An improved understanding of the natural progression of the infection with Human Immunodeficiency Virus (HIV) and its relationship to infectiousness is important for devising treatment and control strategies. Clinicians need to decide how closely to monitor infected patients and when to initiate anti-viral or prophylactic treatment. The purpose of this grant application is to develop statistical methodology that makes use of many different sources of information to study progression and infectiousness. Although repeated measures of biological markers of disease severity are available for HIV-infected people enrolled in prospective studies, analysis of these data is complicated by the fact that time of infection is generally unknown, and follow-up time is generally shorter than the median latency between infection and onset of AIDS. Therefore, standard methodology is not adequate to describe progression throughout the entire latency period. New methods for analyzing repeated measures are required to incorporate external information on the distribution of markers in uninfected people and the distribution of infection-time in the infected. A similar methodologic development is required to use estimates of the latency distribution in such analyses. Estimates of these distributions are available from retrospective analyses of blood donors, homosexually-active men, and hemophiliacs -- groups that are comparable, in important characteristics, to people enrolled in prospective cohorts. Information from different sources can be combined by using empirical Bayes techniques; one approach is to treat the time of infection or latency period as a random effect from a distribution that is either known or can be estimated. New statistical methods are also required to characterize, from studies of sexual partners, the variability in infectiousness of HIV and its relationship to stage of disease. The goal of this research is to determine whether infected people tend to become progressively more infectious over time and/or whether some individuals are highly infectious throughout their latency period. Imputation of infection-time, based on empirical Bayes methods, may also be useful in this research. The general class of problems addressed in this proposal arises in studies of other retroviruses, and in settings in which follow-up time is short compared to average duration of the process of interest.
