Stability of AUTISM Spectrum Diagnoses within A High-RISK Longitudinal Cohort

STABILITY OF AUTISM SPECTRUM DIAGNOSES WITHIN A HIGH-RISK LONGITUDINAL COHORT

Wendy Roberts, Susan Bryson, Isabel Smith, Jessica Brian, Caroline Roncadin, Tracy Vaillancourt, Peter Szatmari, & Lonnie Zwaigenbaum

Background: Longitudinal studies of clinically-referred children indicate a high level of stability of autism spectrum disorder (ASD) diagnoses, with over 80% of children diagnosed with ASD at age 2 years having these diagnoses confirmed independently at age 3. However, the sensitivity and specificity of early ASD diagnoses may be influenced by ascertainment biases, whereby children with more severe symptoms and/or language impairment are referred earlier.

Objectives: Our goal is to examine the stability of diagnostic classification in a high-risk cohort of younger siblings of children with ASD, followed prospectively from infancy.

Methods: ASD symptoms were assessed prospectively in 246 high-risk infants using the Autism Observation Scale for Infants (AOSI) at 6, 12 and 18 months, and using the Autism Diagnostic Observation Schedule (ADOS) at 18, 24 and 36-42 months. The Autism Diagnostic Interview – Revised (ADI) was administered routinely at 36-42 months (hereafter, 3 years), and earlier for selected participants with suspected ASD. Clinical diagnoses were assigned based on expert judgment using the ADI-R (if available), ADOS and DSM-IV-TR. Other aspects of development were assessed using the Mullen Scales of Early Learning (MSEL) at each time point.

Results: Of the 42 clinical ASD diagnoses that were established by age 24 months, 36 were confirmed independently (i.e., blind to prior assessments) at 3 years (85.7%). Six children were noted to have other clinical concerns at 3 years (including 2 with language delays) but were subthreshold for ASD. An additional 28 children who had not been identified at 24 months, were diagnosed with ASD at 3 years. Thus, the specificity of a 24-month ASD diagnosis relative to 3-year diagnosis was very high (96.7%), with a positive predictive value of 85.7%; however, sensitivity was relatively low (56.2%). Children not diagnosed until 3 years were less likely than those diagnosed by 24 months to meet full criteria for autistic disorder, more likely to have phrase speech, and had more advanced cognitive skills as indexed by MSEL at 3, and had fewer symptoms of ASD at age 12 and 18 months.

Conclusions: In our prospective high-risk cohort, 85.7% of ASD diagnoses established at 24 months were confirmed independently at age 3. However, the sensitivity of 24-month diagnostic assessment, relative to independent ASD diagnoses at age 3 years, was only 56.2%. The later-diagnosed group, most of whom had diagnostic subtypes other than autistic disorder, was more likely to have more advanced cognitive and language development than children diagnosed by 24 months. The later diagnosed group also had fewer symptoms at age 12 and 18 months, consistent with previous reports that early detection, screening and diagnosis may be a particular challenge for higher-functioning children with ASD. Ongoing follow-up for this high-risk cohort may identify children with even milder manifestations of ASD.