Selected publication milestones, not an exhaustive history or discovery priority claim.
1996 — randomized pertussis-antigen prevention establishes clinical protection while comparing formulation-specific reactogenicity. Original PMID8538705.
1999 — UK MMR introduction/onset-clustering tests do not reproduce a major autism timing effect. Original PMID10376617.
2000 — individual/community exemption analysis links lost vaccine protection to measles/pertussis risk. Original PMID11135778.
2003 — matched Danish vaccine-formulation/dose comparison finds no increased autism risk from thimerosal; other developmental endpoints remain distinct. Original PMID14519711.
2005 — Yokohama ASD incidence does not fall after MMR withdrawal, with calendar/diagnostic context retained. Original PMID15877763.
2009 — neuropsychological follow-up after randomized infant dose-regimen assignment shows few small differences and does not directly test ASD incidence. Original PMID19171612.
2010 — validated individual thimerosal exposure/ASD comparison is null; later antigen-load analysis(PMID 23545349) reuses the same parent sample. Original PMID20837594.
2010 — first-doseMMRVfebrile-seizure risk is quantified against separate MMR plus varicella, demonstrating measurable formulation-specific safety effects. Original PMID20587679.
2015 — MMR familial-risk cohort shows no increased ASD association in children with affected older siblings. Original PMID25898051.
2019 — larger Danish MMR cohort examines defined risk and postvaccination timing groups with no consistent increased ASD signal. Original PMID30831578.
2022 — ten-year varicella trial follow-up retains pathogen-specific clinical prevention, with dose and parent-trial scope. Original PMID34435933.
2026 — US EHR age-landmark and negative-control study adds2.56 million-child near-null MMR/ASD evidence without eliminating every possible residual bias. Original PMID42520244.
Historical ethylmercury, environmental methylmercury and MMR are different exposures; actual formulation/dose/route must be recorded.
A near-null negative control limits specified bias pathways but cannot eliminate all residual confounding.
Risk of very rare baseline-defined effects is bounded by precision; no responder-defined subgroup can rescue a failed broad prediction.
General neuropsychological tests, tics, biomarkers and animal proxies are not incident clinical autism.
Absolute net-health outcomes require linked infection, adverse-event and diagnostic follow-up rather than one favorable marker or passive-report trend.
Shared registry/VSD/clinic/trial reports require study grouping and distinct error-mode assessment.
Safety surveillance should preserve known formulation/age-specific adverse events and baseline clinical contraindications.
Pre-register a target-trial emulation of existing naturally varying vaccine timing or historical matched formulations without withholding recommended vaccination. Define eligibility and familial risk before outcomes, validate exposure and incident diagnosis, use correct time-at-risk and age landmarks, measure healthcare contact and independent negative controls, and link confirmed infection, serious morbidity and specific adverse events over a stated horizon. Independently replicate absolute net outcomes and proposed dose predictions; no responder-only or postdiagnosis susceptibility scope.