Accuracy of Modified Checklist for Autism in Toddlers (M-CHAT) in Detecting Autism and Other Developmental Disorders in Community Clinics.

Toh et al. (2018) watched how well the 23-item M-CHAT picked up autism and other delays in regular community clinics.

Doctors gave the checklist to parents of toddlers coming in for routine visits, then noted who later got a full diagnosis.

The screen caught only some kids with autism, especially missing those under 21 months.

It did a better job flagging any broader developmental delay, so a fail still tells you "check this child" even if the label is not autism.

Allison et al. (2008) built the earlier Q-CHAT, showing parent answers can quantify traits at 18 months, setting the stage for later checklist work.

Hedley et al. (2015) tested the play-based ADEC in hospital clinics and found it caught 93-94 % of ASD toddlers, a much higher hit rate than the M-CHAT shown here.

The difference is setting: hospital kids were already referred and likely showed clearer signs, while community clinics catch a wider, lower-risk group.

Laguna et al. (2025) pushed detection even earlier, using baby cries and AI to reach 90 % accuracy, hinting that sound analysis could one day back up the M-CHAT.

Keep using the M-CHAT at 18- and 24-month checkups, but treat any fail as a prompt for fast follow-up, not a diagnosis.

If the child is under 21 months, plan a second screen or move straight to an ADOS/ADI-R referral, since many autism cases slip through at this age.