Este artículo estará disponible en español en El Tiempo Latino.
In announcing new autism prevalence data from the Centers for Disease Control and Prevention, Health and Human Services Secretary Robert F. Kennedy Jr. distorted scientific research to argue that there is an “epidemic” of autism that must be due to an “environmental toxin.”
The main finding of the new CDC publication — that among 8-year-olds at the selected study sites, 1 in 31 had autism in 2022, up from 1 in 36 in 2020 — represents the latest increase in estimated prevalence of the neurodevelopmental condition, also referred to as autism spectrum disorder, or ASD.
However, researchers have long warned that the rising numbers do not indicate a commensurate increase in the true prevalence of autism. While there may be some true rise, researchers say that broadening diagnostic criteria, increased awareness, the gradual introduction of universal autism screening by pediatricians and the growing availability of services help explain rising reported rates of the condition.
The CDC report itself lists such factors. But Kennedy said these attempts at rigorous interpretation of the data amounted to an “ideology of epidemic denial.” Among his claims:
- Kennedy mischaracterized early studies of autism prevalence and misleadingly compared them with the CDC’s latest study. For example, he compared modern numbers with those from a 1970 Wisconsin study that used an old definition of autism, and in a later interview, wrongly claimed that “all the kids in Wisconsin were tested.”
- He claimed that at most 25% of the increase in prevalence “can be attributed to better recognition and better diagnosis,” which means “85% are still — or 75% to 80% — still are part of an epidemic.” That’s not what the overall literature shows.
- Kennedy referred to a recent trend of increasing intellectual disability in the CDC survey data to argue that the increased prevalence is “real.” But a CDC author said that the pattern could be due to a change in methodology in 2018.
In the April 16 press conference, Kennedy, who has a history of pushing the debunked idea that vaccines cause autism, expressed false confidence that it is possible to identify an environmental cause of autism and eliminate it. “This is coming from an environmental toxin,” he said, after earlier referring to autism as a “preventable disease.”
Exposure to certain substances — particularly in the womb — may contribute to autism. Many studies have identified correlations to various exposures, but the role of these exposures in autism has been challenging to confirm. Two well-established risk factors include increased parental age and very preterm birth, both of which may have contributed to a small increase in autism over time as babies have been increasingly born to older parents and medical advances have allowed more premature infants to survive. (Babies born to mothers who took the antiseizure drug valproate during pregnancy also appear at higher risk of developing autism, although this would account only for a small number of cases.) What is very clear, however, is that much of autism is genetic.
“I disagree that the increase in prevalence rates must be due to an environmental toxin,” psychologist Helen Tager-Flusberg, director of the Center for Autism Research Excellence at Boston University, told us in an email. “Demographic and other sociological factors (as described in the CDC report itself) along with changes in diagnostic criteria and substitution are major factors. But there’s lots we still don’t know.”
Kennedy spoke alongside a co-author of the CDC report who said he did think the true increase in autism has been significant. “It is a true increase,” Walter Zahorodny, a professor in the department of pediatrics at Rutgers Health New Jersey Medical School, said. “There is better awareness of autism, but better awareness of autism cannot be driving a disability like autism to increase by 300% in 20 years.”
A few other scientists we spoke with agreed that there had been an increase in autism, although not as large as Kennedy claimed, and not necessarily for the reasons he suggested. Others were more skeptical of a true rise.
“It’s not impossible at all, that just these factors added all together might drive the increase entirely, without the need to invoke any other kinds of causal factors or an epidemic due to an environmental toxin,” Dr. Eric Fombonne, a professor emeritus of psychiatry at Oregon Health & Science University, told us, speaking of the factors that affect who is counted as having autism. “Do we know that for sure? No, but has anyone demonstrated that the rise could not be entirely due to that? No.”
Kennedy Misrepresents Decades-Old Autism Studies
In the press conference, as evidence of an “epidemic” of autism, Kennedy mischaracterized early studies of autism prevalence, incorrectly casting them as “exhaustive” and accurate measures of autism that can be compared with today’s figures.
“The baseline for autism in this country was established with the largest epidemiological study in history, a study of all 900,000 children in the state of Wisconsin, children under the age of 12,” he said, referring to a study published in 1970. “They found 0.7 children had autism in every 10,000. That’s less than 1 in 10,000. Today we’re 1 in 31.”
Later, in an April 22 Fox News interview, he incorrectly claimed that “all the kids in Wisconsin were tested” in the study.
