The incidence of autism is similar in boys and girls, although boys are diagnosed earlier

The study published in The BMJ is based on solid foundations, both in terms of the track record of the teams behind it and the quality of the Swedish registry system. It works with a large and well-documented population cohort, allowing it to analyse the evolution of autism diagnoses with a consistency that is difficult to achieve in other countries. 

Although there are clear limitations—including that its conclusions can only be considered fully generalisable to the Swedish population, that it is based on registered diagnoses (passive cases), which reflects detection patterns rather than true prevalence, and that it does not incorporate comorbidities that could influence the age of diagnosis—the results are consistent with the most recent evidence. Overall, the data suggest that the traditional difference between males and females may be less pronounced than has historically been assumed. As age increases, diagnoses in females increase significantly and the gender gap tends to narrow, pointing to a significant component of initial under-identification. 

In childhood, many girls have more elaborate social skills, exhibit fewer repetitive behaviours, and tend to have better language development, all of which can attenuate or mask the signs of autism. During puberty, however, many of them experience a more marked decline in social functioning, at which point the difficulties become more visible and therefore more likely to be detected. Added to this are social expectations and biases in clinical and educational practice that favour earlier recognition in profiles considered typical of males.

Taken together, these findings invite us to review and adapt our screening and diagnostic practices, and also to adjust interventions to ensure more equitable identification and more timely and effective clinical and educational support for girls and adolescents.

Traditionally, it has been assumed that autism spectrum disorder (ASD) is up to four times more common in boys than in girls. Multiple genetic and environmental factors have been suggested as possible causes for these differences. 

In a study recently published in The BMJ, researchers used national records to analyse autism diagnosis rates in 2.7 million people born in Sweden between 1985 and 2020, who were followed from birth to a maximum of 37 years of age. This study shows that boys are more likely to be diagnosed with autism during childhood (maintaining the ratio established in previous studies), while girls are diagnosed later during adolescence, reaching parity with boys by the age of 20. There are probably several reasons behind this late diagnosis in girls; on the one hand, their clinical manifestations may be more moderate or perhaps more easily overlooked, and on the other hand, we cannot rule out the existence of a gender bias in the diagnostic criteria for ASD, traditionally based on boys, which may have contributed to a lower diagnosis in girls.  

The importance and significance of this study lie not only in the enormous size of the population sample, which includes millions of people (albeit limited to Sweden), but also in its analysis of the effect of three different time scales: age at diagnosis, calendar period (year of diagnosis) and birth cohort. It should be noted that this is an observational study that did not consider other neurodevelopmental disorders associated with autism, such as attention deficit hyperactivity disorder and intellectual disability, nor did it take into account related genetic and environmental factors, such as parental age or complications during pregnancy. 

This study extends previous research demonstrating relative increases in autism diagnosis amongst females, particularly in adolescents and young adults, and supporting the idea that historically autism has been under-diagnosed in girls and women. The use of population-level healthcare records means that the data are not influenced by self-reporting or recall bias, however there may have been other factors, such as co-occurring mental health conditions, influencing diagnosis timing which were not measured here. Accounting for the whole picture of individuals being assessed is important considering the high rates of conditions such as anxiety, depression, and eating disorders in this group. The authors are not able to explain why the diagnosis ratio is decreasing, as they only had data on when a diagnosis was received, not why it was given. Diagnostic rates are still changing and it remains to be seen whether the male-to-female ratio will continue decreasing, will stabilise, or even will increase again as we continue to 'catch up' with diagnosis of girls and women who were missed earlier in life.

This paper is timely and provides important support for what autism researchers have known for a long-time: that autism is markedly undiagnosed in people assigned female at birth. The study's longitudinal follow-up of young people is essential, because the signs of autism are often only visible at a slightly later age in girls; this means that studies which only look at toddlers may well miss autistic features which are already present in males but not yet visible in females. Moreover, since the study uses national data, this means that it's unaffected by biases in who chooses to opt in or opt out to the data.

Vitally, this study corroborates what studies have already shown us - it's not the first to show that if you follow children over time so as to account for the later emergence of autistic features in girls, you'll find that the prevalence of autism is far more equal across male and female children (Burrows et al., 2022) - that study actually found a 1:1 ratio.

Equally importantly, since we have records of the proportions of autism diagnosed in males and females over the years, these findings - showing us now that autism actually occurs far more evenly across people assigned male and female at birth - support other convincing evidence that very large numbers of autistic women are undiagnosed, especially older undiagnosed women (Stewart and Happe, 2026). And what we know about undiagnosed autistic people is that being undiagnosed is often associated with severe difficulties and even suicidality (Moseley et al., 2025) - so underdiagnosis of autism, like ADHD, should be of serious concern.