Screening for Autism Spectrum Disorder in Developmentally At-Risk Toddlers

Background: Children who are at-risk developmentally due to circumstances such as prematurity or genetic syndromes are diagnosed with autism spectrum disorders (ASD) at higher rates than those in the general population, making effective ASD screening techniques important for this group. The Modified Checklist for Autism in Toddlers (M-CHAT) is the most common ASD screening tool, yet little is known about its effectiveness in developmentally at-risk children.

Objectives: To evaluate the relation between the M-CHAT and Autism Diagnostic Observation Schedule (ADOS) in developmentally at-risk toddlers recruited from early intervention programs in southern Ontario by (a) examining whether M-CHAT classification was associated with ADOS classification or severity score in the whole sample, or in subgroups based upon extent of developmental delay; and (b) examining the profiles of the subset of participants who received a diagnostic outcome assessment.

Methods: Participants were 71 children recruited prior to their first birthdays for an ongoing longitudinal study. At 24 months, children were administered Module 1 of the ADOS and caregivers completed the M-CHAT (questionnaire and follow-up). The Mullen Scales of Early Learning (AGS Edition) and Vineland Adaptive Behavior Scales (Second Edition) also were administered to identify developmental delay, i.e., scores > 1 SD below the mean. ASD outcome was available for 21 participants (assessed at 3 years by experienced clinicians blind to prior study data using the ADI-R, ADOS, and clinical judgment using DSM-IV criteria).

Results: Twenty-six children had no delays, 19 had a delay in one domain, and 26 had delays in two or more domains. Twenty-four percent of the sample scored above the M-CHAT clinical cut-off. As expected, M-CHAT classification (Pass vs. Fail) was associated with ADOS classification (ASD/Autism vs. Non-ASD) based on the original algorithm (Χ² = 9.54, p = .002). However, when the sample was divided by extent of developmental delay (No Delay, Single Delay, or Multiple Delays), only the Chi-square test for the No Delay subgroup was significant (Χ² = 5.46, p = .019). Similarly, Mann-Whitney U tests using M-CHAT classification and ADOS severity score were significant for the whole sample (U = 307.50, p = .031) and the No Delay subgroup (U = 6.50, p = .031), but not the Single Delay or Multiple Delay subgroups. In the subsample with 3-year outcome data, 29% (6/21) received an ASD diagnosis and 33% (2/6) of those with an ASD diagnosis also had M-CHAT scores above the clinical cut-off at 24 months. Of the five children with 3-year outcome data who had M-CHAT scores above the clinical cut-off, 60% (3/5) did not receive an ASD diagnosis. All three of these children had Multiple Delays.

Conclusions: The M-CHAT has been shown to be a reliable screening tool for ASD in toddlers. The current findings suggest that it is useful with developmentally at-risk toddlers, but may not be specific enough in those with multiple developmental delays, which has implications for referral to diagnostic services. It will be important to assess ASD outcome for the remainder of the sample to determine the reliability of these findings.