Your Kid's Autism Was Not Caused by Tylenol, Vaccines, WiFi, or Your Birth Control

Andrew Wakefield published a study in The Lancet in 1998 claiming a link between the MMR vaccine and autism. The study had twelve subjects. Twelve. It was a case series, not a controlled trial. It contained no statistical analysis. The subjects were referred by lawyers who were building a lawsuit against vaccine manufacturers, a fact Wakefield did not disclose. Wakefield held a patent on a competing single-dose measles vaccine, a fact he also did not disclose. He had received funding from the legal aid board to investigate a vaccine-autism link before the study began, which means the conclusion preceded the data.

The Lancet partially retracted the paper in 2004 and fully retracted it in 2010. The British General Medical Council found Wakefield guilty of serious professional misconduct and struck him from the medical register. Investigations by journalist Brian Deer revealed that Wakefield had falsified data, subjected children to invasive procedures without ethical approval, and manipulated the clinical histories of his subjects to fit his conclusions. The fraud was not subtle. It was methodical, documented, and comprehensive.

The retraction did not retract in public consciousness. This is the part that matters. The Lancet pulled the paper. The medical community disowned it. Every subsequent study; and there have been dozens, involving millions of children across multiple countries; found no link between the MMR vaccine and autism. The evidence is as definitive as epidemiology gets. And yet. A 2024 survey found that 21 percent of American adults believe vaccines cause autism. The number has not budged meaningfully in a decade. The fraud paper is gone. The belief it created is permanent.

The Slot Machine

Wakefield’s paper worked because it fit a template that was already waiting for it. The template goes like this: autism rates are rising. Something in the environment must be causing it. Your child was developing normally and then something happened. The something is the cause.

The template has a slot where the environmental villain goes, and the slot will accept almost anything. Vaccines were the first major occupant of the slot, but they were not the last. When the vaccine hypothesis lost institutional credibility (though not popular credibility), other candidates moved in. The slot does not stay empty. It cannot stay empty, because the template demands an explanation and the actual explanation; diagnostic expansion and improved surveillance; does not feel like an explanation. It feels like a bureaucratic technicality. It does not satisfy the narrative need.

Tylenol took a turn in the slot. A 2021 consensus statement published in Nature Reviews Endocrinology (not a research study; a statement) argued that prenatal acetaminophen exposure could disrupt fetal neurodevelopment. Media coverage treated it as a finding. Facebook groups and parenting forums circulated it as established science. The actual evidence base consisted of observational studies with significant confounding; women who take more Tylenol during pregnancy may differ from women who take less in ways that independently affect child development; and no replication of a causal mechanism had been achieved. The litigation industry moved faster than the science. Over 100,000 lawsuits were filed against Tylenol manufacturers before the scientific question was resolved. A federal judge eventually excluded the expert testimony supporting the link, finding it scientifically insufficient. The lawsuits continued. The preprint circulates in mommy Facebook groups to this day.

WiFi took a turn. Electromagnetic field exposure has been investigated as a potential contributor to autism, ADHD, and various neurological conditions. The studies are largely negative. The World Health Organization’s International Agency for Research on Cancer classifies radiofrequency electromagnetic fields as “possibly carcinogenic to humans” based on limited evidence from cell phone use, but the autism connection is unsupported by any credible epidemiological data. This has not prevented a cottage industry of EMF-shielding products marketed to pregnant women, including belly blankets, router covers, and grounding mats. The products solve a problem that does not exist, which is the ideal business model.

Glyphosate has been in the slot. Fluoride has been in the slot. Artificial food dyes, pesticides, microplastics, seed oils, birth control residues in the water supply. Each candidate follows the same trajectory: a plausible-sounding mechanism is proposed, preliminary or observational data is circulated, the media amplifies the finding, the finding enters parenting discourse as established fact, replication fails or never occurs, and the original claim persists in popular consciousness regardless. The slot always has an occupant. The occupant does not need to be correct. It needs to be scary, environmental, and recent enough to coincide with the observed rise in diagnosis rates.

Why Diagnosis Rates Rose

Autism prevalence estimates have increased from approximately 1 in 2,500 in 1985 to 1 in 36 in 2023. This is a staggering number on its face, and it is the number that drives the entire environmental panic. If 1 in 36 children are autistic now and 1 in 2,500 were autistic forty years ago, something must have happened. Something is causing this.

The something that happened is diagnostic. The definition of autism changed; repeatedly, dramatically, and in ways that expanded the eligible population by orders of magnitude.

The DSM-III in 1980 defined “infantile autism” as a rare, severe condition characterized by profound social withdrawal, language absence, and stereotyped behaviors with onset before 30 months. It described what we would now call Level 3 autism; children with severe impairments who were often institutionalized. The criteria were narrow. The condition was uncommon by definition.

The DSM-III-R in 1987 broadened the criteria. The DSM-IV in 1994 broadened them further and added Asperger’s disorder as a separate diagnosis, capturing individuals with average or above-average intelligence who had social difficulties but intact language. This single addition moved millions of people from “quirky” to “diagnosable.” The DSM-5 in 2013 collapsed everything into autism spectrum disorder with severity levels, which further expanded the boundaries by eliminating the requirement that symptoms cause clinically significant impairment in all cases.

