16689
Validation of the Modified Checklist for Autism in Toddlers-Revised with Follow-up (M-CHAT-R/F)

Friday, May 16, 2014: 11:06 AM
Imperial B (Marriott Marquis Atlanta)
D. L. L. Robins1, K. A. Casagrande2, M. L. Barton3, C. M. A. Chen3, T. Dumont-Mathieu3 and D. A. Fein3, (1)PO Box 5010, Georgia State University, Atlanta, GA, (2)Georgia State University, Atlanta, GA, (3)Psychology, University of Connecticut, Storrs, CT
Background:   Early detection and early intervention for children with Autism Spectrum Disorder (ASD) improve outcomes. Because receiving effective intervention depends on the child being detected and diagnosed, early screening is crucial. The M-CHAT is one of the most widely used screening tools, but left room for improvement in its psychometric properties.

Objectives:   Validate the 2-stage Modified Checklist for Autism in Toddlers, Revised, with Follow-up (M-CHAT-R/F) in a low-risk sample.

Methods:   The M-CHAT was revised by eliminating three items, rearranging items to avoid positive response bias, simplifying language, and providing developmentally-appropriate examples. Children (n=16,115) were screened during 18- and 24-month well-child care visits in metro-Atlanta and Connecticut. Parents of at-risk cases completed the Follow-up with researchers by telephone. Children who continued to show risk for ASD were invited for a free diagnostic evaluation. In addition, missed cases were identified by physician flag for ASD concerns and secondary play-based screening.

Results:   92.6% of children screened negative; of those who completed the Follow-up, 63.2% no longer showed ASD risk . Of the 348 cases that continued to show risk for ASD, 221 attended evaluations; 105 were diagnosed with ASD, 104 were diagnosed with other developmental delays/concerns, and 12 were typically developing. The optimal scoring algorithm, determined by receiver operating characteristic curves was based on total score (area under the curve=.907): Children whose total score was 0-2 were low-risk, and no further assessment was needed. Children whose total score was 3-7 required administration of the Follow-up, which clarified responses and elicited examples for at-risk responses; Follow-up scores of 2 or more warranted evaluation. Children whose initial score was 8 or higher can bypass Follow-up and be referred immediately for evaluation. The M-CHAT-R/F identified 105 cases of ASD; an additional 18 cases were identified based on other strategies. Sensitivity has an upper bound of .85, specificity=.99. Positive predictive value (PPV) for ASD was .48, and PPV for any developmental disorder/concern was .95. Likelihood ratio for positive screens was 114.05. Physician concerns identified 24% of ASD cases. Direct comparison between M-CHAT-R and original M-CHAT indicated that the rate of ASD detection increased (67/10,000 vs 45/10,000; p=.003) while the overall screen positive rate decreased from 9.15% to 7.17% (Χ2 (1, n=35,060)=39.62, p<.001).

Conclusions:   The M-CHAT-R/F is a valid tool to screen for ASD during toddler check-ups. The questionnaire demonstrates improvements compared to the original M-CHAT, and has strong psychometric properties. Although the 2-stage screener identifies many children who do not have ASD, 95% of children who screen positive on the M-CHAT-R/F require evaluation to identify specific early intervention needs. The M-CHAT-R/F identified many more children with ASD than physicians’ clinical judgment of ASD risk (105 vs. 30); however, physicians identified nine ASD cases missed by the M-CHAT-R/F. Also, although physicians were asked to indicate ASD risk, some did not follow this procedure consistently and may not have indicated risk if the screener was clearly positive. Therefore, the combination of surveillance and screening is likely to be the most powerful method for detecting ASD in toddlers.