Outbreaks in highly vaccinated populations: implications for studies of vaccine performance.

Most of the factors associated with the failure of a vaccination to provide protective immunity are not distributed uniformly or randomly within populations. This paper explores the extent to which a nonrandom distribution of vaccination failures and the selection of exceptional situations for investigation may influence estimates of vaccine performance. The authors show that outbreak investigations will tend to underestimate vaccination efficacy, and that the extent of underestimation will be related directly to the size of the epidemic triggering an investigation, the vaccination coverage in the community, and the extent of clustering of vaccination failures in the population; it will be related inversely to the size of and contact intensity within the investigated community. These potential sources of bias are not the only problems that arise in estimating vaccine efficacy, but they should be taken into consideration when analyzing and interpreting outbreak situations. The fact that outbreak investigations carried out within the United States during the past decade have provided estimates of measles vaccination efficacy on the order of 95% is consistent with a somewhat higher overall "true" efficacy of current vaccines and procedures in the total population. It is important to understand better the frequency, distribution, and risk factors for vaccination failures in populations.