A Tool Built to Catch Change Found a Third of an SD. So Did the ADOS.

A rapid review published September 22 in Autism (Howlin et al.) pooled every study through March 2025 that used the Brief Observation of Social Communication Change (BOSCC). That's 19 studies and 789 children. The BOSCC was designed to pick up subtle, short-term change that diagnostic tools miss, and it's increasingly proposed as an outcome measure.

The findings

• BOSCC total, intervention arms: g = −0.32 (95% CI −0.42 to −0.22), with low heterogeneity (I² = 22.15%). Lower scores mean improvement.
• Control arms (11 arms, 443 children): g = −0.15, non-significant.
• Social Communication subscale: g = −0.34 (−0.49 to −0.19).
• Restricted and repetitive behavior subscale: g = −0.40, but with a wide interval (−0.74 to −0.06).

The interventions were mostly caregiver-mediated and naturalistic developmental models: PACT, ESDM, ImPACT. These aren't comprehensive ABA programs, so don't read the numbers as an ABA benchmark.

The turn: no better than the ADOS here

The lay summary says the BOSCC “was also found to be better at picking up changes in social communication” than the ADOS. The results don't show that. In the four studies that used both, the ADOS total moved g = −0.32 and the BOSCC total −0.31. The authors call it “a striking similarity.” What the BOSCC did show was narrower confidence intervals.

That matters for payers who set hours or continued coverage by standardized change. Aetna's ABA guide ends coverage after “improvement of less than one standard deviation in all domains for successive authorization periods.” The yardsticks differ: Aetna uses norm-referenced scores like the Vineland-3, while this review measures change against the sample itself. Still, a measure built to be sensitive moved about a third of a standard deviation across an intervention period.

The limits, in the authors' words

• “not all studies reported sufficient data to estimate effect sizes for each subscale, and outcome reporting was inconsistent.”
• “in line with rapid review methodology, we did not conduct a formal risk of bias assessment.”
• The analysis “relies on within-group pre–post change rather than between-group comparisons,” which “may inflate effect size estimates.”
• The ADOS comparison rests on four studies, and they say to “interpret this analysis with caution.”

BOSCC coding is also done offline by trained raters who are blind to the child, timepoint and group. Most clinics can't replicate that.

What you must know or do

• BCBAs writing reauthorizations: when you report change on a standardized measure, state the size of change you'd expect from it over six months. Cite the measure and the published effect sizes for it, and say how close your client population is to the studies behind them.
• Clinical directors with Aetna clients: for anyone near the under-1-SD trigger, put the measure's expected sensitivity and the domain-level data in the progress report before the payer draws its own conclusion.
• Anyone adopting the BOSCC: if your coders aren't blind to timepoint, say so in the report. Don't present the result as equivalent to the published data.