Diagnostic Stability of Autism Spectrum Disorders and Predictors of Crossover in Toddlers Prospectively Identified in a Community-Based Setting

Background:  Diagnoses of Autism Spectrum Disorders (ASD) at 2-years-of-age has been established as relatively stable across time in studies using high-risk sibling or clinic referred samples. The stability of ASD diagnoses in children who have been prospectively identified from low-risk, community-based samples is not well established. Furthermore, the majority of studies have compared the diagnostic stability of diagnostic tools (e.g., ADOS/ADI-R) and clinical judgement, rather than identifying individual behaviours that predict crossover from ASD to non-ASD diagnoses, and comparing the cognitive profiles of ASD-stable versus ASD-crossover groups.

Objectives:  The primary objective in this longitudinal study was to investigate the diagnostic stability of ASD diagnoses from 2-years to 4/5-years in children prospectively identified through routine developmental surveillance in a community-based setting. A secondary aim was to identify the individual behaviours that were most predictive of crossover from ASD to non-ASD diagnoses, and compare the cognitive profiles of the ASD-stable versus ASD-crossover groups. 

Methods:  A total of 99 children received a best estimate diagnosis of Autistic Disorder (AD), Autism Spectrum Disorder (ASD), or developmental and/or language delay (DD/LD) at 24-months (Time 1) as part of the Social Attention and Communication Study (SACS). 77 children returned for a follow-up diagnostic assessment between 4- to 5-years (Time 2). Diagnoses at Time 1 were based on expert clinical judgement utilising the ADOS, ADI-R, and Mullen Scales. Mean scores at Time 1 on the individual items of the ADOS Module 1 and the separate subscales of the Mullen (minus gross motor) were also compared between ASD-stable and ASD-crossover groups.

Results:  Diagnostic stability was high, with 86% of children retaining an ASD diagnosis from Time 1 to 2. However, stability of diagnoses within the spectrum (i.e., AD vs ASD) was more variable, although no children in the DD/LD group moved onto the spectrum. ANOVAs revealed significant differences in ADOS Time 1 scores for ‘Unusual Eye Contact’, ‘Frequency of Vocalizations Directed to Others’, and ‘Integration of Gaze and Other Behaviors’ between ASD-stable (n = 53) and ASD-crossover (n = 9) groups, with the ASD-stable group displaying higher mean scores (greater abnormality) in these behaviours. Furthermore, investigation of the cognitive profiles of the two groups at Time 1 revealed that the ASD-crossover group had significantly higher visual reception, fine motor, and receptive language age equivalents, with the greatest mean difference in receptive language.

Conclusions:  The current study suggests that, consistent with previous studies utilising high-risk samples, diagnoses of ASDs at 2-years-of-age is stable in children identified through routine developmental surveillance in a low-risk sample. Furthermore, the best individual predictors on the ADOS of diagnostic crossover from Time 1 to 2 were social attention and communication behaviours, and included not only the use of eye contact and direction of vocalisations towards others, but the integration of these behaviours. Higher receptive language in the ASD-crossover versus ASD-stable group is consistent with our previous findings that higher receptive language skills in the latter part of the second year of life may place children on a developmental trajectory away from the autism spectrum.