Get access to over 1,300 articles and reader Q&As.
There is a lot of discussion of late — in the media, among parents, in policy circles — about autism. The CDC recently released data suggesting that autism diagnosis rates among children in 2022 had risen to 1 in 31 children. This has been steadily rising for years, from 1 in 149 in 2000. Similar increases can be seen over time in data based on medical records that cover children of different ages. And there are increases outside the U.S.
For some, this is a substantial cause for concern. Secretary of Health and Human Services RFK Jr. has termed it an “epidemic” and has promised to provide an explanation by September. He and others have blamed vaccines and environmental factors, and dismissed the impact of genetics and differences in how autism is diagnosed.
For others, including the CDC and organizations focused on autism awareness and advocacy, this trend represents progress. Improvements in diagnosis and more widespread screening mean identifying more children who can get the support they need to thrive. Expanding awareness can help people contextualize their own experiences, especially for those who are diagnosed at older ages. Many of these organizations argue that most or even all of the increase in diagnosis rates reflects a broadening in our understanding of the range of symptoms of autism and an expansion in who has access to care.
At the core, this comes down to a question for data. Of the increase in diagnosed autism over time, how much is due to changes in how we diagnose and changes in awareness, and how much is due to other factors within or beyond our control? We have mountains of clear evidence on some factors (vaccines do not cause autism, and genetics play a large role), and others are deserving of more research.
This question is important in part because it helps focus energy. No matter what the answer is, a focus on getting families the support they need should be a huge part of it.
But if most (or all) of the changes are due to diagnosis, that means focusing policy efforts on better access to screening. Right now, a lot of energy is sucked up in trying to figure out what could change in our medical system to lower autism rates, when, in fact, it might be that such a goal is actually counterproductive in terms of supporting families.
On the other hand, if much of this increase remains unexplained by diagnosis and awareness, it would necessitate a greater focus on what has changed in our environment (or otherwise) that has driven this increase. Let’s look at what the data tells us.
The role of genetics
First, I want to address the role of genetics.
Autistic traits have a strong genetic component. The best way to understand this is based on studies on twins. These twin studies look at the overlap in autism diagnosis among identical twins who are raised together — those who share both genes and environment. The researchers compare this to overlap in diagnosis for fraternal twins, who share an environment but not all their genes. Based on the relative amounts of overlap, these studies can evaluate how much of autism is due to genetic heritability.
A meta-analysis of studies like this shows a heritability factor between 65% and 90% — meaning genetic causes explain the majority of variation across the population, possibly almost all of it. There is not a single gene for autism, but there are many genes that contribute.
The strong genetic component of autism is important in predicting who is likely to be affected, but it does not really help us understand trends over time. Changes in the way diagnosis works can drive trends even if a disease is very heritable. Environmental factors could change how genes are expressed, so they could also be consistent with trends. In the end, recognizing the crucial role of genetics is not actually sufficient to figure out why autism has increased.
The impact of diagnosis and awareness
There are a number of changes over time that have meant that people might be diagnosed with autism now who would not have been diagnosed in the past. These include:
- The definition of autism has broadened over time. Diagnosis of autism is based on guidelines in The Diagnostic and Statistical Manual of Mental Disorders (DSM). Autism was first included in 1980, with a number of important changes in the decades since that reflect a wider range of symptoms, severities, and individual differences.
- Screenings have increased. In 2006 the American Academy of Pediatrics recommended that all children be screened for autism at 18 and 24 months, which likely identified children who would have gone without a diagnosis.
- There have been changes over time in school services that are available to children with a diagnosis of autism, which makes schools and parents more likely to seek a specific diagnosis.
- There is increased awareness of autism, which may cause more parents to seek out help if they suspect their child has this diagnosis.
All of these factors are likely to matter — the question is, how much? That question isn’t easy to answer. One way to address it is to look at specific events.
One paper, with data from Denmark, shows that changes in diagnosis criteria in 1994 and 1995 explain about 60% of the increase in diagnosis the authors observe between a cohort born in 1980 and one born in 1991. A paper with data from California showed that 25% of the increase between 1992 and 2005 is accounted for by a single change in which individuals previously coded as “mentally retarded” (the medical term that was used) were given an autism diagnosis.
Both of these findings, though, are in an older period and represent a response to specific changes. It is much harder to quantitatively capture the impact of changes in awareness and screening behavior. Good evidence that this plays some role can be found by looking across demographic groups. This paper looks at trends in diagnosis from 2011 to 2022. It finds that the increases in diagnosis are disproportionately in older people, in girls, and in Black and Hispanic children. These are groups that, historically, were less likely to be evaluated for autism, so this pattern suggests awareness and universal screening are driving some of these increases.
