Topic: Contributions on independence and distance-based methods Title: Estimation of the AIDS incidence rate in Portugal Authors:  Rita Gaio - University of Porto (Portugal) [presenting]
Julio Silva - University of Porto (Portugal)
Alexandra Oliveira - University of Porto (Portugal)
Joaquim Costa - University of Porto (Portugal)
Abstract: The Human Immunodeficiency Virus (HIV) can evolve to severe illness with the late symptoms referred to as Acquire Immune Deficiency Syndrome (AIDS). In Portugal, the epidemic is monitored by a surveillance system based on cases notification by physicians. This process suffers from reporting-delay and no-reporting, and data on incubation times is scarce. Knowledge of the incidence rate is crucial to evaluate prevention campaigns and plans of action. A lower bound for the HIV incidence rate in Portugal (2000-2012), based on national AIDS surveillance data, is provided. Correction for the reporting delays is obtained by conditional likelihood estimation assuming a Poisson distribution for the number of diagnosed cases. The estimation of the HIV incidence rate is based on the back-calculation method. The number of newly infected cases in any given month is assumed to follow a Poisson distribution. The mean number of new infections per month is estimated from the complete likelihood function, given the number of new diagnoses per month. The incubation period is assumed to be time-dependent and to follow a Weibull or a Gamma distribution. Maximization of the complete likelihood uses a combination of the expectation-maximization algorithm and linear regression. Results from the methodology are tested on simulated and real data.