No one did autism evaluations on all the children in the state. Instead, researchers reviewed records from between 1962 and 1967 from various clinical settings to count the number of children who had been given a no-longer-used diagnosis called early infantile autism, then divided the figure by the population to get a rate.
At the time, very few people knew about or diagnosed autism, David Mandell, a psychiatric epidemiologist, health services researcher and director of the Center for Mental Health at the University of Pennsylvania, told us via email.
For that reason, Fombonne said it was “completely absurd” to compare the study’s estimate to current figures. He added that even at the time, better studies arrived at higher estimates. Certain groups “use this study often because it gives them a very low starting point and accentuates the trend,” he said.
Kennedy also cited a 1987 study of children in North Dakota, falsely claiming that researchers “even conducted in-person assessments of the entire population of 180,000 children under 18.”
Researchers did not interview 180,000 children. Rather, they collected records of children who had been identified as having autistic symptoms in the state and did in-person assessments of around 200 children.
Study co-author Larry Burd, a professor of pediatrics at the University of North Dakota who now primarily studies fetal alcohol syndrome, told us that he is “pretty satisfied” that the study “did capture the prevalence rate of autism at that time” but that “many of the cases today would not have been captured using those criteria.”
He said he thought that there has been a true increase in autism, with broader criteria, screening and earlier diagnosis accounting for a “substantial amount of the difference but not nearly all of it.”
Kennedy also referred to a third study, called the National Collaborative Perinatal Project, which followed pregnancies at 14 medical centers beginning in 1959. As of 1965, researchers identified 14 children with infantile autism. “This was no half-baked survey based analysis,” he said, claiming that autism “would have stood out like a neon sign.”
But again, researchers emphasized that at the time of the study, very few people were diagnosing autism, and the definition then was very different.
Today, “we’d have diagnosed a far greater portion of the population than were identified in these earlier [epidemiological] studies because of fundamental changes in the way we conceptualize autism,” Tager-Flusberg said.
Dr. Ezra Susser, an epidemiologist at Columbia University, told us he counts himself among those who believe there has been “some increase” in autism. But “the figures that [Kennedy’s] using to illustrate the scale of the increase,” he said, are “just completely misleading.”
Missing Context on CDC Data
Kennedy called the latest CDC results “shocking” and presented them as showing that “autism is increasing in prevalence at an alarming rate.”
“This is part of an unrelenting upward trend,” he said, adding that “autism has increased by a factor of 4.8” since the first CDC report in 2000. But as the agency itself notes in its latest report, the observed prevalence rates are not directly comparable over time, nor are they representative of the autism rate in the entire country.
Although the CDC’s autism prevalence survey was designed to stay mostly the same over time, there have been changes over the years in how the agency identifies children with the condition. The figures also represent just a few places around the country — in 2022, for example, the data comes from 16 sites, up from 11 two years prior.
The current method for identifying a case of autism involves reviewing school and medical records for autism diagnostic codes or other indicators of autism. The CDC researchers “never see a kid in person,” Mandell said, adding that he thinks the CDC survey overestimates the number of children with the condition.
Catherine Lord, a clinical psychologist and autism researcher at the David Geffen School of Medicine at UCLA, said that there could have been changes over time in how readily physicians or teachers identified autism. When the CDC study began, she said, “almost no chart talked about social communication,” she said, “but today physicians and teachers are much more aware of a child who rarely makes eye contact or doesn’t seek out other children.”
Maureen Durkin, a professor of population health sciences and pediatrics at the University of Wisconsin-Madison School of Medicine and Public Health, who co-authored the most recent CDC report on autism prevalence, said that various changes in the definition of autism, awareness, support and screening, as well as changes in diagnostic practices have contributed to the increase in prevalence in the CDC reports.
Formerly, “children with other known causes of developmental disability such as Down syndrome were not evaluated for autism but today virtually all children referred for comprehensive developmental assessments are evaluated for autism and many children with autism have multiple co-occurring conditions,” she said.
Durkin also said that very preterm births and older parental age could have contributed to increased autism, as well as potentially “other factors yet to be discovered.”
Multiple researchers pointed out the variation in autism rates between study sites as a sign the CDC survey is capturing something other than true changes in autism. “Autism doesn’t respect geopolitical borders,” Mandell said. “We are seeing the results of better identification and services in different places.”
In the 2022 data, around 1 in 19 children were recorded as having autism at a site in the San Diego area, while just 1 in 103 children were recorded as having autism at the Laredo, Texas, site. In other words, the difference between Laredo and San Diego today is slightly greater than the difference Kennedy touts between children born in 2014 and those born in 1992.
