Late-diagnosed autism may be a different condition

Clinical takeaway: Teens and adults presenting for autism assessment may need a broader differential than an autism spectrum disorder (ASD) pathway alone provides, since anxiety, depression, ADHD, and trauma histories run heavier in this group than in children diagnosed early.

Two groups now share one label despite different genetics, different sex ratios, and different psychiatric burden, and the clinician who helped define autism says that has to end. A diagnosis is supposed to narrow the field, but this one now stretches from a nonverbal child needing constant care to a self-referring adult whose difficulties are real but largely internal.

According to the CDC, prevalence among eight years old children is one in 31, up from one in 150 in the early 2000s. Adults are harder to count but diagnosis rates in large US health systems have risen fastest in adults over the past decade; these are the patients filling assessment waiting lists. This editorial reviews UK, Swedish, and US population data, including a cohort of nearly three million births tracked from 1985 to 2020, to argue the two ends of the spectrum are diverging.

Nearly all of the prevalence increase was tied to diagnoses made after childhood, according to UK primary care records covering more than 65,000 patients. A 2025 genetic analysis split the same way: early-diagnosed autism was dense with polygenic markers for social-communication difficulties and repetitive behavior, while the later-diagnosed profile correlated more strongly with ADHD, major depression, and post-traumatic stress disorder.

Sex ratio tracks a similar divide, shedding more light. In a Swedish birth cohort, autism diagnosed before age 15 held the long-established 3:1 male-to-female ratio, while diagnoses after 15 approached 1:1. As the diagnosis reached further into adulthood, its severity profile thinned. Intellectual disability fell to 6.7% of autism cases from 55.8% in one Swedish population between 2001 and 2020. The patients autism was originally defined around are now a small minority of everyone carrying the label.

Uta Frith, emeritus professor of cognitive development at University College London and one of the researchers who defined autism's cognitive profile starting in 1966, builds the case from published population studies plus an analysis of cultural drivers: broader criteria in successive DSM editions, self-diagnosis, social media, and the DSM-5 allowance for symptoms masked by learned strategies. She notes that the studies she cites used different cutoffs for early versus late diagnosis, and that earlier attempts to define autism subgroups produced inconsistent results and little clinical uptake.

This shift would require a revision of the DSM; there is no announced date for a next edition. The manual was last updated in 2022. The only substantive autism update tightened one criterion, requiring all three social-communication deficits rather than some for diagnosis.

She also argues for more weight on direct observation than self-reporting, and for actively ruling out competing explanations rather than defaulting to inclusion. Frith suggests that the late-diagnosed group has treatable problems that the autism label leaves unaddressed: sensory difficulties, burnout, anxiety, and depression.

"When it was first identified in the 1940s, it referred to a small group of children with severe difficulties in social interaction, communication and behaviour. Over time, the definition has widened to include people with milder traits and no language or learning problems," said Uta Frith, emeritus professor of cognitive development at University College London. "Today, anyone of any age or intelligence level can be diagnosed with autism if they meet certain criteria."

Source: Frith U (2026 Aug 3). Psychol Med. Autism spectrum disorder: has it lost its meaning and is it leading to misdiagnosis?