What RFK Jr. Gets Right—and Wrong—About Autism

The April 16 autism press conference featuring Health Secretary Robert F. Kennedy Jr. had the media and internet in a frenzy. Did he offend people with autism, or was he at long last shining a light on harsh, neglected realities? Was he right to call the stunning increase in autism cases an epidemic, or was that unscientific scaremongering?

Some people, including Sen. Elizabeth Warren, raged. Other people, particularly parents of the severely affected, cheered. But mostly, people were confused. And who can blame them? The truth about autism can seem as easy to grasp as a mound of warm Jell-O.

So it’s worth some quality time to closely examine the main bones of contention. After all, a hell of a lot is at stake. Probably the future of our country is at stake. Because if the new CDC autism prevalence estimate of 1 in 31 8-year-olds signals truly escalating rates, the implications for public health, research, safety-net services, and our economy are simply massive.

As someone who has labored in the trenches of autism for two decades, in the realms of both research and advocacy, I offer this little FAQ for your consideration.

Was RFK Jr. right to call autism an “epidemic”?

Score a point for the health secretary here. A half century of troves of very rich data from multiple sources across multiple systems and examining all levels of autism emphatically point to a true increase in autism rates. Despite rampant speculation about “just noticing it better,” the research does not support this conjecture, though it could explain some effects around the higher-functioning edge. Some key data points to consider:

• Even limited to cases of autism with intellectual disability, or ID (formerly called mental retardation, and meaning IQ of 70 or under), childhood autism rates have skyrocketed, from about 0.01 percent to 0.05 percent of children in the 1960s based on early studies to about 1.28 percent of 8-year-olds in 2022, per the latest CDC data.
• Even in 4-year-olds, who tend to have greater impairments than those diagnosed later in childhood or adulthood, prevalence has more than doubled over a short period, reaching 2.93 percent for the recent 2022 surveillance, up from 1.34 percent in 2010, when the CDC began tracking preschoolers.
• Autism cases in our most populous state, California, have exploded. Its developmental disability system—which serves only significantly disabling forms of autism—has seen autism cases rise from about 4,000 in 1989 to about 206,000 in 2024. That’s a 51-fold increase over 35 years, while caseloads of ID and other categories remained fairly flat.
• Careful epidemiology in the state of New Jersey conducted by the respected Rutgers University researcher Dr. Walter Zahorodny (who gave invited remarks at the RFK Jr. press conference confirming the existence of a true increase) has reflected sweeping increases in autism prevalence in the state, to the point that in some regions, about 7 percent of 8-year-olds have autism. Dr. Zahorodny told me the case definitions did not change over time.
• Autism cases are up sharply in special education across all states. To be eligible for special education, a student must demonstrate significant impairments that interfere with learning, and they require specialized instruction as a consequence. For example, in California, cases grew from about 4,000 in 2000-21 to about 167,000 in 2023-34, while cases of ID remained steady. In 2019, Massachusetts reported that its schools were serving four times as many students with autism as they did 15 years prior. In Minnesota, autism cases surged exponentially beginning in the 1991 school year, with the researchers finding “diagnostic substitution does not largely explain the increasing trends.”
• The largest U.S. study of autism prevalence in hospital and medical care systems shows rising autism prevalence across birth cohorts. Autism in adults over the age of 65 was shown to be extremely rare, in the range of .02 percent, compared to rates of about 3 percent for children aged 5 to 8 in 2022. When RFK Jr. says he’s never met anyone his age with significant autism, he’s not wrong. The data, especially from the robust California developmental disability system, show they hardly exist.
• Autism rates are skyrocketing in our adult Medicaid and Social Security safety-net systems, where the disorder is by definition limited to more severe cases. In Medicaid, the prevalence of young adults with autism more than doubled from 2011 to 2019. For SSI, we see staggering autism growth of 336.6 percent from 2005 to 2019, while cases with ID and other mental disorders decreased by 45 percent.
• Further evidence it’s not a game of diagnostic switcheroo: The CDC has found that the prevalence of children with developmental disabilities generally, including ID, has increased over time.

The word epidemic is appropriate to describe the ascension of autism, Alexander MacInnis, MS, an independent epidemiological researcher who has published on California autism data and who has a daughter with profound autism, told me. “Epidemic has a definition, and not just for infectious disease,” he said. “It can be a disorder where more cases are occurring than what you would expect based on history. We have massively increased birth cohort prevalence from every data source I can find showing very consistent increases, even within studies, which removes bias. Overall we see about a 7 percent increase in autism cases per birth year. Does this meet the definition of an epidemic? It does.”

But Scientific American just told me there’s no true increase in autism.

The media loves to cherry-pick a few old studies to defend the idea that there’s been no true increase in autism. A perfect example is this recent gem from Scientific American citing and mischaracterizing papers that are plainly irrelevant.