Each revision brought more people inside the diagnostic boundary. Not because more people became autistic. Because the definition of autism grew. A person born in 1975 who was socially awkward, intensely focused on specific interests, and uncomfortable with eye contact was called shy. Or difficult. Or gifted. Or “a handful.” The same person born in 2005 gets evaluated, meets DSM-5 criteria, and receives an autism spectrum disorder diagnosis. The person did not change. The definition did.

Surveillance also improved. Schools implemented universal screening. Pediatricians added developmental screening to well-child visits. Parents became educated about autism signs through public awareness campaigns. Special education services required a diagnosis for access, which created the same structural incentive that drives ADHD diagnosis: the label opens doors to resources. A child without a diagnosis gets told to try harder. A child with a diagnosis gets an IEP, accommodations, therapeutic services, and legal protections. In a context where support requires a label, labeling increases.

The epidemiological data supports the diagnostic explanation. Studies examining whether autism prevalence has genuinely increased; as opposed to diagnostic prevalence; have consistently found that much of the increase is accounted for by broadened criteria, improved surveillance, diagnostic substitution (children who would previously have been diagnosed with intellectual disability or language disorders being reclassified as autistic), and increased awareness. A 2015 Danish study found that 60 percent of the increase in autism prevalence from 1980 to 2012 was attributable to changes in diagnostic criteria and reporting practices alone.

The 1990 cohort was not less autistic. They were less diagnosed.

The Neurodiversity Irony

Here is where the story turns strange. Two movements are running simultaneously, and they point in opposite directions.

The anti-vaccine movement (and its successors in the environmental causation space) treats autism as a catastrophe. Something caused it. Something broke your child. The framing is inherently pathological; autism is damage, inflicted by an external agent, and the goal is to identify the agent and eliminate it. Prevention is the objective. The implied message is that autism should not exist, and that a world without autism would be better.

The neurodiversity movement, running in parallel, argues that autism is a neurological variation, not a disease. Autistic people are not broken. They are differently wired. The movement advocates for acceptance, accommodation, and the recognition that autistic cognition has value; that the intense focus, pattern recognition, systematic thinking, and sensory sensitivity associated with autism are features, not defects. The neurodiversity framework does not want a cure. It wants a world that accommodates neurological diversity rather than demanding conformity.

These two movements are culturally and politically incompatible, and they are both loud, both growing, and both focused on autism. The anti-vax parent fundraising for autism research to find the cause and prevent future cases is operating in direct opposition to the autistic adult arguing that there is nothing to prevent. The parent wants to know what went wrong. The autistic person is saying nothing went wrong.

The dark irony sits at the intersection. The anti-vaccine movement killed children. Measles outbreaks in communities with low vaccination rates have hospitalized and killed kids. The 2019 measles outbreak in Samoa killed 83 people, mostly children, after vaccination rates dropped below 35 percent following anti-vaccine campaigning. These children died because their parents were trying to prevent a neurodevelopmental variation that the neurodiversity movement is simultaneously arguing is not a disease. Children died to prevent a condition that, according to a growing consensus among the people who actually have it, does not need preventing.

The Wakefield fraud did not just produce bad science. It produced a framework; autism as environmental catastrophe; that has now been filled with a rotating cast of villains, each one more speculative than the last, each one surviving its own debunking because the framework is more durable than any individual claim within it. The framework persists because it serves a psychological need. Parents who watch their child struggle want a reason. “The diagnostic criteria expanded” is not a satisfying reason. “Something in the environment did this” is a satisfying reason, because it implies the possibility of prevention, which implies the possibility of control, which is what terrified parents most desperately want.

What Is Actually Known

The genetics of autism are complex and polygenic. Hundreds of genes contribute to autism risk, each with small individual effects. Twin studies consistently show high heritability; concordance rates for monozygotic twins are 60 to 90 percent, compared to 0 to 30 percent for dizygotic twins. The genetic architecture is not simple Mendelian inheritance; it is a landscape of common variants, rare variants, de novo mutations, and gene-environment interactions that are genuine but far more subtle than “Tylenol did it.”

Environmental factors likely play a role, but the relevant factors are not the ones dominating public discourse. Advanced paternal age is associated with increased autism risk, likely through accumulated de novo mutations in sperm. Extreme prematurity and very low birth weight are associated with increased risk. Prenatal exposure to valproic acid (an anti-seizure medication) is associated with increased risk through a well-characterized mechanism. These are real findings with real data behind them. They are not popular topics in Facebook parenting groups because they do not fit the villain template. An older father is not a satisfying villain. A prescription medication taken for a pre-existing condition is not a satisfying villain. The environmental causation framework needs an external agent; something that was done to you, not something that was part of you.

The people most harmed by the environmental causation panic are autistic people themselves. Every dollar spent investigating debunked vaccine claims is a dollar not spent on services, accommodations, and support for people who are autistic right now. Every news cycle consumed by the latest environmental scare reinforces the message that autism is a thing to be feared and prevented rather than understood and accommodated. The framework that promises to protect children from autism is, in practice, a framework that tells autistic children they are the consequence of something going wrong.

Nothing went wrong. The diagnostic criteria expanded. The surveillance improved. The 1990 cohort grew up undiagnosed, navigating a world that had no name for how they experienced it. The 2020 cohort gets a name, and with the name comes services and identity and community; and also stigma and panic and a billion-dollar industry of false causation.

Nobody gave you the tism. You were always built like this.