Discussion: Vaccine Effect on Susceptibility

The three papers by Drs.Pugh, Albert and Rida represent a diverse spectrum of problems in evaluating HIV vaccines.I enjoyed reading them and am thankful to the organizers for this opportunity to put them briefly into historical and systematic perspective.Vaccine efficacy or effectiveness is generally measured as VE = 1 -RR, where RR is some measure of relative risk.The interpretation of the vaccine efficacy measure depends on the outcome of interest, the choice of parameter, the choice of comparison groups, whether the vaccine effect on susceptibility or infectiousness is of primary interest, and possible heterogeneities in the population.Characteristics such as waning or boosting of protection, as well the indirect effects of a vaccination programme could also be of interest.Marian Pugh's paper on confidence intervals for vaccine efficacy field trials' is motivated by the three-prong decision possibility of intermediate size phase 111 field trials.The three decision choices are either to discard the vaccine candidate, to continue the vaccine trial, or to apply for licensure.The outcome in her paper is infection.The vaccine effect of interest is the relative reduction in susceptibility.The measure of relative risk is based on infections per person years at risk in the vaccinated compared to the unvaccinated groups under the assumption of independent binomial outcomes.This is an unconditional measure of vaccine efficacy' because it does not require information about the contacts made between infectives and susceptibles.This parameter is not specific to infections diseases, but is commonly used in non-infectious disease settings as well.Her calculations are, therefore, applicable beyond the estimation of vaccine efficacy.There are, however, certain assumptions in the use of the person-time measure that are specific to infectious diseases.Since the person-time measure of vaccine efficacy does not condition on a specified exposure to infection in the comparison groups, interpretation of the vaccine efficacy estimate as being due to the difference in susceptibility induced by the vaccine relies on the assumption of equal exposure to infection.This condition was one of three requirements for a valid vaccine efficacy estimate specified by Greenwood and Yule in 1915.3Dr. Pugh is simulating a randomized trial.Randomization has at least two purposes in vaccine field trials.Randomization is supposed to ensure that exposure to infection is equal in the comparison group^.^It is also supposed to ensure that the groups are in all other material respects comparable.Assuming exchangeability of the vaccinated and unvaccinated groups for the duration of a vaccine trial, however, is pr~blematic.~Dr. Pugh compared four different methods of confidence interval calculation: (i) Bayes analysis with independent beta prior; (ii) asymptotic normality of the ratio of the log of risks; (iii) the exact CCC 0277