RFK Jr has vowed to find out what’s responsible, but scientists say he is ignoring answers from decades of research

Autism is on the rise: what’s really behind the increase?

RFK Jr has vowed to find out what’s responsible, but scientists say he is ignoring answers from decades of research

On 16 April, Robert F. Kennedy Jr held a press conference about rising diagnoses of autism. The US Health and Human Services (HHS) secretary pointed to new data showing that autism prevalence in the United States had risen steeply from one in 150 eight-year-olds in 2000 to one in 31 in 2022. He called it an “epidemic” caused by “an environmental toxin” — and said he would soon be announcing a study to find the responsible agent.

The next month, the US National Institutes of Health (NIH), part of the department that Kennedy leads, announced the Autism Data Science Initiative (ADSI). The initiative offered up to US$50 million to fund studies on the causes of autism. The winning applications are expected to be announced in September.

Usually, big investments in research are welcomed by scientists — but not this time. Many were dismayed that these developments seemed to ignore decades of work on the well-documented rise in autism diagnoses and on causes of the developmental condition. Although Kennedy said that environmental factors are the main cause of autism, research has shown that genetics plays a bigger part. Population studies¹ have linked a handful of environmental factors — mostly encountered during pregnancy — to increased chances of autism, but their precise role has been hard to pin down. More than anything, research has shown that the drivers of autism are fiendishly complicated. “There will never be a sound-bite answer to what causes autism,” says Helen Tager-Flusberg, a psychologist who studies neurodevelopmental conditions at Boston University, Massachusetts.

The rise in prevalence, many researchers say, is predominantly caused by an increase in diagnoses rather than a true rise in the underlying symptoms and traits. “We don’t see an epidemic of autism, but we see an ‘epidemic’ of diagnoses,” says Sven Bölte, a specialist in child and adolescent psychiatric science at the Karolinska Institute in Stockholm. Researchers are concerned that Kennedy, an anti-vaccine advocate, will use the ADSI to promote the disproven idea that vaccines are linked to autism.

Kennedy’s comments prompted equally strong reactions from autistic people and autism groups, says Alycia Halladay, chief science officer at the Autism Science Foundation, a non-profit research and advocacy group in New York City. Autistic people have a wide range of traits and symptoms — that means that many live and work independently, whereas others need intensive support. Some people want more research into the causes, she says, and others don’t.

But many people in the autism community found it demeaning and offensive when Kennedy said at the press conference that “these are kids who will never pay taxes, they’ll never hold a job”. These factors do not determine a person’s value, Halladay says — and many autistic people are employed and pay taxes.

Scientists and autistic people alike are also concerned that the administration of US President Donald Trump is actually cutting the overall amount of money going towards autism research and support for autistic people, including through cancelled or delayed grants. “Everyone is really, really upset that money is being taken away from these important research projects,” Halladay says.

What explains the rise?

Many studies suggest that the prevalence of autism is rising, and not just in the United States. The trend has been observed in other high-income countries, including the United Kingdom, Denmark, South Korea and Japan. “Everywhere people are doing research and are looking at the numbers, [they] are going up,” Bölte says (see ‘Tallying autism diagnoses’). In some less wealthy countries, however, a lack of health services and data mean that prevalence is difficult to assess.

Source: Autism and Developmental Disabilities Monitoring NetworkSource: Autism and Developmental Disabilities Monitoring Network

There are several reasons to think that a rise in diagnoses explains a large part of this trend. One is that the diagnostic criteria have changed over time in two periodically updated tools used by health professionals: The Diagnostic and Statistical Manual of Mental Disorders (DSM), used mainly in the United States, and the International Classification of Diseases (ICD).

These describe autistic people generally as showing differences in social interaction and communication, as well as having ‘restricted and repetitive behaviours and interests’. But until the early 1990s, the manuals had a narrow definition for autism, says Diana Schendel, an epidemiologist who studies autism at Drexel University in Philadelphia, Pennsylvania. For instance, the third edition of the DSM recognized autism only in young children and required that they meet a minimum number of criteria.

Updated versions of both handbooks — ICD-10 in 1990 and DSM-IV in 1994 — broadened the diagnostic criteria so that they could be applied to people with fewer symptoms and of different ages. “That dramatically changed the opportunity for individuals to be classified under the autism umbrella,” says Schendel. In 2013, DSM-V dropped some separate diagnoses, such as Asperger’s syndrome, and subsumed them under the term autism spectrum disorder. Autism is commonly diagnosed with co-occurring conditions, such as intellectual disability or attention deficit–hyperactivity disorder (ADHD).

A 2015 study that Schendel co-authored tried to quantify the causes of increased autism prevalence among Danish people born between 1980 and 1991, and diagnosed up to 2011. The authors estimated that 60% of the increase could be explained by changes in the early 1990s to diagnostic criteria and to the way in which diagnoses were reported to national health registries².

Diagnoses have also risen as health professionals have adjusted how they interpret and apply the criteria, and as they have adopted improved standardized diagnostic tools, such as structured interviews and observations. Eva Loth, a cognitive neuroscientist at King’s College London, says that she now diagnoses people in research projects whom she would have excluded ten years ago. “Our understanding of what autism is has changed,” says Loth.

The other change that has turbocharged diagnoses, researchers say, is increased awareness of autism and declining stigma among teachers, health professionals and the broader public, alongside increased availability of diagnostic and support services. Parents are more likely to observe signs of autism in children and are often incentivized to seek a diagnosis if it can help their child to access support, such as at school. “Where before a child might be classified as difficult, problematic in a classroom, now they might go seek a diagnosis,” says Nick Puts, who studies neurodevelopment at King’s College London and is autistic.

