Effectiveness of using the Modified Checklist for Autism in Toddlers in two-stage screening of autism spectrum disorder at the 18-month health check-up in Japan.

Doctors in Japan added the M-CHAT to the regular 18-month check-up.

Parents filled out the 23 yes-no questions while they waited.

A nurse scored the sheet the same day.

Kids who failed were invited back for a short play-based follow-up.

The screen caught about half of the toddlers who later got an ASD diagnosis.

It almost never flagged a child who did not have ASD.

In short: good at avoiding false alarms, weak at finding every case.

Kerub et al. (2020) saw the same pattern in Israel—high specificity, modest sensitivity—so the trade-off looks real across countries.

Scarpa et al. (2013) found the opposite problem in rural, low-income US families: lots of false alarms.

The difference is likely cultural and educational, not the tool itself.

Li et al. (2018) doubled detection in Shanghai by adding a five-minute observation and a phone call, showing you can boost sensitivity without losing specificity.

Use M-CHAT at 18 months, but do not stop there.

Pair it with brief observation or a second questionnaire, especially if families have limited education or speak another language.

Track who never returns for follow-up; they are the kids most likely to slip through the cracks.