Fact Check: Did a CDC Study Show 99% of Serious Vaccine Injuries Go Unreported? No

The claim and verdict

Health and Human Services Secretary Robert F. Kennedy Jr. said at a September 13 Children's Health Defense conference that a CDC study showed the Vaccine Adverse Event Reporting System captures less than 1% of vaccine injuries. The verdict is misleading. The project he invoked was funded by the Agency for Healthcare Research and Quality, not conducted as a CDC study, and its under-1% sentence referred broadly to adverse events after vaccination—not proven serious injuries.

The distinction changes the denominator. Expected fever, soreness or a temporary rash may go unreported because patients and clinicians already know they can occur. That does not support the claim that 99% of serious vaccine-caused injuries disappear from surveillance.

AFP reported that Kennedy used the figure while arguing for a stronger tracking system. Better surveillance can be a legitimate policy goal. The cited number still must be described accurately.

What the old project actually examined

The Harvard Pilgrim Health Care project used electronic health records from 2007 through 2010 to explore automated reporting of possible adverse events. Its report said fewer than 1% of vaccine adverse events were reported to VAERS. It did not say all the unreported events were severe, and it did not determine that every event appearing after a shot was caused by that shot.

Timing is the entry point for safety investigation, not the conclusion. If someone has a heart attack, seizure or infection after vaccination, a report can help analysts look for patterns. Individual records can include coincidental illnesses that would have occurred anyway. Controlled studies, clinical review and comparisons with expected background rates are needed to test causation.

AFP cited published reviews finding wide variation in reporting, with serious-event reporting far higher than the under-1% figure for all events. No single percentage describes every vaccine, symptom, age group and reporting requirement.

VAERS is deliberately open—and deliberately limited

VAERS accepts reports from patients, relatives, clinicians and manufacturers. That breadth helps detect unusual signals early. Federal law also requires certain reports, and manufacturers must submit events brought to their attention. The database warns users that a report does not prove a vaccine caused the event.

A passive system has predictable limitations: underreporting, stimulated reporting after publicity, incomplete medical information and no unvaccinated comparison group in the raw report count. Those limits are reasons to investigate a signal with other methods, not reasons to multiply every report by 100 and label the result confirmed injuries.

The United States also uses active systems such as the Vaccine Safety Datalink, which analyzes health records from participating care organizations, and clinical networks that investigate complex cases. Kennedy's statement treated one line from one project as though it described the entire safety system.

The correct way to read a vaccine-safety number

The cited project supports the unsurprising point that many post-vaccination symptoms are never entered into a passive database. It does not support the much stronger claim that a CDC study found 99% of serious vaccine injuries are hidden.

• Define the event: any symptom after vaccination, a medically serious event or an event judged causally related are not interchangeable.
• Identify the surveillance system and whether it is passive, active, clinical or experimental.
• Compare observed events with the background rate expected without vaccination.
• Check whether the estimate applies to one vaccine and outcome or is being generalized to everything.