The American Academy of Pediatrics recommends screening for autism spectrum disorders (ASD) as routine pediatric practice. Surveillance of ASD has consistently shown lower prevalence rates among Hispanics than Whites, the reasons for which are not clear but may include differences in recognition or diagnosis. Few studies have examined screening methods within the Hispanic population in the United States. To investigate ASD prevalence rates and appropriate screening methods in this population, we initiated a universal screening program in a county healthcare system serving primarily low income Hispanics.
Objectives:
Screen children born in 2006 attending well-child pediatric visits at two county pediatric clinics, and for this report, to examine fail rates of M-CHAT items in Hispanics compared to other ethnic groups.
Methods:
The Modified Checklist for Autism in Toddlers (M-CHAT) was given to parents of children aged 16-30 months at their clinic appointment, in English or Spanish, as requested. These were scored on-site and results were provided to pediatricians for further review or referral. Among subjects who failed the M-CHAT based on M-CHAT cut-off values, we calculated the proportion who failed for each of the 23 items. We compared these by factors including ethnicity and language using the chi-square test (p-value <0.05 indicating statistical significance.) Results:
Overall, 26% failed the M-CHAT, which was significantly higher in Hispanics (28%) than Non-Hispanics (17%), so we examined whether specific items contributed disproportionately to this difference. Examining the first failed instrument only, we compared Hispanics with positive screens (n=403) to Non-Hispanics (n=58). Of the six “critical” questions on the instrument, the only significant difference in was for item 15, which Hispanics were less likely to fail than Non-Hispanics, 6.5% vs. 13.8%, respectively (p=0.04). Children of Hispanic ethnicity had significantly more fails on items 4 (p=0.01), 10 (p=0.02), and 17 (p=0.02), and significantly less fails on items 18 (p=0.02), and 22 (p=0.02) than Non-Hispanics. Within the Hispanic population we compared fail rates of items by instrument language. Hispanics who chose a Spanish translation were more likely to fail items 3 (p=0.001), 4 (p=0.04), and 21 (p=0.01), but less likely to fail item 22 (p=0.006), than those who chose the English version. There was little difference in item fail rates by maternal education.
Conclusions:
This population's M-CHAT fail rate is twice that found in other studies of similar aged children. This may be due to a higher prevalence of ASD in this primarily Hispanic population, or to differences in interpretation of the instruments by these parents. The
higher and lower fail rates on specific M-CHAT items among Hispanics may indicate these questions do not carry the same cultural relevance for Hispanics as for others. For example, item 4 asks if the child plays peek-a-boo, which may not be a common game in this population. Items with differing fail rates for Hispanics using the Spanish translation may indicate differences created upon translation of the original instrument. Further examination of ASD screening in Hispanics and refinement of screening methods and instruments are needed to help serve this growing population.