Skip to main content

Welcome to a Grasp lesson

Create a free account to keep studying and create your own study path.

Lesson 6

Evaluating Vaccine Safety and the Autism Claim

Recommended resources

Hello. The previous lesson showed why a study can support a narrow causal finding without establishing every broader claim made about it. That distinction matters here too. The claim that vaccines cause autism is testable, but it is not the same claim as “vaccines can have adverse effects.” This lesson examines the evidence for autism separately from the evidence for recognized vaccine harms.


Why a compelling sequence is not enough

A child may receive a vaccine shortly before a family notices signs of autism. The sequence is real and can be deeply concerning, but an event occurring first does not establish that it caused the event noticed later. Routine childhood vaccination and the recognition of developmental differences both occur in early childhood. To test causation, researchers need to compare outcomes among children with different vaccination histories, while accounting for other differences between them.

Watch these excerpts from PBS NewsHour’s What research reveals about the rise in autism diagnoses and why vaccines aren't the cause. They explain how the original MMR-autism allegation arose and why investigators moved from individual cases to comparative studies.

What research reveals about the rise in autism diagnoses and why vaccines aren't the cause

PBS NewsHour traces the early MMR-autism claim and the research that followed. Watch it for the difference between an observed sequence of events and a study that can test whether vaccination changes autism risk.

Watch the original claim, from the account of the 1998 paper through its retraction. Continue with the subsequent research on comparison groups and evidence syntheses. In the later discussion of rising diagnoses, watch diagnostic trends for reasons a change in recorded autism prevalence cannot, by itself, identify a cause.

The 1998 paper described only 12 children, lacked a suitable comparison group, and was retracted after serious problems with its research came to light. Even if its case histories had been accurate, they could not have shown whether autism occurred more often after MMR than it otherwise would have. Retraction addresses the paper’s integrity; the larger studies that followed address the underlying causal claim.

It helps to separate three versions of that claim:

  • MMR vaccination causes autism.
  • Thimerosal—also spelled thiomersal, a preservative used in some vaccine formulations—causes autism. MMR does not contain thimerosal.
  • Receiving multiple vaccines or vaccine antigens increases autism risk.

Evidence about one exposure should not silently be treated as evidence about all three. Likewise, changes in autism diagnoses over time require care: broader diagnostic criteria, screening, and awareness can change the number of recorded cases without demonstrating that a particular exposure caused the change.



What the strongest comparisons find

A well-designed observational comparison records vaccine exposure, follows children long enough for an autism diagnosis to be identified, and considers factors that could differ between vaccinated and unvaccinated groups. No single comparison is perfect. Large population cohorts, studies of children with an autistic sibling, and systematic reviews are valuable in part because they test the claim in different ways.

Read selected portions of the World Health Organization’s Vaccines, thimerosal and autism spectrum disorder review. Its discussion distinguishes results from individual studies from conclusions drawn across studies.

Vaccines, thimerosal and autism spectrum disorder

Read the World Health Organization review to see how a large cohort and several evidence syntheses assess the autism claim, including studies that reported contrary findings.

In “Childhood vaccination and ASD (n = 5)” (pp. 23–25), read the cohort example describing the nationwide Danish study. Then find “Evidence in support of a relationship between vaccination and ASD (n = 2 articles)” (p. 26) and read the contrary studies, noting how their exposure and autism outcomes were measured. Finally, in “Results from peer-reviewed meta-analyses (n = 4)” (pp. 26–28), read the review findings. Notice that the total number of children in a broad vaccine review is not necessarily the number assessed for its autism outcome.

In the Danish cohort, researchers followed 657,461 children and identified 6,517 autism diagnoses. The adjusted hazard ratio for autism comparing MMR-vaccinated with unvaccinated children was 0.93, with a 95% confidence interval of 0.85 to 1.02. An estimate close to 1 indicates little observed difference in this comparison; the interval expresses uncertainty around that estimate. It would be wrong to translate this into “proof that the true effect is exactly zero.” It would also be wrong to present it as evidence of a substantial increase. Analyses of children with autistic siblings, children with higher assessed autism risk, and different intervals after vaccination likewise found no increased risk.

Evidence syntheses do more than count papers. They examine which studies measured the exposure and outcome reliably, used appropriate comparison groups, addressed confounding, and produced consistent results. The WHO review reports that major syntheses of MMR and autism find no association. It also describes studies reporting positive associations but rates their evidence very low because, among other limitations, some relied on aggregate vaccination rates or parental reports and had weak autism-outcome measurement.

