Joint modeling of HCV and HIV from cross-sectional serological data

Abstract

In this study co-infection of HCV and HIV is investigated by joint modeling of the two infections using serological data from Italy and Spain. The seroprevalence and force of infection of the diseases are estimated over exposure time using the alternating regression model (ALR) and shared random effect models. The marginal (ALR) and random effect models are fitted with the logic and complementary log-log(clog-log) links. On the basis of the AIC values, Weibull model was chosen as the best fitting model. Significant co-infection of HCV and HIV is observed. Known risk factors such as sharing of syringes and age at first injection were confirmed as risk factors.