Study of 2.5 Million U.S. Children Finds No Link Between Early MMR Vaccination and Autism
The largest U.S. study of its kind has concluded there is no association between receiving the measles-mumps-rubella (MMR) vaccine before age 2 and childhood autism. The paper, published online July 28 in the *Pediatric Infectious Disease Journal*, analyzed electronic health records from 2,560,035 American children with linked birth-parent data, covering January 2015 through December 2024 with follow-up through age 8 or the last recorded visit. The authors — Todd L. Burstain, Dale Jacques and Jennifer M. Burstain of UW Medicine at the University of Washington in Seattle — say the findings "reinforce existing evidence supporting the safety of MMR vaccination."
Rather than simply comparing vaccinated and unvaccinated children, the study asked a finer question: does the *timing* of the first MMR dose affect autism risk? Using age-anchored landmark analysis, the team compared children vaccinated within defined age windows against children unvaccinated at the start of each window.
The primary window, 11.5 to 24 months, produced an adjusted hazard ratio (aHR) of 0.97 with a 99% confidence interval of 0.91–1.03 — consistent with no difference in risk. Narrower windows landed in the same place: 11.5–12.5 months at aHR 0.98 (99% CI 0.93–1.04), 12.5–13.5 months at aHR 0.98 (0.93–1.05), and 13.5–14.5 months at aHR 0.98 (0.91–1.05).
To test whether the near-null findings could be explained by residual confounding, the authors ran a negative control: the pneumococcal conjugate booster given at 12–15 months, which has no plausible biological link to autism. That exposure also returned a near-null result — aHR 1.01 (99% CI 0.99–1.04) — supporting the conclusion that the MMR findings are unlikely to be an artifact of unmeasured bias. The two later MMR windows (14.5–17.5 and 17.5–24 months) showed slightly lower hazard ratios of 0.90 and 0.92, which the authors attributed to residual bias rather than any protective effect.
The MMR-autism scare traces to a 1998 case series of 12 children by Andrew Wakefield, later retracted for scientific misconduct. Since then, every large study has come back negative: a 2002 Danish study of 537,303 children (adjusted RR 0.92), a 2019 Danish cohort of 657,461 children, a 2014 meta-analysis pooling five cohort studies of more than a million children, and a 2021 Cochrane systematic review.
The new study lands amid a heated U.S. vaccine-policy debate. Health Secretary Robert F. Kennedy Jr. has called aluminum adjuvants in vaccines "neurotoxic" and directed the CDC to investigate a possible vaccine-autism link. The CDC updated its MMR webpage in July 2026: where the November 2024 version said no studies had concluded that MMR causes autism, the new wording says the possibility "has not been ruled out." Measles cases, meanwhile, are at roughly a 35-year high, according to media reports.
By 24 months, 91.4% of the study sample had received a first MMR dose. The prevalence of the narrow "childhood autism" diagnosis code in the sample was 2.55%, slightly below the national estimate of about 3.2%. Median follow-up was 853 days for vaccinated children versus 366 days for unvaccinated children — and still no signal emerged.
The study has real limits. Electronic health records are subject to diagnostic coding errors; follow-up ended at age 8, so later diagnoses could be missed; the data include no family-history information, preventing stratification by genetic risk; and as an observational study it can demonstrate association, not causation.
A study of this size and design is about as strong an answer as observational epidemiology can give. Whether it quiets the controversy — with a federal investigation still underway — is another question.