The Autism Diagnostic Observation Schedule: revised algorithms for improved diagnostic validity.

The team rewrote the scoring rules for the Autism Diagnostic Observation Schedule. They blended the old Social Affect and Restricted Repetitive Behavior sections into one cleaner scale.

Kids and adults with and without autism took the new test. The goal was to see if the new math gave sharper yes-or-no answers across ages and language levels.

The revised algorithm raised diagnostic accuracy. It sorted autism from non-autism better than the old version at every age and every language level tested.

In short, the test became less fuzzy and more useful for clinicians.

Oosterling et al. (2010) ran the same new rules on a fresh sample and got the same boost in accuracy. This direct replication gives you confidence the fix works.

Hus et al. (2014) later split the blended score back into two separate, calibrated scales for Social Affect and RRB. Their 2014 method therefore supersedes the 2007 combo score if you want to track each symptom area on its own.

Bennett et al. (2008) sounds a warning: in Hispanic children with mild social delays, the new rules sometimes over-called autism on Modules 2 and 3. The contradiction is only apparent; the 2007 paper averaged many groups, while Terry et al. zoomed in on one ethnicity and symptom level. When you test a child from this background, weigh the score against clinical judgment.

If you still use the original ADOS-G algorithm, switch now. The 2007 revision gives you fewer false positives and false negatives in most populations. Just remember to lower your trust in the score slightly for Hispanic kids with mild social issues, and consider Hus et al. (2014) separate domain scores when you need fine-grained data for treatment planning or research.