Monte Carlo simulation experiments for analysis of HIV vaccine effects and vaccine trial design

Abstract

The field of infectious disease epidemiology has increasingly adopted stochastic simulation technologies to simulate complex infectious disease transmission systems. Such simulations have both increased the scientific understanding of infectious disease transmission dynamics and served as important tools for evaluating epidemiologic study designs and statistical methods. This paper reports on a discrete-event simulation to analyze the recently developed Retrospective Partner Trials (RPT) HIV vaccine trial design. A specially designed simulation system, HIVSIM, was used to simulate data resulting from the RPT design vaccine trials. HIVSIM explicitly models complex HIV transmission dynamics (e.g., sexual partner mixing patterns and concurrent sexual partnerships) and vaccine trial design characteristics. Monte Carlo simulation analyses conducted with HIVSIM indicate that the RPT design is able to produce vaccine effect estimates with acceptably small bias, high precision and excellent statistical power under plausible HIV vaccine trial conditions. Additionally, the explicit simulation of HIV transmission dynamics permits investigations into the common, but unwarranted, statistical independence assumptions routinely used in the estimation of vaccine effects.