Nearly half of these children have an average or above-average IQ

The Autism and Developmental Disabilities Monitoring Network estimates autism
prevalence by reviewing the records of 8-year-olds across 11 US sites — not
by asking parents, and not by counting diagnoses. Trained abstractors read
evaluations from clinics, specialist programmes and, at most sites, school
special education records; **experienced clinicians then decide case status
from the described behaviour.**

For 2014: 16.8 per 1,000 — one in 59.

The distribution people get wrong

Among the nine sites with sufficient data on intellectual ability:

Nearly half are in the average or above-average range, and **more than two
thirds** are above the threshold for intellectual disability. The picture of
autism as inseparable from intellectual disability describes under a third of
the children this survey identified.

The gap between noticing and evaluating

**Although mention of developmental concerns by age 36 months was documented
for 85% of children with ASD, only 42% had a comprehensive evaluation on
record by age 36 months.**

The median age of earliest known diagnosis was 52 months — and **did not
differ significantly by sex or race and ethnicity.**

Somebody wrote down a concern about six children in seven before their third
birthday. Fewer than half were evaluated. The delay is not in noticing; it is
in what happens after someone notices, and that is a service-capacity problem
rather than an awareness one.

Variation that is about places, not children

Prevalence ranged from 13.1 to 29.3 per 1,000 across the 11 sites — more
than double, between communities in the same country. **Boys were four times as
likely as girls to be identified.**

By race and ethnicity, prevalence was **higher for white than Black children,
and both higher than Hispanic children** — but the trend matters: **differences
between Black and white children have diminished in most sites, while the gap
for Hispanic children remained notable.** A gap that closes when identification
improves was a gap in identification.

The ADDM sites are not a representative sample of the United States, so the
combined estimate cannot be generalised to all US 8-year-olds — a
limitation the report states plainly and that is routinely dropped when the
number is quoted.

Changing the ruler without breaking the measurement

These estimates use DSM-IV-TR criteria. The 2013 DSM-5 changed the
criteria considerably, which could shift prevalence for reasons that have
nothing to do with children.

So 85% of the records were reviewed again under a DSM-5-based definition,
and the two were compared:

• DSM-IV-TR counts exceeded DSM-5 counts by less than 5%
• About 86% overlap between the two definitions (kappa = 0.85)

The definitions agree closely — which is what lets the series continue across
the change. **From surveillance year 2016 the DSM-5 definition became the
basis**, with DSM-IV-TR kept in a limited area for comparison.

**With prevalence ranging from 13.1 to 29.3 per 1,000 across communities, the
need for behavioural, educational, residential and occupational services
remains high.**

*Source: Centers for Disease Control and Prevention, MMWR Surveillance
Summaries.*