Using the M-CHAT-R to Identify Developmental Concerns in a High-Risk 18-Month-Old Sibling Sample

Background: With estimates of the recurrence risk of ASD in a younger sibling as high as 18.7% (Ozonoff et al., 2011), identifying early markers of ASD is extremely important for purposes of early diagnosis and intervention. However, screening siblings can also be a complex process that may warrant a more extensive system of identification and testing than current standards dictate. Little is currently known about the sensitivity and specificity of commonly used ASD screening instruments in the general population, much less in the high-risk sibling subset.

Objectives: This project examines the predictive properties of a potential community-based measure, the Modified Checklist for Autism in Toddlers-Revised (M-CHAT-R; Robins and Fein, 2011), as a screening instrument for infant siblings of children with ASD.

Methods: 71 younger siblings of children with ASD completed M-CHAT-Rs at a mean age of 18 months. Of these children, 25 failed the M-CHAT-R + follow-up interview (FUI). An additional 15 children who passed the M-CHAT-R agreed to participate in full diagnostic evaluations (cognitive, adaptive, and autism testing) for a total of 40 participants. No information is available on participants who failed the M-CHAT-R but declined further evaluation. Three children who failed the screener but passed the FUI did not come in for evaluations.

Results: We completed full diagnostic evaluations on 25 children who failed the M-CHAT-R+FUI and 15 children who passed. Diagnoses were as follows: No Diagnosis (ND; n=15), Other Developmental concern (OD; n=3), Autism Spectrum Disorder (ASD; n = 12), and At Risk (AR=10). This At Risk designation included children where the clinician had ASD-related concerns but deferred diagnosis for future evaluation. In ASD and OD, 100% failed the M-CHAT-R+FUI; in AR, 50%; and in ND, 20% of children failed. We examined the M-CHAT-R+FUI sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) two ways. When differentiating between autism risk (ASD+AR), non-autism (OD) and no concerns (ND),  the M-CHAT-R+FUI had sensitivity of 77.27%, specificity of 66.67%, PPV of 73.91%, and NPV of 70.59%. When differentiating between any developmental concern (ASD, OD, AR) and no concerns (ND), sensitivity and specificity were 80%, PPV was 86.96%, and NPV was 70.59%.

Conclusions: The M-CHAT-R+FUI demonstrated better sensitivity, specificity, and PPV when identifying children with any developmental concern. NPV remained the same across both groups. Differences in sensitivity across groups were modest, whereas differences in the specificity and PPV were larger. Our results mirror Chlebowski et al.’s (2013) finding using the original M-CHAT+FUI in a population-based sample, which indicated that it performed better at picking up general developmental concern (PPV = 98%) than ASD risk (PPV = 54%). Results from our high-risk sample suggest that the M-CHAT-R+FUI continues to pick up broader developmental concerns in addition to ASD-related vulnerabilities, underscoring the need for clinician training to facilitate appropriate post-screening follow-up evaluations for children.