“Research has not demonstrated that living in certain communities puts children at greater risk for developing ASD,” the authors of the new CDC study wrote in their paper.
The Laredo site had the oldest average age of diagnosis and the highest rate of cognitive impairment in the children identified as autistic, Craig Newschaffer, a professor of biobehavioral health at Penn State University, told us via email.
“One hypothesis could be that there is some absence of an environmental toxin in Laredo that is causally linked to autism and especially to early emerging cases of autism,” Newschaffer said. “However, it is much more likely that this descriptive picture supports the alternative hypothesis that there is a lowered diagnostic tendency in Laredo (kids are diagnosed at later ages, and higher-functioning children are missed), perhaps because this is an area where access to strong diagnostic services may be lacking.”
One factor that could drive the high rate of identified cases in the San Diego area, the authors of the CDC report wrote, is an initiative in which “hundreds of local pediatricians have been trained to screen and refer children for assessment as early as possible.”
Inaccurate Summary of Existing Studies
During the press conference for the CDC findings, Kennedy selectively and inaccurately described the scientific literature, presenting it as clearly showing that the steep rise in observed autism rates is real — something it does not do.
Kennedy claimed that after being tasked by the California state Legislature to investigate the issue in 2009, Irva Hertz-Picciotto, a University of California, Davis scientist, “came back with a definitive answer: the epidemic is real. Only a very, very small portion of it can be charged to better recognition or better diagnostic criteria.” He went on to say that “many, many” other studies “affirm” this and “the answer is very clear.”
Later, after being pressed by a reporter, Kennedy acknowledged that a larger fraction of the increase might not be real, but maintained that the overwhelming majority was.
“There are small slivers of the autism epidemic, maybe 10% to 25%, according to the studies — highest studies are around 25% — that can be attributed to better recognition and better diagnosis,” he said. “That means 85% are still — or 75% to 80% — still are part of an epidemic, and that’s too many.”
But that’s not what the overall literature shows. It’s true that a 2002 report, conducted by UC Davis researchers — but not by the researcher Kennedy named — found “no evidence” that widened diagnostic criteria contributed to California’s rising autism rate and concluded that “some, if not all, of the observed increase represents a true increase.” It was presented in the popular press at the time as being conclusive. The report, however, was not peer-reviewed and has been criticized by other scientists as having “unwarranted” conclusions.
The researcher Kennedy named, Hertz-Picciotto, specializes in identifying possible environmental causes of autism and did publish a study in 2009. It concluded that about one-third of the increase in autism reported in California was due to diagnostic criteria changes, including an earlier age at diagnosis and inclusion of milder cases.
That’s already more than Kennedy claimed. But the paper also noted that it had not included other potential contributing factors, such as increased awareness and desire to access autism services. “Other artifacts have yet to be quantified, and as a result, the extent to which the continued rise represents a true increase in the occurrence of autism remains unclear,” the study said. (We reached out to Hertz-Picciotto but did not receive a reply.)
When asked for support for Kennedy’s claims, in addition to the Hertz-Picciotto study, HHS pointed us to two papers authored or co-authored by Cynthia Nevison, an environmental research associate at the University of Colorado Boulder who has published about autism, including as a volunteer for SafeMinds, an anti-vaccine group. She has previously espoused anti-vaccine views and written articles for Children’s Health Defense, the anti-vaccine nonprofit Kennedy founded and previously led. (In his remarks, Kennedy appeared to reference another paper Nevison co-authored when he claimed a study found that the cost of treating autism by 2035 “will be $1 trillion a year.” The paper was retracted nearly two years ago.)
One of her analyses, published in 2014 in Environmental Health, concluded its data “suggests that ~75-80% of the tracked increase in autism since 1988 is due to an actual increase in the disorder rather than to changing diagnostic criteria.”
Experts, however, told us the analysis could not make such a claim, as information does not exist about the degree to which greater awareness, screening and diagnostic services may have driven autism increases, for example. Among other issues, Fombonne said, the analysis relied on a database that only includes children referred to services, and therefore is confounded and biased. “You cannot be drawing inferences by studying just that referred population,” he said.
A few other studies have attempted to quantify the impact of certain changes, generally finding between one-quarter to one–third of the increase in autism prevalence is attributable to a single factor, such as diagnostic substitution or a change in diagnostic criteria. But no study has tried to account for all factors — and therefore, Kennedy is wrong to say that research has shown 75% or more of the increase remains part of an autism “epidemic.”
Experts were generally skeptical of these efforts, with Durkin calling it a “dubious exercise.” But many other studies have documented or implicated various diagnostic changes in the observed rise in autism, although they did not quantify these effects.
