Autistic Women Have Always Been Here

And the evidence is catching up!

For most of autism's diagnostic history, clinicians have repeated some version of the same number: autism is about four times more common in boys than in girls. That number has shaped which children get screened, which symptoms get recognized, and which ones get explained away as anxiety or shyness or a hormonal thing. Generations of autistic women and girls have been missed because the people doing the looking were taught to look for boys.

In this episode, I walk through a 2026 study published in the BMJ that followed 2.7 million Swedish birth records over 35 years. The findings suggest the four-to-one ratio is collapsing — and in adolescent and adult diagnosis, it has either evened out or flipped. Autistic women were never rare. We were just being missed.

Topics covered include:

• Why the four-to-one male-to-female ratio has dominated autism research and clinical practice
• What "masking" or "camouflaging" means, and why it has cost so much
• The biological vs. diagnostic explanations for the apparent gender gap — and what this study tells us about both
• Why this looks like a catch-up effect rather than a sudden surge in autistic girls
• What this means for autistic adults who got missed for decades

If you'd like to know more about topics discussed in this episode, check out:

"Time Trends in the Male to Female Ratio for Autism Incidence: Population Based, Prospectively Collected, Birth Cohort Study" by Caroline Fyfe et al.

"What Is the Male-to-Female Ratio in Autism Spectrum Disorder? A Systematic Review and Meta-Analysis" by Rachel Loomes et al.

"'Putting on My Best Normal:' Social Camouflaging in Adults With Autism Spectrum Conditions" by Laura Hull et al.

"Clinical Characteristics and Problems Diagnosing Autism Spectrum Disorder in Girls" by Hannah Young et al.

Theme music: "Everything Feels New" by Evgeny Bardyuzha.

All episodes written and produced by Kristen Hovet.

Send in your questions or thoughts via email or audio or video recording for a chance to be featured on the show! My email address is otherautism@gmail.com 

Large files can be sent for free via WeTransfer.

Buy me a coffee!

Buy The Other Autism merch. Use code FREESHIP for free shipping on orders over $75 USD!

The views, opinions, and experiences shared by guests on this podcast are their own and do not necessarily reflect those of the host or production team. The content is intended for informational purposes only and should not be taken as medical or professional advice. Please consult with a qualified healthcare provider before making any decisions related to your health, fitness, or wellness. 

