Short and medium term projections of the AIDS/HIV epidemic by a dynamic model with an application to the risk group of homo/bisexual men in Amsterdam.

Several methods exist for short term projection of the numbers of AIDS cases. Some use extrapolation of empirical curves fitted to data up to a given time, whereas others such as the popular method of 'back projection' or actuarial methods also use information about the process. In this paper we describe a dynamic model based on a distributed modelling technique allowing for variability both in infectiousness and in age distribution of the population at risk. Some model parameters are taken from the literature, others are estimated from AIDS incidence data from the homo/bisexual population in Amsterdam. The model described here simulates prevalence and incidence of HIV infection. We present prediction intervals for two years from January 1990 onwards. We discuss three scenarios based on the estimated model, two of which consider early treatment with anti-viral drugs. Given the model and the state of the epidemic in Amsterdam, early treatment intervention must be combined with very drastic measures for reducing infectivity in order to have any serious impact on the course of the epidemic.