10 for Some Words 21–30, sensitivity in the original sample was .94 for children who received the 12–20/Nonverbal 21–30 algorithm and .88 for children receiving the Some Words 21–30 algorithm. Sensitivity in our replication sample was 0.88 for the 12–20/Nonverbal 21–30 group and 0.71 for the Some Words 21–30 group. In the original sample, 82.2% fell within the moderate-to-severe concern range, 14.4% fell into the mild-to-moderate range, and 3.4% fell into the little-to-no concern range. In the replication sample here, 72.2% fell within the moderate-to-severe range, 19.1% were in the mild-to-moderate range, and 8.7% were in the little-to-no concern range.
In the current study, the ADOS-2 Toddler Module was conducted as part of a clinic or research evaluation. A similar battery of assessment measures was used across sites and projects. The University of Michigan, University of Minnesota, and University of Wisconsin-Madison administered the Toddler Autism Diagnostic Interview-Revised (Toddler ADI-R; Kim & Lord, 2012 (link); Lord, Rutter, & LeCouteur, 1994 (link)) to inform diagnosis; children seen at FSU were given a developmental history interview and parent-report measures of ASD symptoms. Children at all sites received psychometric measures of cognitive and adaptive development, including Mullen Scales of Early Learning (MSEL; Mullen, 1995), and Vineland Adaptive Behavior Scales, 2^nd edition (Vineland-II; Sparrow, Cicchetti, & Balla, 2005 ). Additionally, language skills were assessed at the Universities of Michigan, Minnesota, and Wisconsin-Madison using the Preschool Language Scales (PLS, 4^th and 5^th editions; Zimmerman, Steiner, & Pond, 2002 ; 2011 ) and/or MacArthur-Bates Communication Development Inventories, 2^nd edition (Fenson et al., 1993 ). Diagnostic distinctions of autism and non-autism ASD were made at the Universities of Michigan and Wisconsin-Madison; at FSU and University of Minnesota, subcategories were not assigned, and children meeting criteria for DSM-IV diagnoses of Autistic Disorder, PDD-NOS, or Asperger’s Disorder were given a best estimate diagnosis of ASD. To be consistent with DSM-5 (APA, 2013 ), and because clinical subcategories have been found to be unstable over time (e.g., Lord et al., 2006 ), unreliable across clinicians, and not representative of meaningful differences in symptom presentation (Lord et al., 2012c (link)), children with any autism spectrum diagnosis were grouped into one ASD category for the present analyses.
Clinic-referred patients received oral feedback and a written report without financial compensation. Participants recruited only for the purpose of research received financial compensation and a written summary of evaluation results. Institutional Review Boards at the University of Michigan, FSU, University of Minnesota, and University of Wisconsin-Madison approved all procedures related to this project.
Site differences emerged in demographic and child variables. Differences in child variables across sites were expected due to differences in recruitment patterns and study design across sites. We viewed these site differences as beneficial to the purpose of this study, as we sought to include children with varied levels of impairment and symptom characteristics. The University of Wisconsin sample generally was older, had lower verbal skills, and showed greater impairment in IQ and ADOS-2 scores than children from other sites. Families in the FSU sample self-identified as more racially and ethnically diverse than families from other sites. See Supplemental Tables 1 and 2 for further details on site differences.