First Measles–Mumps–Rubella Vaccination and Childhood Autism

Background: Concerns about potential associations between measles–mumps–rubella (MMR) vaccination and autism persist despite extensive prior research. Observational studies evaluating vaccine timing are susceptible to bias from healthcare utilization patterns, neurodevelopmental concerns and time-dependent exposure classification. Methods: We conducted a retrospective cohort study using the Cosmos electronic health record database, including 2,560,035 US children with linked birth-parent data and follow-up through age 8 years. To minimize bias related to exposure timing and healthcare utilization, we performed prespecified age-based landmark analyses evaluating the first MMR vaccination administered between 11.5 and 24 months. Adjusted hazard ratios (aHRs) for autism were estimated using Cox proportional hazards models with covariates selected via directed acyclic graphs. A negative control exposure (pneumococcal conjugate vaccine booster administered at 12–15 months) was analyzed to assess residual confounding. Results: In the primary 11.5–24 month landmark cohort, the first MMR vaccination was not associated with increased autism risk (aHR 0.97; 99% confidence interval 0.91–1.03). Across all age-specific landmark cohorts, aHRs remained near the null. The negative control analysis yielded a similarly near-null estimate (aHR 1.01; 99% confidence interval 0.99–1.04), supporting that residual bias is unlikely to explain the findings. Conclusions: In this large US cohort, age-anchored landmark analyses show no association between MMR vaccination before 24 months and childhood autism. The concordant near-null findings for both MMR and a negative control exposure support that the observed results are unlikely to be explained by residual confounding. These findings highlight the importance of age-specific analytic approaches in evaluating vaccine safety using observational data.