The selection diagram below belongs to one vaccine-autism meta-analysis discussed by the review. It illustrates how many identified studies did not meet that analysis’s inclusion criteria. Selection is part of a transparent review; the diagram alone does not tell us whether the included studies were unbiased or what they found.

Selection flow for a vaccine-autism meta-analysis: 1,112 potentially relevant studies were initially screened and 10 were included after successive exclusions. The counts describe study selection, not the size or direction of a vaccine effect.

Syntheses may include overlapping primary studies, so their participant totals should not be added together as though each review supplied an entirely new group of children. Their strength lies instead in the quality of the underlying comparisons and the consistency of findings across approaches.



Testing the preservative and “too many vaccines” claims

The thimerosal claim is distinct from the MMR claim. It asks whether exposure to a particular preservative in certain formulations changes autism risk. It is also important not to treat all sources of mercury as interchangeable: evidence about an environmental mercury exposure does not automatically establish an effect from thimerosal in a vaccine.

Return to the WHO review for its comparison of thimerosal studies. This portion is useful because some studies did report an association; the task is to assess how much weight those findings bear.

Vaccines, thimerosal and autism spectrum disorder

The WHO review compares studies reporting positive thimerosal associations with studies and meta-analyses reporting no association. Read it to see why study quality matters more than a simple tally of results.

In “Reported association between thiomersal and ASD / NDDs (n =9)” (pp. 19–21), read the study appraisal. Focus on how cases, controls, birth periods, and vaccine exposure were defined. Then, in “Results from peer-reviewed meta-analyses (n = 2)” (pp. 21–22), read the pooled findings.

A large database does not automatically yield a reliable causal answer. For example, if children receiving different vaccine formulations also come from different birth periods, changes in autism diagnosis and reporting may distort the comparison. A reported dose-response pattern does not fix that problem if dose estimates and comparison groups inherit the same bias. Conversely, independently conducted studies with more reliable vaccination records and outcome assessment, together with the meta-analyses discussed in the review, do not support an association between thimerosal exposure and autism.

The “too many vaccines” version requires its own measure of exposure. The WHO review also discusses a study that estimated children’s cumulative exposure to vaccine antigens from immunization records and compared children with validated autism diagnoses to matched controls. It found no increased autism risk with greater measured antigen exposure. As with any observational study, limitations remain; its finding is most informative when read alongside the larger, consistent body of evidence.



A rejected autism claim does not mean “no adverse effects”

A vaccine can be beneficial overall and still cause an adverse reaction. The crucial distinction is between an adverse event following vaccination—something reported after a shot—and an adverse reaction causally linked to vaccination. Timing is a reason to investigate, not sufficient proof of the link. Safety assessment compares rates, timing, and patterns across suitable groups and continues after vaccines are in use.

The WHO review’s discussion of vaccine safety notes recognized concerns including febrile seizures after MMR-containing vaccines and idiopathic thrombocytopenic purpura, a rare condition involving low platelet counts. It also mentions aseptic meningitis in evidence covering particular vaccine formulations. These findings should not be generalized to every vaccine or formulation: the relevant product, outcome, and size of risk matter. Nor does recognizing these effects provide evidence for autism, which is a different outcome investigated separately in the same review.

The evidence-based judgment is therefore more precise than either “vaccines are risk-free” or “a reported harm proves the autism claim.” High-quality comparisons and syntheses do not support vaccines causing autism across the major exposures studied. Recognized adverse effects still belong in vaccine safety monitoring and in assessments of benefits and risks.

Public wording can make this distinction harder to see. A news report about changes to CDC website language provides an example of why an agency statement and a new research result must be evaluated separately.

CDC's new language regarding vaccines and autism

ABC News reports on a change in CDC wording about vaccines and autism. Watch this brief clip as an example of public communication about uncertainty, not as a substitute for the study comparisons and reviews above.

Watch the wording change, then the response. Focus on the difference between saying that research cannot eliminate every conceivable possibility and presenting positive evidence that vaccines cause autism. Treat numerical claims made in the interview as claims to check against the underlying reviews.

Science rarely proves that an effect of exactly zero is impossible in every population and circumstance. That logical limit does not make a substantial causal claim equally plausible. The relevant assessment is what the best available studies can detect, how well they address bias, and whether their findings consistently support the proposed effect.



Conclusion

The original MMR case report could not establish causation, and it was retracted. Much stronger comparisons and evidence syntheses have since examined MMR, thimerosal, and broader vaccine-exposure claims without finding support for increased autism risk. That conclusion is specific to autism; it does not deny recognized vaccine adverse effects or remove the need for safety monitoring.

The next lesson applies the same discipline to synthetic food dyes: identifying a specific exposure and outcome, then judging whether the evidence supports the health claim being made.