Transcript

Kristen Hovet 0:00
If you're an autistic woman or someone born female who got diagnosed late, or someone still wrestling with whether the label even fits, you've probably encountered some version of this sentence in your life. Autism is more common in boys or in males. And then a number, usually somewhere around four to one, four boys for every one girl, is diagnosed as autistic. That number has shaped how clinicians look for autism. It has shaped which children get screened. It's shaped which symptoms get recognized and which ones get explained away as anxiety or shyness or just a bad attitude or a hormonal thing or a sensitive personality. Generations of autistic women and girls and those born female have been missed because the people doing the looking were taught to look for boys. Today I want to tell you about a 2026 study that came out in the BMJ, formerly the British Medical Journal. The researchers looked at 2.7 million birth records from Sweden spanning 35 years, and what they found will, I hope, finally start putting that four to one number to bed. Today we're going to talk about how a number that's defined autism for decades is quietly and on a massive scale falling apart. Today's episode is about a study published in the BMJ this year. It's titled Time Trends in the Male to Female Ratio for Autism Incidence: Population-Based, Prospectively Collected Birth Cohort Study. The lead author is Caroline Fyfe, and the senior author is Sven Sandin. Both are at the Karolinska Institutet in Sweden. This is a big ambitious piece of epidemiology. They followed every single child born in Sweden between 1985 and 2020, 2.7 million people in total, and tracked who received an autism diagnosis through the end of 2022. That gives them follow-up out to age 37 for the oldest people in the cohort. And it lets them ask a question that until now no study has been able to answer at this scale. How has the male to female ratio for autism diagnosis actually changed over time, once you separate out age at diagnosis, calendar year, and birth cohort? The answer is going to surprise some of you. Maybe not all of you, especially those who've been listening to the podcast for a while, but it's going to potentially surprise a lot of clinicians. Let me set the table. For most of autism's diagnostic history, the male to female ratio, clinicians call it the MFR, has been reported as somewhere between 3 to 1 and 4 to 1. Most clinical training materials until quite recently taught somewhere around that 4 to 1 ratio as the working assumption. And that number has led to real consequences. When you assume autism is four times more common in males, those born male, you build screening tools that are validated mostly on boys. They look at male populations to validate those tools, in other words. You publish case studies and research about boys. And then when a girl walks into the office, a person who's born female, and they're presenting differently, the system has nothing to match her against. She gets called anxious or shy or sensitive. And those were words that were used to describe me when I was a kid. Or when that little girl gets older, depressed, or borderline, as in borderline personality disorder, anything but autism. Now researchers have been pushing back on the four to one number for years. There's a 2017 meta-analysis by Loomes, Hull, and Mandy. They looked at 54 studies. The overall pooled ratio was 4.2 to 1, but the high quality studies in their analysis showed a ratio of about 3.3 to 1. And when researchers actively went out and screened the general population instead of waiting for kids to show up at the clinics, the ratio dropped to about 3.21 to 1. That was a clue, going out into the real world. Then there are the explanations researchers have offered for why the ratio appeared so skewed in the first place. There are basically two camps here. Camp one, it's biological. There's a theory called the female protective effect, which proposes that female bodies need a higher genetic load for autistic traits to become noticeable. There's the greater male variance theory, which says males just have more genetic variation in autism-related alleles. So more of them end up above the diagnostic threshold by chance. And there's the extreme male brain theory popularized by Simon Baron Cohen, which frames autism as an exaggeration of typically masculine cognitive traits. I have my own thoughts about that one. I think I probably have gone into that before in a previous episode. And I would love to sit down with Simon Baron Cohen one day and see where he's at with that theory. Camp two, it's diagnostic. The ratio looks skewed because we're bad at recognizing autism in girls, in women, in those born female. They simply present differently. They mask more. Their interests look more socially acceptable. So they don't trigger the same red flags. Their distress gets attributed to other things, like I said, anxiety, even eating disorders, depression, obsessive-compulsive disorder. By the time anyone considers autism, they've been in the mental health system for years. And quick definition while we're here: masking, sometimes called camouflaging, is a term for the constant, often unconscious effort that some autistic people make to hide their autistic traits in order to fit in. Copying other people's facial expressions, rehearsing conversations before they happen, suppressing stims, forcing eye contact even when it hurts or is really uncomfortable, researching social rules the way other people research stocks. Masking, camouflaging, it's exhausting. It's expensive in terms of energy, and it tends to be more common, though absolutely not exclusive to autistic women and girls. We mask because we figured out very young that the consequences of being visibly autistic were worse than the consequences of pretending not to be. The cost of all that masking, by the way, includes burnout, late diagnosis, identity confusion, and a whole lot of mental health diagnoses that turn out eventually to have been autism the whole time. And I want to be clear that masking and camouflaging isn't like a conscious thing or it's something that, you know, when I'm doing it, I don't know I'm