Two-Step Screening of the Modified Checklist for Autism in Toddlers in Thai Children with Language Delay and Typically Developing Children.

The team gave the Thai version of the M-CHAT to toddlers who had language delays and to typically developing peers.

They used the two-step method: parents filled out the 23 items first, then a clinician followed up on failed items.

All children were then checked with gold-standard tests so the team could see how well the screener worked.

Counting any child who failed three or more items gave 91% sensitivity and 99.7% specificity for autism.

That cutoff caught most kids with ASD and rarely flagged typical children.

Tsai et al. (2019) ran a near-copy study in Taiwanese toddlers and got similar high accuracy with the newer M-CHAT-R/F, showing the two-step logic travels well across East-Asian cultures.

Samadi et al. (2015) looked like a contradiction: in Iranian preschoolers their local Hiva scale caught more ASD cases than the M-CHAT. The gap disappears when you note they screened older, community children while Pornchada et al. tested younger, clinic-referred toddlers—age and setting drive the difference.

Pandey et al. (2008) had already warned that screening before 24 months misses more cases; the Thai study took that cue and focused on the 24-30-month sweet spot, aligning its strong results with the age advice from the earlier US work.

If you screen toddlers in Thailand—or any Thai-speaking family—use the total-score cutoff of three failed items; it is simple, free and highly accurate.

For BCBAs elsewhere, the study is a reminder to follow the two-step process and to pick age-appropriate cutoffs rather than relying on follow-up interviews alone.