Estimating vaccine effects from studies of outbreaks in household pairs.

The traditional way to measure efficacy of a vaccine, with respect to reduced susceptibility and reduced infectivity once infected, is to look at relative attack rates. Although straightforward to apply, such measures do not take disease transmission into account, with the consequence that they can depend strongly on the community setting, the duration of the study period, the way participants are recruited into the study and the virulence of the infection. Sometimes they give a very misleading assessment of the vaccine, as we illustrate by examples. Here measures of vaccine efficacy are considered that avoid these defects, and estimation procedures are presented for studies based on outbreaks in household pairs. Such studies enable estimation of vaccine effects on susceptibility, infectivity and transmission. We propose that the vaccine efficacy measures be estimated, without making any assumptions about the nature of the vaccine response, by consistent estimates of bounds for the measures.