Vaccine Safety Forum: Summaries of Two Workshops (1997)

Adverse events associated with vaccines are currently detected through the Public Health Service's Vaccine Adverse Event Reporting System (VAERS), the Centers for Disease Control and Prevention's (CDC's) Large Linked Data Base (LLDB),³ and through surveillance measures undertaken by vaccine manufacturers. The federal agencies primarily responsible for responding to the adverse events associated with vaccines are the Food and Drug Administration (FDA) and CDC.

Under the aegis of CDC and FDA, VAERS was established in 1990. VAERS requests physicians to report any adverse events observed following vaccination of their patients. Health care providers are required by law⁵ to report specific serious events occurring within 30 days after vaccination with a particular set of vaccines, as outlined in a published reportable events table. Patients, parents, or even nonrelatives who witnessed or who are aware of a reaction can also submit a report to VAERS. Although some VAERS reports can include symptoms that occurred years after a vaccine was given, most VAERS reports are of adverse events tightly linked in time to vaccination—usually occurring within a few days after the vaccination.

The narrative description of an adverse event reported on a VAERS form may include a discussion of the symptoms, the time course, the laboratory tests that were performed, and the treatment provided for the adverse event. This description is coded and entered into a computer database. Adverse events are also categorized as to whether they were fatal, serious but nonfatal (i.e., those that were life-threatening, required or prolonged hospitalization, or involved permanent disability), or less serious.

VAERS is a passive surveillance system, and thus, adverse events are likely to be underreported. In addition, it lacks much of the information needed for an epidemiologic study to assess whether a specific vaccine caused a specific adverse event. It can, however, generate signals of potential new adverse events

(e.g., hair loss following receipt of hepatitis B vaccine) or a change in the rate of reporting of previously known adverse events associated with vaccines (e.g., decreased numbers of reports of both serious and less serious events after receipt of the acellular pertussis vaccine versus receipt of the whole-cell vaccine. Confusing peaks in the reporting of adverse events can also stem from recent publicity about an adverse event associated with a vaccine. It is difficult to ascertain whether these peaks in adverse event reports are due to a true increase in adverse events related to the vaccine or result only from increased awareness of a potential problem. Spurious peaks in VAERS reports may also occur from changes in the background rate of adverse events. A recent study has shown, for instance, that an increase in VAERS reports of Guillain-Barré syndrome (GBS) during the 1993–1994 influenza vaccination season was due largely to an increase in the rate of GBS in the general population (Lasky, 1997).

To improve detection of adverse events associated with vaccination and to conduct more active surveillance studies to assess the causality of adverse events, CDC established LLDB in 1990. This database is a CDC-coordinated linkage of large databases of four large health maintenance organizations. Comprising automated data from more than 500,000 children (ages 0 to 6 years) in Oregon, Washington, and California, LLDB enables researchers to link vaccine exposures to medical outcomes and thereby to estimate the rates of occurrence of adverse events following vaccination as well as the background incidence. An elevated rate following vaccination, in comparison with the rate at other times, would suggest a possible causal relation to vaccination. Over the next 5 years, LLDB will also be expanded to include adolescents and adults. Unlike VAERS, LLDB contains all of the information necessary for epidemiologic studies of vaccine adverse effects (vaccination exposures, outcomes among vaccinated and unvaccinated person times, and potential confounders).

The large size of LLDB improves the ability to detect rare adverse events, and extensive quality control of the clinical and immunization data in the system increases its reliability. LLDB also allows for the study of confounding variables and for the exploration of groups at risk for adverse events following vaccination.

Vaccine manufacturers, which share a responsibility for detecting adverse events associated with vaccination, described their procedures as follows. Adverse event reports are reviewed by a team representing safety surveillance, clinical research, regulatory affairs, and quality control. The review team examines events individually and with an eye to repeated reports. The review committee may use a threshold level of, for example, 10 reports during a vaccine's lifetime as a signal that follow-up is needed. Manufacturers also monitor the pattern of adverse event reports for particular lots or batches of vaccine over time. Any atypical patterns are carefully reviewed. The entire surveillance process is carried out in close collaboration with and regular reporting to FDA, and thereby VAERS.

In addition to conducting periodic in-house investigations of vaccines for which a company and VAERS have received large numbers of adverse event reports, local investigations are sometimes necessary. An evaluation of the large numbers of adverse events reported from immunization clinics in South Africa, for example, revealed that the vaccines were improperly administered and that some of the reactions were labeled inaccurately.

Manufacturers noted that postmarketing reporting is voluntary and consequently represents incomplete reporting of the true numbers of adverse events that are actually occurring in the population. Furthemore, these reports are coded in the terminology used by the reporter, and there are no standard case definitions for various adverse events. If a health care provider reports a case of encephalitis, for example, it is recorded as a case of encephalitis.

Because of these limitations and because the standard premarketing clinical trials are not sufficiently large to detect rare adverse events, manufacturers also conduct postmarketing surveillance studies. These studies either search for a particular adverse event in a specific population or scrutinize a large number of vaccines for general safety issues. As an example, an epidemiologic analysis of selected reported neurologic adverse events associated with the administration of the hepatitis B vaccine was presented. The manufacturer reported that, although a causal association could not be entirely ruled out, it found no evidence from the analysis of postmarketing reporting data to suggest that Guillain-Barré syndrome, myelitis, transverse myelitis, multiple sclerosis, optic neuritis, or peripheral neuropathies were causally associated with the administration of the hepatitis B vaccine.

FDA monitors adverse event reporting rates for individual vaccine lots on the basis of crude estimates of lot size and the amount of time that the vaccine lot has been in distribution. The information available to the FDA includes VAERS data and manufacturers' reports, as described above. At weekly meetings, FDA personnel review all the data for lots for which a report of a serious adverse event (any event that requires hospitalization, results in permanent disability, is life threatening, or prolongs hospitalization) has been made during the past week. At these meetings the individual reports are discussed, as are the patterns of reporting, that is, the number of reports made during a specific time frame. Particular scrutiny is given to lots for which there have been relatively large numbers of adverse event reports early in their circulation.

At FDA, further investigation of lots with which adverse events have been associated includes a detailed review of individual reports to look for clusters of similar cases, syndromes that connect the reports, or other patterns. Additional information is sometimes collected from the reporter, primary physician, or medical examiner. The adverse event reporting patterns for other final lots filled from the same bulk vaccine are also considered. Of greater concern is if elevated adverse event reports are received for several final lots from the same batch bulk material than if the number of reports for one lot is higher and that for the others is lower. Lot release data are reviewed to ensure that the lots in question have passed all the required tests.

According to FDA representatives, reports of fatal events and reports of certain serious events are routinely followed up by FDA physicians or by VAERS personnel. Some consumer representatives, however, expressed discontent with the followup that they are aware of. Follow-up includes assessment of the current condition of the patient for nonfatal reports, confirmation of the information provided by the reporter, and requests for additional information, including a more detailed chronology of events and relevant facts about the child or family of the child who experienced the adverse event. Sometimes, additional laboratory results, hospital discharge summaries, or autopsy reports are requested. Sometimes the patient or family is contacted directly; at other times the medical records suffice. Reports of less serious events are occasionally investigated for unusual events or patterns.

Once the review process is complete, FDA has several options. The agency may determine that the vaccine lot in question is not likely to be faulty but will