Few questions in child health have been studied as often as this one, and few still cause as much worry: can the MMR vaccine (measles, mumps, rubella) cause autism? If you are a parent who has asked it, you are far from alone. In July 2026, a study of more than 2.5 million American children, published in The Pediatric Infectious Disease Journal and covered in October by PsyPost, added one of the largest answers so far. Here is what it did, what it found, and what it cannot tell us.
What the Study Did
Todd Burstain, a physician at UW Medicine in Seattle, and two co-authors used Cosmos, a research database built from Epic electronic health records across US hospitals and clinics. They included 2,560,035 children who had at least one routine primary care visit between 12 and 24 months of age. Each child was followed until age 8 or the end of the data. By their second birthday, 91.4% had received a first MMR dose.
Comparing vaccinated and unvaccinated children sounds simple, but it hides a trap called immortal time bias: depending on how and when children are sorted into groups, the comparison can be quietly distorted. To avoid it, the team used a landmark analysis: in each age window of the second year of life, children vaccinated during that window were compared with children of the same age not yet vaccinated, counting only autism diagnoses made afterwards.
The researchers also adjusted for factors chosen in advance with a causal diagram (such as the mother's age, gestational age, pregnancy complications, neighbourhood disadvantage and region) and ran a negative control: the same analysis for the pneumococcal booster given around 12 to 15 months, a vaccine with no proposed link to autism. If the method were biased, it would likely show up there too.
What They Found
- Main result: children who received their first MMR dose between 11.5 and 24 months were no more likely to be diagnosed with autism. The adjusted hazard ratio was 0.97 (99% confidence interval 0.91 to 1.03), meaning no measurable difference.
- How small a risk could it miss? The upper end of that interval rules out an increase in risk of more than about 3%.
- Negative control: the pneumococcal booster showed nothing either (hazard ratio 1.01, 99% CI 0.99 to 1.04), a sign that the method was not creating false signals.
- Consistency: results stayed near no difference across age windows and sensitivity analyses. Two later windows showed slightly fewer diagnoses in vaccinated children (0.90 and 0.92), which the authors attribute to remaining bias, not protection.
What the Study Cannot Say
Good research states its limits, and this paper does:
- It is observational. It does not assign vaccines at random. Burstain told PsyPost that "causation can only be claimed with randomized prospective trials." A trial that deliberately leaves young children unprotected against measles would not be ethical, which is why large, carefully designed cohorts like this one are the strongest tool available.
- No family history of autism. The records did not include it, a real gap given how strongly autism runs in families, although the authors note it would more likely make the vaccine look protective than harmful.
- Unequal follow-up. Vaccinated children were followed longer (median 853 versus 366 days), giving more time for a diagnosis to be recorded, yet no increase appeared.
- A narrower definition. Only the "childhood autism" code was studied, not the whole spectrum: 2.55% of the children, below the CDC's latest US estimate of about 1 in 31 eight-year-olds.
Not the First Study to Look
This result fits a long line of research. In 2019, a Danish study in Annals of Internal Medicine followed 657,461 children and found no increased risk of autism after MMR (hazard ratio 0.93), including in children with a sibling with autism, who are at higher risk. A 2015 US study in JAMA of 95,727 children with older siblings reached the same conclusion, whether or not the older sibling was autistic.
The fear itself began with a 1998 paper in The Lancet describing just 12 children. In 2010 the journal retracted it, after the UK's General Medical Council found serious ethical failings and the editors concluded that key statements in the paper were false.
Why the Question Keeps Coming Back
If the evidence is this consistent, why does the worry persist? Two reasons are worth understanding, and neither is about parents being careless.
Timing. The first MMR dose is usually given early in the second year of life, the very period when the first signs of autism (fewer words, less shared attention, a child who seems to withdraw) often become noticeable. When two things happen close together, it is deeply human to connect them. That is exactly why studies like this one compare children of the same age, vaccinated or not.
Rising numbers. Autism diagnoses have increased a lot in recent decades. A 2026 Danish genetic study suggests this is largely because diagnostic criteria widened and awareness grew, so that people with milder traits are now recognised. We covered it in why ADHD and autism diagnoses are rising.
What We Know About Where Autism Comes From
Autism is one of the most heritable conditions studied in psychiatry. A 2019 study of about 2 million people in five countries, published in JAMA Psychiatry, estimated that around 80% of the variation in autism risk comes from inherited genetic factors. Hundreds of genes have been linked to autism, many of them involved in how the brain develops and how neurons communicate (see our article on autism genes and brain pathways). The evidence describes autism as a different way of developing, rooted mostly in genetics and early brain development, rather than something a vaccine switches on.
What This Means for You
If you are a parent with questions about vaccines, those questions deserve a real conversation: talk to your doctor or your child's doctor about vaccination. If you are the parent of an autistic child, nothing in this research suggests that a vaccination decision caused your child's autism.
And because autism is so often inherited, many adults first recognise their own traits when a child or relative is diagnosed. If that sounds familiar, our free AQ-10 autism test (10 questions, about 3 minutes) is a quick first step, and the adult autism test (RAADS) (15 questions, about 7 minutes) looks more closely at social, sensory and repetitive traits in adults. Screening tests are not a diagnosis, but they can help you put words to what you notice.
Source: Burstain TL, Jacques D, Burstain JM. Association Between First MMR Vaccination Before Age 2 Years and Childhood Autism in a U.S. EHR Cohort of 2.5 Million Children. The Pediatric Infectious Disease Journal, 2026 (published online 28 July 2026, ahead of print). DOI 10.1097/INF.0000000000005349. Retrospective cohort of 2,560,035 US children (Cosmos electronic health records, 2015-2024), age-anchored landmark analyses with a negative control exposure. Reported by PsyPost, 4 October 2026. Background: Hviid A et al. Annals of Internal Medicine 170(8):513-520 (2019), DOI 10.7326/M18-2101 (Danish nationwide cohort, 657,461 children). Jain A et al. JAMA 313(15):1534-1540 (2015), DOI 10.1001/jama.2015.3077 (95,727 US children with older siblings). Bai D et al. JAMA Psychiatry 76(10):1035-1043 (2019), DOI 10.1001/jamapsychiatry.2019.1411 (five-country cohort, 2,001,631 people).