There “is a very large body of informal descriptive evidence continuously reemphasizing the role of diagnostic tendency as a major driver of autism prevalence trends,” Newschaffer told us, referring to factors that determine who gets diagnosed.
In a 2012 study, which Fombonne co-authored, researchers reexamined records from a 1980s autism prevalence study in Utah, finding that 59% of the individuals originally designated as not having autism did have autism according to the more modern definition. The result was not primarily due to recognizing mild cases, as the average IQ of the newly identified cases was significantly lower than the originally identified group.
A 2013 study in California found that children who moved to a neighborhood with more diagnostic resources were more likely to then receive an autism diagnosis, suggesting social drivers of diagnosis were at play.
Taking another approach to address the question, a 2020 JAMA Psychiatry paper used twin cohorts in Sweden and found the relative importance of environmental factors associated with autism did not change over time, as might be expected if such exposures were indeed driving a surge in the condition. “These results thus do not suggest that environmental factors explain the increasing prevalence of ASD,” the study concluded.
Claim of ‘Alarming Escalation’ in Severity
As supposed evidence that artifacts cannot explain the observed increase in autism prevalence, Kennedy referred to a recent trend in the CDC data that shows a rise in the proportion of autistic children in the survey who have lower IQs. After gradually decreasing to 31% in 2014 from a high of roughly 50% in 2000, the proportion of kids with intellectual disability has ticked back upward, to nearly 40% in 2022.
“If you look at table 3 of the ADDM report” — the table with IQ information — “it’s clear that the rates are real,” Kennedy said during the press conference. The HHS press release on the CDC report also noted the recent trend, calling it “an alarming escalation in case severity” and stating that it demonstrates the increase in autism “cannot be solely attributed to the expansion of diagnoses to include higher functioning children.”
But that’s a straw man argument. No one has claimed that the observed increase in autism is due to that single factor. The trend doesn’t prove that the true prevalence of autism has increased, nor does it negate the larger pattern over decades that shows a broader definition of autism over time has captured more mild cases.
Fombonne said that when he trained as a psychiatrist, as many as three-quarters of children with autism also had intellectual disability, compared with 40% in the current survey. Surveys from around the world have documented a large decline in this percentage over the decades, he said, and the current CDC percentage is similar to what has been observed in other high-income countries. He said the trend “is not particularly alarming” and cannot be interpreted without taking into account a variety of factors.
Cautioning that the data on intellectual disability in the CDC reports may not be representative, Durkin pointed to better identification of autism among low-income children, who consistently show higher rates of intellectual disability and previously were underrepresented in the CDC surveys. That underrepresentation has recently reversed, beginning with the CDC’s 2020 survey, she said, with autism now being slightly more common among low-income versus high-income children.
Durkin said the change in the pattern “could well be due” to a change in the CDC’s methods in 2018 and “could explain the uptick” in the percentage of children with co-occurring intellectual disability in the most recent CDC reports. For surveillance year 2018, the CDC no longer required clinical review to identify an autism case, which Durkin’s work showed had contributed to disparities in autism prevalence.
Other experts also warned against reading too much into the numbers, noting that only a subset of children had IQ information, which varies greatly by site, and there could be different reasons why this information is or isn’t reported.
“You wouldn’t have an IQ because the kid is so low that nobody knows how to test them, or you wouldn’t have an IQ because the child went to a psychiatrist and was clearly very verbal and articulate, and they didn’t want to bother getting an IQ,” Lord said. “The trouble with IQ data is it’s very hard to interpret it unless you know exactly what the tests were and when the kids were seen and the competence level of the person giving the tests.”
She added that “there has been a big push by parent advocates to pay more attention to the more severely affected profoundly autistic kids” and that people may have improved over the past decade in knowing the tests that can be given to a child with profound autism, which could lower overall scores.
Despite HHS’ emphasis on this trend, the CDC report itself does not directly identify or discuss it. The report, however, does suggest that various so-called social determinants of health “could lead to higher rates of disability among certain populations,” noting that more Black babies are born preterm than white babies, and that lead poisoning and traumatic brain injuries are other causes of intellectual disability. Disparities in access to early therapies, which can improve cognition, could also contribute, the report added.
Editor’s note: FactCheck.org does not accept advertising. We rely on grants and individual donations from people like you. Please consider a donation. Credit card donations may be made through our “Donate” page. If you prefer to give by check, send to: FactCheck.org, Annenberg Public Policy Center, P.O. Box 58100, Philadelphia, PA 19102.