Multivariate models for predicting progression to AIDS and survival in human immunodeficiency virus-infected persons.

Nine hundred thirty persons enrolled in the US Air Force Human Immunodeficiency Virus (HIV) Natural History Study were evaluated with a standard battery of 30 potential surrogate markers of disease progression. A risk score for predicting progression to AIDS was then calculated for each patient in the cohort by using the four highest-ranking variables from multivariate analysis: percentage of CD4 CD29 cells, anergy status, age, and hemoglobin. For predicting survival, beta 2-microglobulin replaced age in the Cox model. Stratification according to the risk score demonstrated that rates of progression to AIDS and survival were significantly different between risk groups (P < .0001). The novel combination of these markers results in extremely accurate risk scores, which may serve as the basis for the development of true surrogate markers of disease progression.