Another suggestive piece of evidence is the cross-state variation in the estimates provided by the CDC tracking. In the most recent data release, covering 8-year-olds in 2022, the prevalence estimates ranged across states — they were (to give one example) twice as high in California as in Maryland. There is no strong reason to think that actual prevalence would vary this much across states, so this suggests differences in diagnosis.
(Online I have seen claims that this is due to mandatory vaccination policies in California. However, there is no evidence that vaccines cause autism, and Maryland and California have almost identical vaccination rates.)
All this together indicates that the majority — possibly the large majority — of the change over time is due to changes in how we diagnose and in awareness. What is missing is a concrete answer to whether this is 60%, 80%, 95%; this feels like a very important research question, which I hope someone is working on.
If the answer is 60%, that leaves more of a role for other factors.
What other possible factors are there?
There are a number of other factors that are often mentioned as possible causes of increases in autism, some of which are more policy-relevant than others.
One is paternal age. On average, fathers over 50 are slightly more likely to have children with autism. It is difficult to know how much of this relationship is causal, as opposed to correlation. But even if this relationship is causal, the share of fathers in this age group has increased only a tiny bit over time, probably not enough to be an important part of the explanation.
A second explanation is prematurity. Babies born preterm, especially very preterm, are much more likely to have autism. Those born between 22 and 27 weeks are about six times as likely to be diagnosed as children born full-term. Over time, survival at these very early gestational ages has increased. Again, though, the share of babies here is just too small to be an important part of the overall trend.
With both of these explanations, even if they played a role in the trends in autism, there is not any clear change we would make to “fix” them.
A final explanation that comes up a lot is “toxins” or other environmental factors. Some of these claims have been debunked (like the link with Tylenol in pregnancy). In other cases, there is a link (like with lead exposure), but, since those exposures have decreased over time, they do not fit the trend. There are also a number of studies that look at urban air pollution and autism, but the evidence shows correlation, not causation. It is hard to rule out these explanations, since someone can always point to a change over time in some chemical and then say that must be the explanation. However, there is nothing concrete in the data that would point to a particular environmental factor driving these trends.
Closing thoughts
There are two things that feel clear here.
First, there are many parents with autistic kids who need continued support. An autism diagnosis can often really help families find the resources and advocate for the accommodations they need, and making those resources available through policy should be our top focus.
Second, we need more research on the importance of diagnosis and awareness. This is a case where the numbers matter, and they are not precise enough yet.
Beyond this, there is a need to reframe the debate away from a fear-based approach. A rising autism rate has been used by the current administration as a political tactic to promote a number of non-evidence-based policies. In response, others have pushed back with a narrative suggesting that the question of what explains trends is completely settled, which it is not. We need to reframe this debate to focus on understanding, and then to generate better measurement and evidence.
The bottom line
- Autism diagnosis rates among children have been steadily rising for years — in 2022 the rate had risen to 1 in 31 children.
- The increase in diagnoses has sparked debate over whether there is evidence of preventable causes or it’s the result of improvements in awareness and screening.
- Autistic traits have a strong genetic component. This is important in predicting who is likely to be affected, but it does not really help us understand trends over time.
- The majority of the change over time is due to changes in how we diagnose autism and in awareness. What is missing is a concrete answer to how significant of a role this plays, and whether or not it leaves room for other factors.
- An autism diagnosis can often help families find the resources they need, and making those resources available through policy should be our top focus.
Log in
As always, I appreciate your thorough look at and interpretation of the data. However, I think a fundamental piece of this conversation is missing. With all of this research and speculation about autism rates and causes, we are sending the message that having an autistic child is a bad thing. The fear mongering is perpetuated with his kind of rhetoric. As a parent, I understand the fear that coincides with the possibility of having a “disabled” child. But maybe it time we stop labeling autistic people as disordered, and started actually accepting that there are just different types of people in the world. We are not all the same, and we don’t have to do a bunch of research to figure out why. Can you imagine if we had various parenting/pregnancy experts telling us all about the data on the rise in numbers of LGBTQ people? Or speculating on the genetic and or environmental causes differing sexual orientations? As is some type of brain damage must have occurred prenatally to create such a person? I’m here to offer a different perspective. Maybe being autistic isnt the end of the world. Just a thought. https://open.substack.com/pub/divergentbydesign/p/neurodivergent-not-broken?r=cxnxk&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true
Emily, regarding Tylenol, it’s frustrating that the debunking focus is still on pregnancy use, when there are stronger unanswered concerns around its use in babies/young children (eg as set out here: (https://pmc.ncbi.nlm.nih.gov/articles/PMC5536672/). This research also questions whether the vaccine concern is actually acetaminophen given around vaccines to young babies. It would be so
helpful if you could you share your thoughts? Is there anything in it, or would you debunk it also?
Primary care pediatrician and healthcare justice organizer here.
Great conversation here, and quite civil too, gently educating and offering different opinions. That’s why I appreciate this community.
