Contested causes: vaccines, acetaminophen and leucovorin

Some claims about what causes autism carry the weight of the research literature, and others are still argued over.

Vaccines do not cause autism. Large cohorts and pooled analyses agree, across the measles-mumps-rubella (MMR) vaccine, thimerosal-containing vaccines and aluminum-adjuvanted vaccines, and no substantial body of human evidence points the other way.^1,3,5

On prenatal acetaminophen, the strongest study designs find no effect. Studies that compare children within the same family, one exposed before birth and one not, find no association, while ordinary cohort studies that cannot control for family factors report a small one, and some reviews still argue the case.^8,9,10,24

A third question, whether leucovorin eases autism symptoms, is not settled, and one trial that supported it has been retracted.¹⁴ It gets a short pointer below; a separate entry covers unproven treatments in full.

A Danish cohort of 657,461 children found no increase in autism with MMR vaccination (hazard ratio 0.93, 95% confidence interval 0.85 to 1.02).¹ A US electronic-records study of 2,560,035 children reached the same result for a first MMR before age two (adjusted hazard ratio 0.97, 99% confidence interval 0.91 to 1.03), and a comparison vaccine gave a near-identical estimate, which argues against hidden bias.² A meta-analysis of five cohort studies (1,256,407 children) and five case-control studies (9,920 children) found no link between vaccination and autism, and none either for the MMR vaccine, thimerosal or mercury on their own.³ A Cochrane review of the MMR vaccine, given alone or with a second vaccine, found no evidence of an increased risk of autism.⁴

The aluminum additives used in some vaccines have been examined separately. A Danish cohort of 1,224,176 children found no association between the amount of vaccine aluminum a child received in the first two years of life and any of 50 chronic conditions, including autism and other neurodevelopmental conditions; the authors note that small effects on rarer conditions could not be ruled out.⁵ A 2026 systematic review of 59 studies found high-quality evidence of no association between aluminum-adjuvanted vaccines and serious or long-term outcomes including autism.⁶

The claim still circulates. In 2026 a reanalysis by Karl Jablonowski and Brian Hooker, both of the group Children's Health Defense, argued that the Danish aluminum cohort does show an association for several autism diagnoses;⁷ the cohort's authors, Andersson and colleagues, reject that reading. The US move to have the study retracted is described under what was claimed.

A sibling-controlled study compares children in the same family who differ in whether they were exposed before birth. This removes the family factors, such as shared genes and the home a child grows up in, that ordinary cohort studies cannot separate from the drug. In Hong Kong, researchers built sibling-matched cohorts from 708,020 mother-child pairs; among 124,333 children the association between prenatal acetaminophen and autism was null (adjusted hazard ratio 1.00, 95% confidence interval 0.91 to 1.11), and it stayed null across timing, cumulative dose and pattern of use.⁸ A Swedish nationwide cohort of about 2.4 million children found a small association in the whole cohort, but it disappeared once siblings were compared (hazard ratio 0.98, 95% confidence interval 0.93 to 1.04).⁹

Whole-cohort studies, which do not control for family factors, report a small positive association, and two meta-analyses report a "1.17" for these figures, though for different outcomes: one gives a risk ratio of 1.17 for autism,¹¹ the other an odds ratio of 1.17 for ADHD, and no significant association with autism.¹² In the Hong Kong study the same signal appeared for acetaminophen taken before pregnancy, which points to family factors rather than to the drug.⁸ Both meta-analytic estimates shrink toward no difference once higher-quality studies or family controls are applied. A 2025 umbrella review of nine systematic reviews found that the only sibling-controlled analyses in the literature did not reproduce the association, and concluded that the evidence "does not clearly link" prenatal acetaminophen with autism or ADHD.¹⁰

Not every synthesis agrees. A 2025 review using the Navigation Guide method, a structured qualitative synthesis of 46 studies covering both whole-cohort and sibling-comparison designs, concluded the evidence is consistent with an association and called for pregnant women to be advised to limit acetaminophen use.²⁴ It judges the sibling analyses underpowered; the sibling cohorts above hold that the family comparison is what removes the shared family factors the whole-cohort studies cannot.

In September 2025 the US Department of Health and Human Services named leucovorin (folinic acid, a form of folate) a "promising treatment for symptoms of autism" and asked the Food and Drug Administration to make it available.¹³ The largest trial cited in support, a 2024 randomized trial in 80 children, was retracted by its journal on 29 January 2026, after a statistical review could not reproduce its results from the data.¹⁴ A 2026 preprint that pooled five small randomized trials found no clear overall benefit, with a possible effect on language and communication; it has not been peer reviewed.²⁵ The FDA's approval in March 2026 was for cerebral folate deficiency, an ultra-rare inherited condition, not for autism itself.^15,23

On both topics, two camps appear. The no-link position is held by the research groups in the sections above and by pediatric and obstetric bodies; the linking position is promoted by the group Children's Health Defense and, since 2025, by parts of the US federal government.

