Using the M-CHAT Best 7 Score In Screening for Autism Spectrum Disorders In Young Children Referred for Developmental Assessment

Background:   The M-CHAT was developed for use in primary care settings to identify toddlers at risk for ASD and has undergone several revisions to administration and scoring procedures (e.g., the use of a follow-up interview) to increase clinical utility as well as sensitivity and specificity as a screening tool. However, the additional time required by the follow-up interview is potentially limiting for everyday practitioners.  Dr. Diana Robins (IMFAR 2010) presented promising data regarding the use of a “Best 7” approach for scoring the M-CHAT in a primary care sample to reduce the need for a follow-up interview. The current study evaluates this refined approach in a sample of high-risk children who are referred for evaluation due to known or suspected developmental delays.

Objectives:   To evaluate the Best 7 scoring and identification approach for the M-CHAT in young children in a tertiary care clinic setting among children at risk for ASD or DD.

Methods:   As part of the intake process for children under the age of 4 years, caregivers completed the M-CHAT in addition to a general intake form requesting information about development. Based on referral question and intake materials, including scores on the M-CHAT, children received either ASD-specific or more general developmental evaluations. Rule out or diagnosis of an ASD utilized ADI-R/ADOS and final diagnostic conclusions were determined by members of the interdisciplinary team.

Results:   Diagnostic evaluations are complete for 95 children whose caregivers completed the M-CHAT at intake (mean age = 33 months). Among the children with confirmed ASD, 86% were accurately identified by the original M-CHAT scoring procedure, while 67% were identified with the Best 7 approach. Specifically, the Best 7 approach did not identify 10 children with ASD who were identified by the original approach. Among the children whose ASD evaluations resulted in a non-spectrum diagnosis, 31% were accurately classified as not being at risk for ASD using the standard M-CHAT scoring; however, 78% were accurately classified using the Best 7 approach. That is, within the non-spectrum group, 27 more children were accurately classified as not needing an ASD evaluation using the Best 7 approach versus the original approach. Positive predictive values (PPV) were 43% for the original approach and 61% for the Best 7 approach. Evaluations are in progress or will be scheduled for at least 50 more children.

Conclusions:  Preliminary data suggest that the Best 7 approach may be a useful tool in a high-risk population for identifying which children are in need of ASD-specific evaluation versus evaluation for other developmental delays and disorders. Given the high demand for specialized ASD evaluation and the limited number of skilled clinicians who can provide these services, identification of an efficient yet effective screening tool could improve appropriate use of these specialized evaluation appointments, which also has potential to reduce waiting time for young children in strong need of ASD diagnosis and intervention. Further analyses utilizing a larger sample will also explore whether a different cutoff for the Best 7 score can increase sensitivity for identifying ASD in young children.