A Spanish-Mexican translation of the Modified Checklist for Autism in Toddlers, Revised (M-CHAT-R*; Robins, Fein, & Barton, 2009) was administered to a sample of 420 Mexican children aged 18 to 36 months. The importance of early detection in the outcome of individuals with Autistic Spectrum Disorder (ASD) is now well established; however in Mexico there is often a significant time lag between parent concerns and confirmation of diagnosis.
The M–CHAT is a screening instrument designed to detect ASD in young children that was introduced into Mexico by the author and Dr. Deborah Fein as part of Autism Speaks’ Pan American Autism Awareness Training Initiative 2009 and Clínica Mexicana de Autismo.
Objectives:
Methods:
The M-CHAT-R Spanish-Mex. version was applied to a sample of 300 Mexican typical children (control group) aged 18 to 36 months compared with a sample of 120 Mexican children with ASD of the same age. A binary logistic regression was used in order to verify the statistical reliability, sensitivity and predictive ability of the M-CHAT-R Spanish-Mex., in which the dependent variable is a dichotomous measure (“ASD children =1” and “control group=0”). The independent variables consist of the 20 items contained in the M-CHAT-R Spanish Mex. version which also are dichotomous (0 if the answer to the question suggests no ASD risk and “1” if the answer to the question suggests ASD risk).
Results:
The results of binary logistic regression model suggest that ten out of the 20 items of the M-CHAT-R Spanish-Mex. version instrument are enough to predict children with ASD accurately. These items included 1 (follows point) , 2 (pretend play), 6 & 7 (points), 8 (peer interest), 9 (brings to show), 11 social smile), 14 (eye contact), 17 (seeks parent’s attention), & 18 (understands). Predictive ability of the ten most sensitive items indicates that the selected questions will predict accurately 80% of the ASD cases, and also 97% of the control group cases would be classified correctly (global percentage of correct prediction was 92%); (goodness of fit test Cox and Snell R2=.632; Nagelkeke´s R2=.843; cutoff value=.5). The measure of the internal consistency of the test score for this sample indicates a good reliability, Cronbach's α (alpha) =.837.
Conclusions:
This study highlights some of the difficulties in diagnosing very young children reliably and emphasizes the need for the development and evaluation of instruments that are sensitive to ASD in a very young population. The M-CHAT R is, however, available in a Spanish version for a Mexican study and may provide a suitable “Level 1 screen” for the toddlers in Mexico; to date, this study is the first quantitative/empirical research study available in Mexico that reports the statistical reliability, sensitivity and predictive ability regarding the results of the M-CHAT-R Spanish Mex. version. Future research will examine the performance of the M-CHAT-R, Spanish Mexican version in a Level 1 sample.