doing it. It's not like I'm trying to uh trick anyone or dupe anyone. And I also assume that that's what everyone else is doing. I assume that this is the same effort that everyone else has to put in. Once I learned about autism, I realized that's not true. That's not the case. I think it's been like portrayed by some people as this like thing that autistic people do to trick others, but it's not like that. It's kind of like um an overextension or exaggeration of the natural inclination to mimic others. And we do it because we don't naturally know always how to enact social behaviors in order to fit in, or we feel that we've been judged, and or we feel like we've been judged harshly in social circumstances, and we're very sensitive to that. We tend to be. And so we learn ways and strategies that take a lot of like cognitive energy basically to enact. And we're doing it because when we actually do it and we, you know, get by in a social context and people talk to us more, people like us more, we avoid being ostracized, we avoid having, you know, hurtful comments flung our way, then that helps perpetuate those camouflaging and masking behaviors. They, it's like this feedback loop. We're gonna keep doing this because obviously it protects us. Okay, with that on the table, let's talk about what Fyfe and her team actually did. Quick methods overview. Sweden is a researcher's dream when it comes to this kind of study. They have a publicly funded health system, they have national registers that go back to the 1970s. Every Swedish resident gets a unique personal identification number, which means researchers can link birth records to medical records to outcomes across someone's entire life. The medical birth register covers 99% of all births in Sweden. You cannot do this kind of work in most countries. You can do it here, though. The team pulled every live-born child registered in Sweden between 1985 and 2020. That's 2.7 million people. They restricted the analysis to children whose parents were both born in Sweden just to keep the population homogenous across that 35-year window. Then they linked those records to the National Patient Register and identified everyone who received an autism diagnosis from age two onward through the end of 2022. And then they did something statistically clever. They ran what's called an age-period cohort analysis. The idea is that when you see autism rates change over time, that change can come from three different sources, and you want to be able to tell them apart. There's the age effect, the fact that autism gets diagnosed at different rates depending on how old someone is. There's the period effect, changes in any given year that affect everyone, like the introduction of new diagnostic criteria. And there's the cohort effect, differences between generations, like older parental ages or improved screening for certain birth years. Untangling those three threads is hard. And most previous studies haven't had the data to do it. Fyfe's team did. They also ran sex as an interaction term across all three of those effects, which is the part that lets them answer the question we actually care about. How has the male to female ratio changed separately with age, with calendar year, and across birth cohorts? One thing I want to flag before we get into the findings: in the patient and public involvement section of the paper, the authors note that no patients or members of the public were involved in the study. They acknowledge that this is a gap, and they say it's being addressed in future work. I appreciate the honesty here. It does mean the study is a clinical epidemiological view of autism, not a community-informed one. So that's worth keeping in mind. Here's what they found. Finding one: Autism diagnoses are way up in everyone. So that's the fun headline finding, right? Autism diagnosis rates went up roughly tenfold over the study period. Ah, tenfold. That's... to give you a concrete example, in the 10 to 14 age group, the diagnostic rate in 2000 was about 56 per 100,000 person years. That's epidemiology speak for per 100,000 people per year of observation. That's a bit complex. I'm not an epidemiologist, so I'll just leave it at that. By 2022, it was 564 per 100,000 person years. That's the same age range, same country, same diagnostic system, just two decades apart. Now that's not because autism itself became 10 times more common. The researchers are clear about this: that the increase reflects three things: better awareness, broader diagnostic criteria, especially after the introduction of ICD 10 in 1997 and the DSM 5 in 2013, and more comprehensive screening systems, particularly in Sweden, where every child gets multiple developmental check-ins between birth and age five. By the end of follow-up, 2.8% of the entire study cohort had received an autism diagnosis. That's 78,522 people. Just to put that in context, the most commonly cited global figure is around 1%. Sweden is finding nearly three times that just by virtue of looking properly. Ring any bells from episode 48? It should. Finding two: the male to female ratio is collapsing. This is the finding that justifies the whole paper. Ready? In the youngest age group, zero to four years old, the male to female ratio is still high, around three or four to one. Roughly what we've always seen. So for very young children getting diagnosed, the historical pattern still holds. But watch what happens as age at diagnosis goes up. By age 10 to 14, the ratio drops to about 1.3 to 1. By age 15 to 19, it's 0.7 to 1. Let me say that again. 