Autism diagnoses are rising, but detection may be driving much of the increase

Growing diagnosis rates appear to reflect better detection, broader criteria and changing clinical practice more than a clear surge in underlying autism, though researchers say a biological increase cannot yet be ruled out.

Autism diagnoses have risen sharply over the past two decades, but the most important question is not simply whether more people are being labelled. It is why the numbers are climbing and how much of that rise reflects changes in detection, record-keeping and diagnostic boundaries rather than a true increase in the underlying condition. Recent research suggests the answer is likely a mixture, with the balance tilting heavily towards how autism is being found and defined. According to a 2026 study in the Journal of Autism and Developmental Disorders, diagnoses in a US national health survey rose among children and adolescents from 2013 to 2022, with nearly 4% of children estimated to have received an autism diagnosis by the end of the period, although the figures were based on parent reports and a survey response rate of about 50%.

That pattern fits with broader global evidence. A 2022 systematic review in Autism Research found a median prevalence of about 1% across the studies it examined, but the estimates varied enormously, from 1.09 to 436 per 10,000 people. The authors said that spread was better explained by differences in study design, case-finding methods and social context than by biology alone. In other words, autism does not appear to be a single fixed number waiting to be discovered; the way researchers search for it can dramatically change what they find.

This matters because prevalence is not the same as incidence, and neither is the same as the actual biological rate at which autism develops. Administrative records usually capture diagnosed cases, not every person who meets diagnostic criteria. That distinction helps explain why some studies report dramatic increases while others, using more consistent screening methods, find a steadier picture. The challenge is especially clear when comparing health-system data with population screening, since changes in access to assessment, clinician awareness and referral pathways can all push the figures upwards without any shift in biology.

The profile of those being diagnosed has also changed. Research discussed in the article suggests that the increase is not confined to children, or even to boys, which is one reason some scientists argue that earlier under-recognition is being corrected rather than a wholly new wave of autism emerging. At the same time, other researchers warn that broader awareness can lower the threshold for diagnosis and pull in people whose difficulties may stem from other conditions. That tension helps explain why the debate has become so fraught: expanding recognition can be both a form of progress and a source of diagnostic drift.

A second source of uncertainty is that the diagnosis itself has changed over time. As clinical manuals have been revised, the autism spectrum has widened, narrowed and been redefined in ways that affect who qualifies for a label. A 2022 review in Autism Research also found that study methods strongly shaped reported prevalence, with record-review surveillance producing higher rates than insurance databases. The implication is that the rise in autism counts cannot be read as a simple thermometer of disease; it is also a measure of changing clinical practice, social awareness and the mechanics of diagnosis.

Even so, the possibility of a real increase in risk cannot be dismissed entirely. Researchers continue to examine genetic susceptibility, parental age, prenatal health and environmental exposures, but those factors are best understood as influences on individual risk rather than proof of a sudden population-wide surge. For now, the strongest evidence suggests that much of the apparent rise comes from better detection, broader criteria and altered expectations about who autism can affect. The harder question, and the one science is still trying to answer, is how much of the increase remains once all those measurement effects are stripped away.