Joint modelling of HBV and HCV infections from cross-sectional serological data

Abstract

The aim of this project is to study the co-infection by regressing marginal association and subject heterogeneity of hepatitis C virus (HCV) and hepatitis B virus (HBV) on behavioral risk factors among drug users within drug treatment centers and prisons in Belgium, using a joint modeling approach that deals with multivariate nature of the response. Using marginal(Bivariate Dale Model (BDM),Bivariate Probit Model(BPM) and Alternating Logistic Regression(ALR)) models, the association measures between HCV and HBV infections estimated at individual level (cluster) in terms of odds ratios and correlation coefficients was regressed against behavioral risk factors. Shared random-effects models that take into account the individual heterogeneity in the acquisition of the infections were fitted as well. The analysis used cross-sectional data from 972 drug users who agreed to participate in a sero-behavioural study between 2004-2005. The results showed that the infections are positively ass