We propose to develop new statistical methods for the analysis of hepatitic C (HCV) disease progression as measured by fibrosis scores from liver biopsies, with particular attention to how much human immunodeficiency virus (HIV) co-infection accelerates progression. About 3 millon persons are chronically infected with HCV, including 150,000 to 300,000 who are HIV co-infected. Good estimates of progression rates and factors that influence them are needed for clinical decisions and for projecting public health burden. Although many studies have addressed these issues, estimates vary widely and there are many potential problems with the statistical methods that have been used. In contrast to the substanial effort devoted to measuring and studying HCV progression, the statistical problems involved have been largely neglected. We will develop new methods to address these problems. First, we will develop a discrete-time multi-state modeling that will avoid problems with previous methods while accounting for: measurement error in fibrosis scores, uncertainty about the time of HCV infection, non-linear progression over time, exclusion of persons who progressed so rapidly that they had already died by the time of the study, and a possible tendency for those who progress rapidly through early fibrosis stages to also progress rapidly through later stages. A second method will model observed fibrosis scores as resulting from an unobserved underlying continuous process. Both methods will permit multivariate modeling with covariates that change over time, which will be crucial for assessing the impact of HIV and its treatments on the course of HCV disease. We will also estimate the association of HIV with time of HCV infection and perform other specialized analyses to further assess the influence of HIV. The methods will be applied to at least seven different data sets and will also be evaluated using simulated data sets where the true parameters of interest are known. Hepatitis C is an infection that can lead to fatal liver problems. Taking out a very small piece of the liver and examining it under a microscope is the best way to tell how far problems have developed, and many studies have done this on many people in order to learn about how often and how soon the infection causes problems. Because methods previously used to analyze the data that those studies provide have been prone to inaccuracy, this proposal will develop better methods and apply them to several studies.
