Short-term predictions of HIV prevalence and AIDS incidence.

Reports of AIDS cases in Amsterdam up to February 1990 were used to make predictions of future cases up to 1993. Two published methods were applied, which make extrapolations from current cases and simultaneously estimate the extent of delay in reporting. The choice of the exact model greatly influenced the predictions, as did predictions for distinct transmission groups. We present results for the homo/bisexual male group, and the total population of Amsterdam. The AIDS case predictions are used to predict the HIV prevalence using the ratio of HIV prevalence to AIDS incidence and through 'back calculation'. We suggest that the ratio is a simple technique that may be used to estimate HIV prevalence. The estimated number of cumulative HIV infected homo/bisexual males in Amsterdam in January 1990 was between 2100 and 4100 in a total of 2200-4600 infected people.