Infectious disease outbreaks and the continuing HIV/AIDS epidemic require novel interventions to prevent transmission and effectively treat individuals with these diseases. Because the treatment status of one individual can affect the outcome of another in infectious disease trials, innovative clinical trial designs are needed to study these interventions. Cluster-randomized trials (CRTs), both parallel-arm and stepped-wedge CRTs (SW- CRTs), are often used to assess these interventions because of their desirable statistical and logistical properties. Methods to determine the required sample size for stratified CRTs and determine the benefits of stratification in terms of increased power, however, are poorly developed and require unreasonable assumptions. In addition, methods of analysis of SW-CRTs either make assumptions that are inappropriate for infectious disease contexts or are severely underpowered. Invalid or underpowered studies waste precious public health resources and lead to inconclusive results that slow the development of effective interventions. This proposal addresses these gaps by proposing biostatistical methods for designing clinical trials for HIV/AIDS and other infectious diseases. Specifically, the research will develop easy-to-use, flexible formulae for determining the required sample size for stratified CRTs with binary outcomes and for determining the reduction in required sample size due to stratification compared to a comparably-powered unstratified CRT. Additionally, the research will develop a new analysis method for SW-CRTs that is appropriate for outbreak settings and that yields more interpretable treatment effect estimates than existing methods. The value of these methods will be demonstrated by using data from two trials, one completed and one currently being planned, of interventions to prevent tuberculosis in high-risk, HIV-positive individuals. To ensure that they are used to improve public health research, these methods will be published in journals that reach both statistical and public health audiences, and the software developed will be made available to investigators planning clinical trials of novel interventions. The proposed research will be conducted at the Harvard T.H. Chan School of Public Health, a site of collaboration between basic science researchers, biostatisticians, and clinical investigators that encourages the development of such high-impact methods. Working with professors from the Departments of Biostatistics and Epidemiology, the Center for Biostatistics in AIDS Research, and the Center for Communicable Disease Dynamics, the trainee will learn how to conduct biostatistical research that addresses major public health challenges. He will also develop his written and oral communication skills, learn the manuscript drafting and publication process from experienced researchers, improve his computational skills, and increase his effectiveness as a teacher of statistics. The fellowship will thus prepare the trainee for a productive career as a biostatistics professor and help him develop a research agenda that is statistically rigorous and relevant to pressing public health needs.
Cluster-randomized trials (CRTs) are one of the most appropriate and feasible ways of assessing the effectiveness of interventions for the prevention and treatment of HIV/AIDS and other infectious diseases. In order to plan, size, conduct, and analyze such trials, investigators must consider the unique design features of cluster randomization and infectious disease dynamics. This proposal will allow the development of methods to ensure that CRTs yield statistically valid estimates of the effect of intervention and are adequately powered to use public health resources most efficiently.
