Prospective evaluation of a model of risk for HIV infection among injecting drug users.

Abstract

Data on 3016 out-of-treatment injecting drug users (IDUs) were analyzed in order to replicate findings from an earlier study on risk factors for HIV infection (Iguchi et al., 1992) and evaluate a model for estimating probability of infection. Logistic regression analyses yielded a set of risk factors highly consistent with previous findings. A logistic function was used to estimate subjects' probabilities of infection, and these estimates were strongly correlated with actual HIV prevalence in both the original and current samples. The current study represents a successful replication of earlier findings and supports the validity of the risk model.