0.7 to 1. That means in Sweden, between 2020 and 2022, more female teenagers in the 15 to 19 age group were being diagnosed as autistic than males. By age 20 to 24, it's 0.8 to 1. And that's still tilted towards females. By age 25 to 29, it's one to one, equal. By age 30 to 34, it's 0.9 to 1, slightly more female than male. In other words, the four to one ratio is a story about young children. Once you start looking at adolescent and adult diagnosis in 2020, Sweden, the ratio either evens out or actively flips. There are years in age groups where more women are getting diagnosed than men. Finding three: cumulative diagnosis is approaching parity. Now the per year, per age numbers above are what we call period rates. They tell you about one slice of time. The bigger question is across an entire lifespan, how many men and how many women, how many males and how many females end up with an autism diagnosis? That's the cumulative male to female ratio, the CMFR in the paper's shorthand. By the end of 2022, the cumulative ratio for autism diagnosis by age 20 in Sweden was 1.2 to 1. Not 4 to 1, not 3 to 1, 1.2 to 1. And when the researchers projected the trends toward 2024, the cumulative ratio at age 20 dropped to 1 to 1. Parity. By age 20 in Sweden, males and females are now essentially just as likely to be diagnosed autistic. The authors put it this way. I'm going to read directly from the paper here. They wrote, quote, this male-to-female ratio may therefore be substantially lower than previously thought, to the extent that in Sweden it may no longer be distinguishable by adulthood. End quote. That sentence sitting in the abstract of the BMJ in 2026 is one of the quietest revolutions I've seen in autism research. Woo! Finding four: This is a catch-up effect, not a flood of new autistics. Okay. Now I want to be careful here because I can already hear the bad faith readings. Are autism rates really skyrocketing in girls? That's the framing certain corners of the internet are going to put on this, and it's wrong. What the researchers found is what they call a catch-up effect. Not catch up as in the red gooey stuff that people put on food, but catch up, catching up with, catch-up effect. The number of girls and women being diagnosed isn't increasing because females are suddenly becoming autistic. The autistic females have been there the whole time. We're just finally being counted. The pattern in the data is unmistakable. The diagnosis is happening later for females than for males. The peak age of diagnosis for males in the most recent period was 10 to 14. For females, it was 15 to 19. So girls are still being missed in childhood on average more often. But by adolescence and adulthood, when their masking starts to break down, when they end up in mental health services, when they encounter clinicians who've been trained on the newer literature about autistic females, they're finally getting recognized. I don't know, for those of you who haven't heard the oldest episodes where I talked about my diagnosis, it came after I had cancer and I had major surgery. And then my margins were not clear. So I ended up getting radiation and chemotherapy. And then yeah, it was it was a series of uh medical traumas. And I I became for a time incapable of masking. I had no idea what was going on with me. And the trauma of the medical treatments didn't explain everything. So I sought answers and I ended up with an autism diagnosis. When I looked into it, when I read about it, it was like a series of aha moments, a series of like, it felt like self-discovery that I had never felt before. And I also describe a few times in the podcast looking back over my life and seeing it with new, with new lenses, basically, having more compassion for myself, having more understanding, just feeling, I don't know. There were a lot of low moments too. I don't want to make it sunshine and the unicorns. I mean, there there were definitely sunshine and unicorns as part of it, but it's hard too, because you look back and you go, what!? You think about the things that you missed and what could have happened had you have known all of this all along. So yeah, for me, it definitely took having bad things happen over and over again. It wasn't just the medical stuff, it was also interpersonal stuff and past traumas all kind of converging. And um, it led to my autism diagnosis. And that is a very common story. That doesn't mean that everyone has to go through similar to get their diagnosis, but it just happens to be a very common story, one that a lot of people in our community can relate to. Okay, back to the research. Autistic females are finally getting recognized. The researchers also ran a separate analysis on what they call autistic disorder, which is the older, narrower diagnostic category that typically requires higher support needs. The ratio there is more stable across age. It's the broader autism spectrum that's seeing this catch-up effect, the population that matches more of what we hear about as the female autistic phenotype. Also, I just want to put a note there that there's a lot of males, those born male, who identify with what is called in the scientific medical literature, female autistic phenotype or female autism phenotype. What does this all mean? So let's zoom out a little bit. This study is by far the largest and most methodologically rigorous look at the autism male to female ratio that's ever been done, to my knowledge. It's not the first study to suggest the ratio is lower than four to one, but it's the first one with the size and the statistical machinery to really answer the question with confidence. And the answer is unambiguous. The ratio has been collapsing for years, and in Sweden, it's now functionally gone. There are a few implications I want to draw out because I think they matter for autistic adults specifically, especially anyone in the audience who got diagnosed late or who's still wondering whether to pursue a diagnosis. First, if you got missed as a kid, that wasn't on you. The four to one ratio that shaped your childhood evaluation, or your parents' evaluation of whether to bring you in, or your teacher's decision about whether to flag you to the school psychologist, that ratio was wrong, not slightly wrong, substantively wrong. The autism research field built its early infrastructure on a sample that was drawn primarily from boys. And the diagnostic tools and screening practices that flowed from that work systematically failed to identify autistic girls. You weren't too subtle. You weren't a special case. You were one of the enormous, mostly invisible cohort. And I've read research that indicates that that presentation of autism is potentially the most common one. Second, if you're currently considering whether to seek diagnosis as an adult, this paper is part of a growing pile of evidence