Sources of uncertainty in estimating HIV infection rates by back-calculation: an application to Italian data.

The effects on back-calculated HIV diffusion estimates of the uncertainties in a number of factors that may potentially influence the estimation process are examined and compared. More specifically, the effect on HIV infection estimates of uncertainties in: (i) the sequence of periodical AIDS incidence cases; (ii) the estimation of the AIDS incubation period, including the influence of therapeutic treatment on the same; and (iii) the choice of a particular back-calculation method, including the stochastic uncertainty associated with each method, are examined. The choice of one particular back-calculation model over another, among three different methods considered, seems to be a minor component of the uncertainty, when compared to the uncertainty, due to all other sources. HIV incidence and prevalence in Italy for different risk categories are estimated.