For example, it points to a 2015 study on special education enrollment for the idea that autism’s increase was only caused by diagnostic shifts away from ID to Autism Spectrum Disorder (ASD). That paper was not an epidemiological study on autism increase over time but instead intended to look for hints of common molecular origin for comorbid conditions. It was conducted by geneticists, not epidemiologists using proper tools of that trade, and performed a clumping of all ages 3 to 21, which is meaningless to determine autism rates over time. In any event, the study found that a large percentage of the autism increase was not accounted for by declining ID, and many states did not reflect this trend at all, including the most populous states, California and Texas. Furthermore, the data is now 15 years old, and autism rates have clearly continued surging over that time.

Moreover, the federal Office of Special Education Programs (OSEP) shows a dramatic increase in autism in special education, to 909,055 in 2022-23, up from 513,688 in 2014-15. Meanwhile, OSEP data shows ID cases have flatlined, not decreased, 420,000 in 2022-23 from 415,335 in 2014-15. But none of this stops journalists from citing this old, shoddy study.

The other oft-cherry-picked paper cited in the SciAm piece is a 2015 study finding an increase in autism in Denmark in the very earliest years of the surge, looking at children born from 1980 to 1991, when autism rates were still vanishingly low. The study found that a change in diagnostic criteria in 1994 and the inclusion of outpatient diagnoses in 1995 specific to Denmark accounted for perhaps 40 percent of the increase over that specific period. There’s no indication that more recent, and vastly higher, rates in Denmark, including a finding of 2.8 percent in a 2018 study, are owing to diagnostic shifts.

But isn’t the increase in the United States due to the broadening of diagnostic concepts?

Here we can say a hard no. RFK Jr. is right to reject that rationalization. This is often speculated, even taken as dogma, but not backed up by evidence. The CDC network, for example, actually narrowed its ascertainment after 2016. In the study cycles between 2000 and 2016, the researchers used an exhaustive, expensive, and time-intensive “active case finding” method, which identified not just diagnosed cases but also children without a formal diagnosis. In New Jersey and elsewhere, “it yielded approximately 20 to 25 percent additional cases,” Dr. Zahorodny told me.

But for the 2018 ascertainment cycle through the 2022 cycle announced this month, to save on time and expense, the CDC switched to relying solely on preexisting diagnoses, whether through medical or school systems. Zahorodny had predicted that autism rates would fall because of the more restrictive methods. But that didn’t happen.

He also criticized the commonly invoked idea that the switch from DSM-4 (referring to the Diagnostic and Statistical Manual used by psychiatrists to diagnose mental disorders) to DSM-5 criteria in 2013 opened the floodgates to milder cases in the CDC system. He said that as applied in the CDC cycles, “the DSM-5 criteria are more restrictive in DSM-5 than DSM-4” and that a CDC study found no significant differences in ascertainment between the two diagnostic schemes.

“ASD is a serious neurodevelopmental disorder that has increased without interruption across all communities and subgroups,” Zahorodny told me. And he says we should brace ourselves for numbers to go up: “The 3.22 percent number is, if anything, likely an underestimate.” Citing emerging data from New Jersey and other sources, he says that soon autism’s “new normal will be 5 percent.”

Do the research questions RFK Jr. mentioned make any sense?

This was RFK Jr.’s major fail. For the most part, no. While he’s right to say “external factors” should be a focus, there’s no reason to believe that “environmental toxins” are playing much of a role—not mold, food additives, pesticides, air, water, or ultrasounds. He did mention “medicines” as well, and perhaps that was a nod to the discredited vaccine hypothesis, or perhaps a reference to maternal medications like the anticonvulsant valproic acid that have been validated to increase autism risk.

But the idea of “toxins” being a meaningful driver of risk is not backed by two decades of environmental epidemiology, nor does it make sense biologically. Neuroscience research has found that autism, at least for the most part, begins with mis-wiring in the fetal brain, resulting in microstructural defects, particularly in the cerebral cortex. The defective circuitry leads to distinctive impairments in learning and behavior. There’s no known biological mechanism to link low-level environmental toxicants such as those mentioned by Kennedy to these neurodevelopmental pathologies.

That Kennedy hired David Geier, a disgraced clinician, to help lead autism research at the National Institutes of Health should send chills down our collective spines. It shows that despite having some wisdom about autism’s increase, he still clings to conspiracy theories, and their peddlers.

But autism is genetic, right?

Research is clear that autism is a strongly heritable condition, a phenomenon detected in both twin and sibling studies. Identical twins can have a 90 percent concordance rate of autism, and younger-born children of an older autistic sibling have up to a 20 percent risk for having autism themselves.

However, in contrast with the commonly heard “Autism is genetic!” rally cry, there’s only weak evidence that autism’s strong heritability is genetic in origin. I know that sounds strange. But for example, a huge study of more than 21,000 individuals with autism yielded genetic findings in only 8.6 percent of cases. That said, genetic findings can be higher—for example, 16 percent to 26 percent—in small samples with patients with severe presentations and physical anomalies.

