Replication and Comparison of the Newly Proposed ADOS-2, Module 4 Algorithm in ASD Without ID: A Multi-site Study.

Austin et al. (2015) tested the new ADOS-2 Module 4 rules on adults with autism who do not have intellectual disability.

Teams at several clinics gave the same people both the old and new scoring rules. They checked how many true autism cases each rule caught and how many non-autism cases it wrongly flagged.

The updated rule caught more real autism cases, but it also called more non-autistic people autistic.

Accuracy was best for women, people with very high or very low verbal IQ, and anyone age 16 or older.

Hus et al. (2014) first showed the rule worked well in a single-site study. Austin et al. (2015) now repeats that test across many sites and still finds the trade-off: higher sensitivity, lower specificity.

Bennett et al. (2008) saw the same problem in Hispanic children with mild social delays. Both papers warn that revised ADOS rules can over-label certain groups, so you must check the person’s full picture.

Hong et al. (2021) and Oosterling et al. (2010) show toddler and child modules replicate well. Module 4 now joins that list, proving the whole ADOS-2 family can be trusted if you watch the cut-offs.

If you test verbally fluent teens or adults, the Module 4 update will help you miss fewer autism cases, especially in women and extreme IQ ranges. Just remember it also raises false positives, so always pair the score with clinical judgment and extra data before you share results with families or refer for services.