Autism's four-to-one sex ratio may be an artefact of when girls get diagnosed

For as long as autism has been counted, it has been counted as a predominantly male condition, with ratios of roughly four boys to every girl repeated in textbooks, screening guidance and research design. A population-based birth cohort study published in The BMJ suggests that ratio is not a fixed biological fact but a moving number that depends heavily on when in life you measure it — and that in Sweden it has almost closed.

The cohort

The study used the Swedish medical birth register to follow 2,756,779 children born alive between 1985 and 2020, with follow-up to 2022 [s1]. Autism spectrum disorder was diagnosed in 78,522 of them, or 2.8%, by the end of follow-up [s1].

The analysis was an age-period-cohort model, which separates three things that are easily confused: how old a person is at diagnosis, what calendar year it is, and which birth cohort they belong to [s1]. Diagnostic criteria, service availability and clinical awareness all change over calendar time, and a study that does not separate those from age effects will misattribute one to the other.

What the numbers show

Age-specific incidence of autism rose for every calendar period and every birth cohort between 1985 and 2020 [s1]. In 2020–2022, incidence peaked at 645.5 per 100,000 person-years for males at ages 10–14, and 602.6 for females at ages 15–19, then declined [s1].

That five-year offset in the peak is the study's core observation, stated as a rate. Girls are not diagnosed less often so much as later.

Consistent with that, the male-to-female ratio decreased with increasing age at diagnosis, and — for those older than 10 — decreased by calendar period as well [s1]. By the final year of follow-up in 2022, the cumulative male-to-female ratio for autism incidence was 1.2 by age 20 [s1]. Projecting the trend forward, the authors suggest the cumulative ratio would reach parity at age 20 by 2024 [s1].

A ratio of 1.2 is not four to one. The authors conclude that the male-to-female ratio may be substantially lower than previously thought, to the extent that in Sweden it may no longer be distinguishable by adulthood, and that this highlights a need to investigate why girls and women receive diagnoses later [s1].

What this does and does not settle

The finding is about diagnosed autism, and it is about Sweden. A registry study cannot distinguish between three explanations for a narrowing ratio: that autistic girls were always there and are now being found; that broadening criteria are capturing people who would not previously have met them, disproportionately female; or some combination. The rising incidence in every period and cohort [s1] is compatible with all three.

The projection to parity by 2024 is an extrapolation of an observed trend, not an observation [s1]. Trends flatten.

Sweden also has a national registry, near-universal health coverage and a specific diagnostic culture. Whether the same convergence is occurring in health systems with different referral pathways is an open question this study cannot answer.

An editorial published alongside the paper frames the implication as a question of equal recognition of autism in girls and women [s2].

Why girls might be diagnosed later

A study published in the journal Autism six days later offers one mechanism, at the level of the assessment itself.

Researchers examined "social words" — words referring to other people — produced by 138 age- and IQ-matched autistic and non-autistic young people aged 6 to 15, during a single task from the Autism Diagnostic Observation Schedule, Second Edition [s3].

Girls used significantly more social words than boys across both diagnostic groups [s3]. But the gap between autistic and non-autistic participants was larger among girls than among boys, with non-autistic girls using the most social words of any group [s3].

The implication the authors draw is about the yardstick rather than the child: non-autistic girls' social language sets an especially high comparison bar for autistic girls, who often report trying to blend in with other girls [s3]. An instrument calibrated on a mixed-sex or male-typical reference will register an autistic girl as closer to typical than she is relative to her own peer group.

The authors argue that assessment should be sensitive to sex-associated differences, and that interventions should consider the norms of a young person's actual social circle [s3].

This is a single task from a single instrument in 138 matched participants, and it measures word use, not diagnostic outcome. It does not demonstrate that ADOS scoring causes delayed diagnosis in girls. It describes a plausible route by which it could.

What to watch

The specific claim to test elsewhere is the age offset — female incidence peaking at 15–19 against male incidence peaking at 10–14 [s1]. If registries in other countries show the same five-year lag, then the historical sex ratio was substantially a measurement artefact of following people only through childhood, and adult diagnostic services should expect demand that current planning assumptions do not anticipate.

Sources

• [s1] Time trends in the male to female ratio for autism incidence: population based, prospectively collected, birth cohort study. BMJ, 4 February 2026. https://doi.org/10.1136/bmj-2025-084164
• [s2] Towards the equal recognition of autism in girls and women. BMJ, 4 February 2026. https://doi.org/10.1136/bmj.s120
• [s3] Sex-specific metrics for success: gaps in social word use are bigger for autistic girls than boys. Autism, 10 February 2026. https://doi.org/10.1177/13623613261417535