Objective: To investigate the dependence of the hazard of symptomatic AIDS on various markers using a non-parametric method. The markers we consider are measures of time (time since infection and calendar date), measures of immune function (numbers and percentage of CD4 T cells) and serological activation markers (neopterin and beta 2-microglobulin).
Methods: We adapted a non-parametric statistical method to estimate the hazard of AIDS. We considered both univariate analyses, in which each marker was considered separately and bivariate analyses of pairs of markers.
Conclusions: Using data from 356 seroconverters from the Multicenter AIDS Cohort Study, we found that in the univariate analyses the hazard of AIDS is dependent on all markers, with the strongest dependence for CD4 count and CD4 percentage. In the bivariate analyses we found that the time since infection is of little importance in determining the hazard of AIDS if the CD4 count or percentage are known, and is of minor additional value if one of the serological markers is known. In contrast, we found that both beta 2-microglobulin and neopterin do add some additional information to the hazard of AIDS if CD4 count or CD4 percentage are known.