On 1 August 2025 the US health secretary, Robert F. Kennedy Jr., called for the retraction of the Danish aluminum cohort study in an opinion piece. On 11 August 2025 the editor of Annals of Internal Medicine declined, writing that retraction is warranted only for serious errors or misconduct, "neither of which occurred here."¹⁶

On 22 September 2025 the Food and Drug Administration said it had "initiated the process for a label change" for acetaminophen, citing evidence of a possible association with autism and ADHD, and sent a warning letter to physicians. In the same announcement the agency said that "a causal relationship has not been established."¹⁷ As of October 2026 the over-the-counter label is unchanged: the current Kenvue label for Tylenol Regular Strength, revised 27 May 2026, still carries only the standard advice to ask a health professional before use in pregnancy.¹⁸

Professional bodies and the drug's maker disputed the acetaminophen claim. The American College of Obstetricians and Gynecologists said the announcement was "not backed by the full body of scientific evidence,"¹⁹ and Kenvue, which makes Tylenol, asked the FDA to deny the citizen petition behind the change.²⁰

On 20 November 2025 the Centers for Disease Control and Prevention changed its vaccine-safety page to say that the statement "vaccines do not cause autism" is "not an evidence-based claim because studies have not ruled out the possibility that infant vaccines cause autism," and said the Department of Health and Human Services had begun an assessment of the causes of autism.²² The cohorts in the vaccines section test that claim: their estimates sit close to no difference, with confidence intervals that exclude a large increase in risk.^1,2

On 10 September 2026 Kenvue and 14 retailer co-defendants asked a federal appeals court to rehear the July 2026 ruling that revived more than 500 lawsuits over prenatal acetaminophen; the request was still pending at the end of September.²¹

A single observational study can show that two things occur together. It cannot show that one causes the other. The reason is confounding: a third factor, measured or not, can drive both. Random assignment is the design that balances the confounding factors between groups, whether or not the researchers measured them. No trial can randomize a pregnancy exposure, and no vaccine trial has been powered to test autism as an outcome, so these questions are answered with the strongest observational designs available instead.

Two designs do most of the work here. Large cohorts follow many people and can show an association. Family designs, which compare siblings who differ in exposure, subtract the factors the siblings share. Each has limits: no cohort can measure everything, and no sibling comparison can remove factors that vary between siblings. This is why a small association in a single cohort means little on its own, and why a result that survives in several designs and several populations carries more weight.^8,10

Dose and timing matter too. A real effect usually shows a pattern, appearing at higher exposure or within a sensitive period. Where a study finds an association whatever the dose and whenever the exposure, confounding is the likelier explanation.

A retraction is a separate kind of signal. It does not mean the authors were dishonest; a retraction notice states the journal's own grounds. In the folinic acid case, the notice reports that the results could not be reproduced from the data.¹⁴ When a claim rests on its largest trial and that trial is retracted, the claim loses its strongest support, leaving only the smaller, inconsistent ones.

What would settle these questions differs by subject. For vaccines, the evidence is already conclusive, and what remains is its acceptance rather than more studies. For acetaminophen, a randomized trial is not feasible, so the question turns on whether the null result holds in more populations and on a fuller account of why some pregnancies were treated. For leucovorin, an adequately powered randomized trial, repeated in a second group, with outcomes that matter to autistic people, would move it from a handful of small studies to a real answer.

Autistic writers treat the vaccine claim as a settled question that cost them years. Eric Garcia, an autistic journalist, writes that anti-vaccine "quackery infected our politics" and that although the retracted study was withdrawn, "the damage was already done."^c1 Steve Silberman's history records that parents caught in the fight over vaccines and mercury came to call it the "Autism Wars."^c2

The deeper argument is about the premise. Kayla Rodriguez, writing in Sincerely, Your Autistic Child, says autism "isn't a bad 'disease' that needs to be 'cured' or 'prevented'."^c3 Garcia, recalling a campaign that promised to end autism, writes that it was "corrosive" to hear it.^c4 Asking what causes autism, these writers suggest, slides easily into asking how to prevent it. Mitzi Walz, a parent writing in a critical-autism collection, describes how parents defending themselves against blame for their children's autism were drawn toward treatments that ranged "from elimination diets to mercury chelation."^c5