I have many autistic patients across a full range of abilities and family backgrounds. Universally, the kids benefit from robust support in schools, but even more so, the kids benefit from the family and caregivers having support and training.
As a couple folks have said, this administration, as well as the previous one in more subtle ways, have been looking for excuses to divert public funds away from the health and home supports that these kids and families so desperately need, and put the onus on struggling everyday Americans to buy Dr. Oz’s magic pills to cure their kids of who they are at their core. If you have friends, family, or community members with autistic kids in your life, now is the time to show up, in big ways or small. We have to believe in each other! No politician will no save us.
I’ve been debating writing this all day; it’s probably not my place. But here goes. I just wanted to acknowledge the parents and caregivers out there who are dealing with children and loved ones with autism and who are deeply struggling. I see so many comments on socials about how they finally feel seen and possibly even hopeful. RFK jr is likely a flawed source but the discussion and focus on autism likely can’t be a bad thing; if at least to help get the support and services you and your family need. My bro is disabled likely also autistic but undiagnosed. He gets support and we are so so so lucky for this. I hope you can hold on to hope that good will come of this and that you find some solace in this community as I often do. Take care.
Sorry to say but I don’t think this administration is going to be giving parents with autistic children or people with disabilities any additional support. Hopefully they don’t take away what little people currently get.
They already have. https://time.com/7279068/trump-administration-autism-research-cuts/
I’ve heard from kindergarten teacher friends that many parents are quite resistant to the suggestion that their child might be different and should be screened, so I’d be hesitant to take that as a general given without evidence that parents are pushing for diagnoses.
RFK is crazy, but there *IS* a huge increase in the actual amount of autism. https://www.ncsautism.org/blog//autism-explosion-2024
Arguing about whether there is more is not that helpful, though. Most research currently is on causes of autism. There should be a lot more on treatment and intervention.
Emily, you’re usually more data-driven– surprised you go mostly with the party line when it’s clear there are more autistic people in the world than there used to be.
I’m just going to quote from that article, nothing more needed, “While observations like these are hardly gold-standard evidence . . . With that in mind, let’s run an observational logic experiment “
I totally agree that the article is crappy and clickbaity. (original commenter)
This article is data-driven. Not only is the essential question is spelled out directly, it clearly acknowledges the fact that there is an increased diagnosis: “Of the increase in diagnosed autism over time, how much is due to changes in how we diagnose and changes in awareness, and how much is due to other factors within or beyond our control?”
I’d also like to point out that the headline of your included article says, “devastating increase in autism, a surge that began with births in the late 1980s and has continued unabated.“
autism was added to the DSM in 1980, and Emily pointed out in this article. So, yeah, I’d say that’s a really great argument for the “change in awareness” camp.
Also any article that uses the term “devastating” and “autism” in the same sentence should be considered suspect.
That is literally what she is talking about. Reread more carefully
She says “The majority of the change over time is due to changes in how we diagnose autism and in awareness,” but also that we have no idea if that accounts for 60% of the increase or some other number. What I’m saying is that there is an increase in the number of autistic people in the world, independent of the fact that it is now more commonly diagnosed in people who are autistic.
I really have a hard time believing this is true. I think a lot of people who would’ve just been considered awkward or eccentric in previous generations get diagnosed with autism nowadays. I always wonder if my brother would get diagnosed with it if he’d been screened with current criteria instead of whatever was being used in the early 1990s.
Karen, thanks for clarifying your point.
How can we possibly know that there is an underlying increase in the number or rate of autistic people, separate from changes in diagnosis? I’m not even sure how one could distinguish between the two, but if you can provide a source from the research literature, I’ll read it.
If the increase is largely due to diagnosis then that doesn’t mean there are more autistic people in the world, it means there are more with a diagnosis. More potentially receiving services. More spending “government money”
Also, are rates of profound autism and rates of high functioning autism, or whatever we prefer to call them, rising at the same rates? Is it thought that both are mainly caused by the same set of genes?
This is THE important question and I would love to know if there is more data. I am a special service provider in a large urban school district and when we talk about “autism” it is so highly variable I am continually shocked we even use it as a diagnose to describe a child. On one hand, we have children, often many in one family (genetics) from all ends of the SES spectrum with severe needs- aggressive behaviors, non-verbal, serious sensory processing impacts, impacted daily living skills. On the other hand we have a school of high achieving white parents and a student with autism can look like this- highly verbal and very high IQ score with sensitivities to chaotic environments, lack of interest in friends and difficulty sharing, rigid and inflexible in thinking patterns and extreme focus on areas of interest – which can sometimes be highly intellectual subjects. These are two different types of kids with VERY different trajectories. I don’t know what the data says but my anecdotal experience is that the increase is multifactorial. For some situations, an autism diagnosis is gentler than a serious emotional disability diagnosis and parents are more open to it so instead of a conduct disorder label which may be more appropriate a student is labeled autistic. For others, like mentioned in the article, many kids labeled MR or now ID are probably being labeled autistic because it has a more positive connotation than those labels. And for others, a student isn’t textbook in every way so a parent wants a way to describe them and get them additional accommodations or services.