So what can explain the more than 74 percent of autism cases that are “idiopathic,” or without a known cause? Here, hypotheses must be rooted in solid science and should steer close to the puzzle of the missing heritability not explained by genetics. For my money, a fruitful path forward is contemplating genetic and epigenetic (meaning non-DNA molecules that help control gene expression) glitches—which may be induced by certain toxicant exposures—in the parents’ germ cells. By germ cells, to be blunt, I’m talking about our sperm and eggs and all their precursor cells, way back to when we were little embryos.

The “germline” hypothesis is too complicated for elaboration here, but those who may be interested might watch my recent presentation at Harvard or read a paper I co-authored on the topic with leading scientists. But other ideas around maternal drugs and medications, perinatal risks, maternal autoimmunity, and assisted reproduction techniques also deserve greater scrutiny. In short, when it comes to autism causation, there are still plenty of stones left to turn over.

What about regression in autism? Doesn’t that point to post-natal events like vaccination?

RFK Jr. implied that children start out developing normally, and then we are “doing something” to them to cause a regression and then autism. Regression in autism is certainly real in many cases, but whether it’s caused by contemporaneous exposures is altogether another question. While my son Jonathan, now 26, seemed to be severely autistic from the day he was born; my daughter, Sophie, now almost 19, developed limited, albeit nonfunctional, language before she lost her words completely.

Brain development unfolds over time based on a complicated spatio-temporal program. The brain can have deeply embedded programming errors that do not manifest immediately but instead become apparent later in development as different parts of the brain wire up, shed excess connections, and even shift electrical balances. It is likely this cascade of developmental events that precipitates regression around 18 months, and not any particular exposure.

If vaccination caused brain damage, then we would see these effects in 2-, 3-, 4-, and 5-year-olds and even older children receiving shots, but we don’t. Autism is generally apparent by 18 months, a time that coincides with some childhood vaccinations.

Now, to be fair, can there be cases of adverse vaccine reactions here and there? Possibly—every pharmaceutical product on the planet carries some risks. But autism’s scientific breadcrumbs do not lead us anywhere close to postnatal exposures like vaccines. In fact, childhood vaccines protect children from infectious diseases that can cause brain damage.

Does RFK Jr.’s blunt rhetoric about autism brush up against “eugenics”?

Advocates online went to town claiming RFK Jr.’s rhetoric used eugenicist logic for talking about autism as a life-limiting disorder imposing financial burdens and also for insisting on research to find risk factors.

Yes, Kennedy did overgeneralize when saying kids with autism will “never play baseball. They’ll never write a poem. They’ll never go out on a date. Many of them will never use a toilet unassisted.” But while his words understandably ruffled feathers, they reflected at least partial reality for a great many, including my own two adult children with nonverbal profound autism. While autism does indeed “destroy families” in many cases, it certainly does not for others. The secretary attempted to clarify his remarks on Fox News after receiving backlash.

His comments on the financial burdens of autism, however, are unfortunately correct and supported in every study examining the question. For example, a 2015 study from UC Davis found autism’s economic burden could reach $1 trillion per year by 2025. Group homes for those with severe autism can easily exceed $400,000 per year, just to serve one client. Talking honestly about the financial costs of autism, or any disorder or disease for that matter, may be painful, but it is hardly eugenicist in orientation.

Likewise, insisting on research to find risk factors can hardly be considered a eugenicist mission. Is it eugenicist to point out that alcohol can cause fetal alcohol syndrome? Is it eugenicist to give pregnant women vitamins to reduce the risk the fetus will suffer spina bifida? No, identifying factors that raise the risk for neurodevelopmental disorders in children can only be seen as a moral imperative, quite the opposite of eugenics. Many say they don’t trust RFK Jr., and I get it. He’s hardly making these proclamations accompanied by a “So we must greatly increase options and care capacity for every American disabled by autism.”

On the flip side of those whose feelings were hurt by Kennedy’s “hate speech,” many parents of the severely autistic were thrilled to see their everyday realities reflected in a talk by a Health and Human Services (HHS) secretary. “Our kids that are profoundly affected can’t advocate for themselves, and they are the ones in most need,” said one. “We want answers, prevention, and effective treatments. My son has cost millions of dollars.”

And another: “I feel so frustrated by the ASD = awesome narrative when those voices are not dealing with adult diapers, seizures, countless medical complications, etc, etc. Those with severe ASD aren’t on X and their lives are overshadowed by those claiming it is some superpower … I’ve never had a two-way conversation with my kid.”

Amy Lutz, parent of a profoundly autistic son, responded to Sen. Warren’s claim that RFK Jr.’s words were “disgusting” with this: “Truly stunned to see so many responses like this to RFK’s description of profound autism, which affects 27 percent of autistic people, including my son. What’s disgusting is sanitizing this devastatingly impaired population out of public discourse.”

Or as I posted on X:

In the end, we should be far less worried about Kennedy’s inartful words than about the grim fact that autism has ballooned into a full-blown public health emergency. We need unwavering national leadership, whether it be Kennedy or others, to unleash efforts to finally find answers. But spotlighting the true surge in autism should also come with the responsibility to insist on appropriate funding for the inevitable lifespan support needs. Let’s not forget that Kennedy’s HHS also governs Medicaid.