### Department of Biostatistics Seminar/Workshop Series

# **Vaccine Efficacy from the Perspective of a Virus**** **

**
**## Dean Follmann, PhD

##### Assistant Director for Biostatistics, National Institute of Allergy and Infectious Diseases & Chief of the Biostatistics Research Branch

**Vaccine Efficacy from the Perspective of a Virus**
Traditional vaccine efficacy clincal trials use the time to consequential human infection as the primary endpoint. A common method of analysis for such trials is to compare the times to infection between the vaccine and placebo groups using a Cox regression model. With new technology we can additionally record the number of virions that infect cells, say $X$, rather than just the indicator of infection $X>0$. In this paper we develop a unifed approach for randomized placebo controlled vaccine trials that couples the time to infection data with X. We assume that the time from randomization to a potentially infectious exposure follows the same proportional hazards model in both the placebo and vaccine groups. We allow unspecified distributions for X=0,1,2,... in the placebo and vaccine groups which have mean mu, mu D respectively. Thus D is a measure of vaccine efficacy which complements the more traditional measure based on Cox regression. If X follows a mixed Poisson model with an unspecified mixing distribution, D can be viewed as the (multiplicative) reduction in the probability of infection for a single virion encountering a vaccinated human compared to a single virion encountering an unvaccinated human. We develop fully parametric and semi-parametric methods of estimating D. Interestingly; we can recover D, which is based on the Xs from all potentially infectious encounters, even though potentially infectious encounters that result in no infection are unknowable. Put another way, we can recover the ratio of untruncated means of X even though only truncated X data is observed. Simulations of clinical trials of candiate HIV vaccines show that the method can reliably recover D in realistic settings. We further show that incorporating X in HIV vaccine trails may be 25% more efficient than traditional approaches.
October 30,20103 MRBIII Room 1220
-- AudreyCarvajal - 21 Oct 2013 **
****
****
** **
** **
**

** Copyright © 2013-2022 by the contributing authors. All material on this collaboration platform is the property of the contributing authors. **

Ideas, requests, problems regarding Vanderbilt Biostatistics Wiki? Send feedback**
** **
** **
** **
**