that the system is finally, finally beginning to see you. The Swedish data shows clearly that adult diagnosis is way up. Female adolescent diagnosis is way up. The clinicians being trained today are being trained on different reference cases than the ones who trained 20 years ago. That doesn't mean the process is easy or accessible everywhere yet. It absolutely is not. And I get emails in my inbox every week showing me that that is the case, especially outside countries with universal health systems. But the science is moving in your direction. Third, there's an implication here for clinicians and for parents of younger children. If a girl is presenting with social difficulties, intense interests that could be normal, quote unquote normal, anxiety, sensory issues, or any of the other markers that we now associate with the female autism phenotype, the four to one prior ratio is doing damage. It's telling clinicians that autism is unlikely in this presentation. The Swedish data says, stop using that. Stop it. The base rate's something much closer to one-to-one across a lifespan. Adjust yourselves. Fourth, and this is the one I keep coming back to. This study reframes a question that has dominated autism research for decades. The question used to be, why is autism more common in males? The answer to that question turns out to have been in significant part because we weren't looking properly. The new question is, why does autism in girls and women get caught later? That's a much more useful question. It points us toward research on presentation, masking, diagnostic instruments, clinician training, and the social environments that delay recognition. It treats the ratio as a problem of measurement rather than a feature of biology, and that is a real shift. Now, what this study can't tell us, a few limitations here that are worth naming. First , as you've noticed, as is obvious, and this is in Sweden. Sweden has a national health system, mandatory developmental screening, as we've talked about, and universal access to specialist evaluation. The findings will not necessarily generalize to countries where autism diagnosis is gatekept by private insurance, geography, or wealth. In the United States, for example, the picture for autistic females remains substantially worse than in Sweden and surrounding countries, and the ratios you see in the United States data still skew higher. Some of that is real difference in how the diagnostic ecosystem functions. Second, the study restricted itself to people whose parents were both born in Sweden. The authors did this to keep the population homogeneous across 35 years, which is a defensible methodological choice. But it means we don't learn anything about how male to female ratio looks in immigrant or racialized communities within Sweden. But it means we don't learn anything about how the male to female ratio looks in immigrant or racialized communities within Sweden. Those are populations where autism is often even more underdiagnosed, and we know almost nothing about gender breakdowns in those groups. Third, the study uses sex assigned at birth. It does not break out trans, non-binary, or gender-diverse autistic people, which we know there are many, many. Yes, we have research telling us that the overlap between autism and gender diversity is incredibly substantial, but this paper isn't designed to speak to that. Fourth, as I mentioned earlier, no patients or community members were involved in shaping the study. That's a standard limitation in this type of epidemiological work, but it's just worth naming. There's a particular grief that comes with reading a study like this. If you're an autistic adult who got missed for decades, raises hand, the data confirms something you've probably suspected for a long time, and we've talked about this before. The system was looking past you, not maliciously on purpose. It's a long, accumulated failure of attention. Autistic females were never rare. The eyes that should have seen us were trained somewhere else. And there's also in this study something super hopeful. The fact that the ratio is collapsing means the eyes are turning, the attention is turning, slowly, unevenly, with a lot still left to do, but it's it's happening. The next generation of autistic kids are hopefully going to grow up in a diagnostic world that's much closer to recognizing the full range of how autism actually presents. Some of them will avoid the decades of misdiagnoses that a lot of us went through. And that is worth a lot. If you're newly diagnosed or identified or self-identified, or somewhere in the middle of figuring it out, the study is part of your evidence base. Maybe print it out, carry it with you, take it to your appointment if you need to. Heaven forbid you have to defend yourself to a mental health care provider, but just in case you do, it's a part of what you can also hand to a skeptical family member or read again when your own internalized doubt shows up and it will. But you know, the science is catching up to our experiences. If anything in this episode raised a question for you or made you want to share something from your own experience, either just with me or with the listeners, I am happy to play it on the episode, to read it on the episode. You let me know what you prefer. Email address is otherautism at gmail.com. That's all one word, otherautism at gmail.com. Also, oh my gosh, I would like to say a huge thank you to Joe, my first monthly coffee supplier. So, in addition to buying a one-time coffee, apparently the Buy Me a Coffee website also allows a person to become a monthly supporter. These aren't literal coffees. If you're not familiar with Buy Me a Coffee, these aren't literal coffees that you're buying me. I do love coffee, but the donation increments are roughly the price of a nice cup of coffee. So it starts at around $5, depending on where in the world you're located, and goes up from there. If you would also like to keep the show well caffeinated, please find the link to my Buy Me a Coffee page in the show notes. I keep whacking this microphone. My gosh. Sorry, baby. Well, that's all I have for you today. Thank you so much for being here. Until next time, bye.