Children are also receiving diagnoses at a younger age than in the past. More adults are seeking diagnoses, and they have become more common in girls and women, in whom autism has historically been overlooked.

Bölte argues that the high demands of modern schools — such as extensive group work and digital information — could also play a part. “The learning environment is extremely complex today,” he says, “and there’s more and more children with maybe lesser traits or symptoms who are just kind of overwhelmed.” Similarly, more adults — navigating a highly complex world — might be seeking help and advice, leading to more autism diagnoses.

How much of the rise in autism is accounted for by increased diagnoses and awareness? “It’s certainly the vast majority,” Schendel says, although the exact proportion is difficult to calculate.

How prevalent is autism now?

The figure that Kennedy cited — a prevalence of one in 31 among eight-year-olds — stems from an April report by the Autism and Developmental Disabilities Monitoring Network³, a US surveillance programme that tallies autism in children using health and education records (see ‘Autism prevalence across the United States’). These show who has received a diagnosis and who is eligible for autism-related education. The results from this type of study are sometimes called registry or administrative prevalence, because they are based on administrative databases.

But some researchers view such studies with caution, says Damian Santomauro, a mental-health epidemiologist at the University of Queensland in Brisbane, Australia. Identification of autism can come from various sources, including psychiatrists, psychologists and educational settings. Some children who have extra behavioural or educational needs might be included in an administrative database, but would not meet stringent criteria for diagnosis in an independent clinical evaluation, Santomauro says. Conversely, some people with undiagnosed autism might be excluded from registries.

A more accurate way to estimate prevalence is through a population survey, in which researchers screen everyone in a representative group of children or adults for autism. Santomauro led the most recent report on autism by the Global Burden of Disease study (the largest scientific project on impacts of health conditions), which is based only on survey data. This estimated that autism spectrum disorder had a prevalence of one in 127 people globally (less than 1%) in 2021. That’s 62 million people overall⁴.

A long-term study of Swedish children, published this year, compared registry and survey prevalence in children who were born between 1993 and 2001. It found that the prevalence of autism symptoms that parents reported when their children reached 18 had remained stable, even as the prevalence of registered diagnoses had gone up⁵. And most of the increase in prevalence seen in studies comes from diagnoses of autistic people who do not have language impairment or a co-occurring intellectual disability, Tager-Flusberg says, whereas numbers of diagnosed autistic people with these conditions have changed little, if at all. All of this adds weight to the argument that spiking autism statistics are driven mostly by improved detection and diagnoses.

But Santomauro says that other high-quality surveys have made him think twice. A study published in 2020 that used rigorous survey methods reported a prevalence of more than 3% in five-year-old children in Japan — higher than previous estimates for autism in some other Asian countries⁶. So, although Santomauro agrees that most of the rise in autism is explained by increased detection and diagnosis, he says it remains possible that something could be pushing up “true prevalence” as well. But what?

What causes autism?

Researchers have been hunting for the causes of autism for decades — and have run up against difficulties at every step. “We just don’t really understand a lot about how any human brain functions,” says James Cusack, chief executive of Autistica, a UK autism research and campaigning charity in London, who has worked in autism research and is autistic. “So trying to find causal mechanisms for autism is, I think, much harder than people thought it was going to be 20 years ago.”

One broad conclusion to emerge from the available data is that genetics plays a huge part. “Family-history risk is probably the strongest risk factor there is,” says Schendel.

Researchers capture the genetic contribution using measures of heritability – an estimate of how much of the differences in autism traits between individuals can be explained by inherited genes rather than by environmental factors. A large, five-country study⁷ published in 2019 estimated the heritability of autism as about 80% — equivalent to the heritability of height. For comparison, the heritability estimates for depression are roughly 30–50%.

Identifying the individual genes involved, however, has proved tough. “The genetics themselves are wildly complex,” says Craig Newschaffer, an autism researcher at Pennsylvania State University in University Park.

Researchers have sought genes by comparing the genomes of autistic people and people who are not autistic in families and in bigger population studies. These have helped to pinpoint a selection of rare genetic variants — variations in DNA sequence — that cause a relatively large increase in the chances of autism. Many of these are de novo mutations that probably arose in a sperm or egg cell and were passed to a child. These large-impact variants are thought to be a major driver of autism in an estimated 10–20% of autistic people.

But “the genetic architecture is very different from one individual to another”, says geneticist Thomas Bourgeron, who studies autism at the Pasteur Institute in Paris. Researchers have also found that hundreds, perhaps thousands, of common variants have a tiny effect individually — making them hard to pinpoint — but a big effect when combined. If someone has enough variants, they could end up with symptoms and a diagnosis.

After genetics, environmental factors are “a smaller slice of the pie” in driving autism, Schendel says, although there is debate about the size of the slice. Some researchers argue that the influence of environmental factors is vanishingly small, and others disagree. Most research suggests that any environmental factors exert their influence before birth, however, and studies have convincingly shown no link to vaccines⁸^,⁹.

Robert F. Kennedy Jr held an April press conference about the increase in autism prevalence.Credit: Elizabeth Frantz/Reuters

Many studies suggest that older parental age — for mothers or fathers — is linked to an increased likelihood of autism in children¹⁰. The average childbearing age has been increasing in many high-income countries, so this could be playing a small part in increasing autism prevalence, researchers say. One possible explanation is that children of older parents are more likely to have de novo mutations than are children of younger parents.

Infections during pregnancy have also been linked in many studies to a greater chance of autism, as has exposure to air pollution before birth. For instance, a US study published in June, involving more than 8,000 children and their mothers, found that exposure to higher levels of ozone was associated with autism¹¹.