Detecting adverse events associated with vaccination and determining whether the health outcome is a result of the vaccination are complex tasks due to a number of factors (IOM, 1997) including the following:
Lack of long-term follow-up. Many controlled studies are geared toward monitoring immediate reactions to the administration of the vaccine; subjects are often followed for 6 months at the most.
Small sample sizes. Vaccine trials to determine immunogenicity often involve sample sizes of no more than several hundred individuals. Trials of this size are unlikely to detect rare effects since large sample sizes are needed to detect rare occurrences.
Multiple vaccinations. Individuals often receive several vaccines at a time or over a short period, which makes it difficult to identify the culprit vaccine in the event of an adverse effect. Controlled safety trials of vaccine combinations would have to include as many study groups as there are combinations of vaccines under study, plus at least one reference group, and thus would require large sample sizes.
Multiple end points. The large number of symptoms potentially associated with vaccination complicates surveillance because the reporting mechanism must allow for numerous symptom categories in addition to as-yet-unreported symptom types.
Lack of symptoms specific to vaccination. Since there is no unique clinical syndrome or laboratory diagnosis associated with vaccination, it is difficult to differentiate whether symptoms, such as fatigue or seizures, are due to receiving the vaccine or to some unrelated factor coincident with vaccination.
Passive reporting systems. Passive surveillance systems are most useful as a sentinel for identifying rare or previously unrecognized side effects of newly marketed vaccines and for monitoring the safety of individual vaccine lots. However, these systems do not provide information about the rates of reactions to vaccines. As discussed above, VAERS is a passive system that relies on health care providers, those receiving vaccinations, and others to report health outcomes that may be linked to vaccine exposure in the recent or more distant past. Reporting is likely to depend on the gravity of the effect, the time lapsed since exposure, and the diligence in symptom reporting by the patient’s health care workers. Thus, underreporting is an inherent issue. Furthermore, supporting information (e.g., laboratory results) to infer causality may be inaccurate or missing.
High vaccination rates. For widely administered vaccines, it is difficult to find a comparable control group that has not received the vaccine. Unvaccinated individuals constitute a small, highly selected group that may differ from those vaccinated in other aspects and, thus, are not generally suitable as a control group. Further, their small number is unlikely to allow for the study of background rates of rare medical events.