Predictive Validity of the MCHAT-R in a Clinical Sample

Background:

Extant literature upholds the importance of early, intensive intervention in maximizing positive outcomes for children with autism spectrum disorder (ASD; Sullivan, Stone, and Dawson, 2014; Warren et al., 2011). Research suggests that the most efficacious interventions for children with ASD start as early as risk is identified and persist during early childhood (Wallace & Rogers, 2010). Individuals with risk markers for ASD must be identified as soon as possible to initiate services. The Modified Checklist for Autism in Toddlers, Revised with Follow-Up (MCHAT-R/F) has demonstrated reliability and validity, as well as improved utility over the original MCHAT, in detecting ASD in low-risk toddlers (Robins, et al., 2014). MCHAT-R/F was designed to facilitate reduced age of diagnosis and commencement of early intervention.

Objectives:

The current research examines diagnostic agreement between the MCHAT-R and clinical diagnosis in clinical sample to further investigate validity of this instrument.

Methods:

Data was collected from young children (n=201; 33 = female, 168 male; and mean age= 28.69 months) who visited a hospital-based early diagnostic screening clinic regarding autism spectrum disorder-related concerns (9/1/2015- 9/1/2016). These patients participated in a full battery of testing, including diagnostic interview, MCHAT-R/F, ADOS-2, Mullen, and Vineland-II. The battery did not incorporate the MCHAT-R/F follow-up questions as clinicians conducted thorough clinical interviews with caregiver(s). All clinicians (n=6) were licensed clinical psychologists who have demonstrated research reliability regarding the ADOS-2.

Diagnostic outcomes included 133 diagnoses of autism spectrum disorder (ASD) and 68 diagnoses of No ASD. In order to examine diagnostic agreement, we calculated the M-CHAT-R sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) compared to clinical diagnostic decision. A 2x2 table was constructed based on the presence and absence of ASD as well as low risk (≤2) and medium-to-high risk scores (≥3) on the MCHAT-R.

Results:

With the current sample, MCHAT-R sensitivity was 86.46%, specificity was 32.35%, PPV was 70.7%, and NPV was 50%. Therefore, the probability of the MCHAT-R correctly identifying individuals with ASD was 86.56% and correctly identifying individuals with No ASD is 32.35%. With this sample, the probability of ASD with a positive MCHAT-R screen is 70.70% and the probability of no ASDwith a negative MCHAT-R screen is 50%.

Conclusions:

Based on the current clinical sample of young children with red flags for ASD, the MCHAT-R demonstrates greater sensitivity than specificity and greater PPV rather than NPV. This suggests that the MCHAT-R often captures individuals with developmental concerns, including ASD; however, it indicates that a low risk score on the MCHAT-R may not be sufficient to rule-out an ASD diagnosis. Results from our sample suggest that the M-CHAT-R continues to be helpful in identifying broader developmental concerns in addition to ASD-related vulnerabilities. These results also highlight the importance of post-screening evaluation for differential diagnosis, particularly with complex presentations of developmental concerns. The present data supports further investigation regarding what factors might influence MCHAT-R scores/reporting to inform clinical practice.