Research Strategies for Assessing Adverse Events Associated with Vaccines: A Workshop Summary (1994)

Vaccines undergo premarketing clinical trials similar to those forpharmaceutical products. The purpose of phase I trials is to evaluatethe basic safety and immunogenicity of a vaccine. Such trials aredone on a small scale and identify only very serious or very commonadverse events. Phase II trials are larger and garner more informationon safety and immunogenicity.

Most safety information comes from phase III trials; the data collectedin this phase are generally of higher quality and greater quantitythan those collected in the first two phases. Some aspects of safety,for example, local reactions, are usually assessed; some trials arelarge enough to look at more serious and less infrequent adverseevents, such as SIDS, emergency room visits, hospitalization, andthe like. FDA encourages manufacturers to communicate with the FDAoften during the process. When planning phase III trials, FDA helpsinvestigators to consider the power of the study, adverse eventsto be considered, and the like. FDA ensures that statisticians areinvolved in the oversight of phase III study protocols. In addition,FDA prefers randomized, blinded studies whenever possible; this,however, is not always feasible. Some of the types of controls usedin trials include the following: vaccine is given on odd-numbereddays; control children receive vaccine at a later date than the othergroup; control or vaccinated status is assigned randomly. Trial studydesigns become increasingly complex as more vaccines are combined.FDA is currently developing “points to consider” when designing studies; these willbe disseminated when they are complete.

There are no uniform requirements for postmarketing surveillanceof vaccines as such. Manufacturers are, however, required to forwardto the FDA any adverse event reports that they receive. More recently,manufacturers are being asked to agree, while still in the licensingphase, to carry out postmarketing surveillance, particularly prospectivepostmarketing studies of rare adverse events.

Participants expressed the opinion that, with all the limitationsand problems of passive surveillance and LLDBs, as discussed earlier,the premarketing period is the best time to obtain data on the safetyof vaccines. However, premarketing studies conducted to support licensureoften use sample sizes more appropriate for studying efficacy andcommonly occurring adverse events and usually do not study sampleslarge enough to detect rarely occurring adverse events.

Participants expressed frustration with the lack of randomizationin vaccine trials. The results obtained from nonrandomized trialsare difficult to interpret. This is particularly important for pre-marketingstudies, because it is almost impossible to do randomized trialspost-licensure. Postlicensure research usually is of a cohort orcase-control design.

A participant commented that trials of therapeutic products almostalways include randomization. Manufacturers were questioned why thiswas not the case for vaccine trials. They responded that in general,prevention trials usually require much larger sample sizes and longertime periods for their completion than treatment trials and thatthe cost is correspondingly greater. In addition, it is difficultto get sufficient people to accept randomization in vaccine trialsfor a period sufficient to detect rare adverse events (this is discussedin a subsequent section).

The example of the Urabe-strain mumps vaccine was cited to illustratethe long period needed to confirm a causal relation. The first caseof aseptic meningitis was reported only after 7 million doses ofthe vaccine had been distributed, and it took 5 years after causalitywas suggested to estimate the frequency of occurrence. A randomizedcontrolled trial of sufficient magnitude to have detected this adverseevent would have been prohibitively costly.

A participant cited the example of an expensive 40,000-subject studyof hepatitis A vaccine that resulted in what some considered to belittle additional information, that is, the trial confirmed clinicalsuspicions that the vaccine was safe. Some participants stressedthat confirmation of the safety of a vaccine is worth the resourcesnecessary to do a definitive study. Even if the trials that are performedare not large enough to detect very rare adverse events, randomizationwill help interpret the safety data that have been collected.

The importance of using standardized forms to record adverse reactionsin clinical trials was stressed. Currently, there is great variabilityin the ways in which adverse reactions are assessed in clinical trials.It was also thought that streamlining of forms would be helpful;better information might be obtained if fewer questions were askedand emphasis was placed on the quality of answers to those few questions.Identification of the most relevant questions would not be a trivialmatter, however.

A participant agreed that it is necessary to standardize the reportingof particular conditions, but questioned how that could be accomplishedand how the information could be disseminated to those who do thereporting. The use of a detailed text sent with the reporting formsis a possibility, but that could be a somewhat unrealistic possibility.