To sum up- a neurodivergent high functioning student labeled with autism is in no way similar to a child with severe autistic behaviors and lumping them all together under one diagnosis for research and policymaking purposes is our first and largest mistake.
While I understand what you’re saying, this is quite the ableist statement. Functioning labels are not exactly welcomed in the autistic community. Let’s do better. https://autisticadvocacy.org/2021/12/functioning-labels-harm-autistic-people/
Currently there is no one gene that is related to Autism, but about 100 different genes, in many combinations. I can’t find the study now, but I read recently that the growth in diagnosis is mostly lower-needs (what used to be referred to as High Functioning). Hope that helps.
Really interesting, thanks. Can you say more about what is going on in other countries? How much are their rates going up, compared to in the US?
I suspect that a lot of the change is due to parents being more likely to push for an autism diagnosis. Partly because parents are much more involved than a few decades ago, partly because there is less of a stigma around disabilities, and partly because such a diagnosis qualifies parents for additional public education services. I was recently at a five year old’s birthday party and there were at least four kids with autism there, but all were basically “shy and good at math” type autism, not “can’t feed themselves” autism. My kids were behind with speech, and many people told us we should have gotten them officially diagnosed with a speech impediment in order to qualify for free preschool in my state. (Unfortunately for my pocketbook, this advice came after it would have been useful so we had to pay for preschool, but otherwise we would have). I suspect it’s a similar story with autism.
I have some firsthand experience with the world of severe intellectual disabilities, and that is a very different world from the “neurodivergent” type of autism to the point where they are essentially different things.
With rates of profound autism going up, it certainly seems like it’s not just a matter of broader diagnosis, although maybe even profound autism would have been categorized as some other impairment in the past.
https://rutgershealth.org/news/first-large-study-profound-autism-finds-rising-problem-disparate-impacts
It’s too bad RFK is driving this discourse, making it polarizing. Vaccines clearly don’t cause autism but if profound autism can be prevented by environmental changes that would obviously be a good thing.
Great article! Could I suggest in the future to switch to person first language? For example, instead of “autistic kid,” you could say “a kid with autism.” It tends to dehumanize a person and is not used in other context as such as cancer. You would not write a cancerous kid, rather a kid with cancer.
I’ve actually heard the opposite from parents of autistic kids. They said autism is a part of someone’s identity and what makes them who they are, rather than something they have and could/want to fix. Some parents think “autistic kid” allows them to feel proud of their identity.
That said, your point is a good one and I’m sure many agree with you. So just throwing this out there as food for thought — not to invalidate your opinion!
Hi, totally agree. There is actually a huge discourse in autistic circles about this. My wife and I (high functioning autistics) have two kids (3 months and 2.5 years) that we are like 80% sure will also be high functioning autistics. We are highly on the side that autistic is a useful adjective. Part of the difference in the cancer example above is that cancer is a disease to be cured, while we do not wish to “cure” ourselves or our children’s autism. We see our autism as making us capable of generating thoughts and ideas that neurotypical aren’t allowing us to be creative and capable in a modern workplace.
I second this that the push for person first language was actually a push from parents and not from people with I/DD themselves.
People choose if they prefer person first or not. As well some don’t want to be identified with a disability at all.
I will say to your point I’ve also heard the preference instead of “high functioning” to say “lower support needs” or “higher support needs”
Nope. This isn’t true for everyone, as nothing is true for everyone, but most autistic people prefer to be called autistic people, not people with autism.
As a parent of a kid who fought leukemia from 1-4 years old and monthly oncology appts still- many oncology parents actually say “cancer kid.” That said, I can’t speak for autism.
Autistic person here. Identity first language is generally considered the preference. It’s not dehumanizing. Autism is not a disease to be cured, like cancer– its a way of being and is a core part of my personal identity . I am not a person with the disease of Autism. I am an autistic person, just someone who is gay is a gay person, not a person with gayness.
https://autisticadvocacy.org/about-asan/identity-first-language/
There doesn’t seem to be much data backing up the “majority of increase is disgnosis.” Would like to see a more quantitative treatment before making a statement like that. I’m predisposed to believe the statement so I’m extra cautious of confirmation bias. I don’t think the studies cited have enough compelling, extrapolatable information to make the claim.
As a parent of an autistic child. The awareness and screening helped us get our son diagnosed early and start early intervention. I really believe even 10-20years ago he would’ve flown under the radar until kindergarten or older and would’ve had such a harder time in school and daily life.