Association of VL parameters with risk of AIDS development after initiation of HAARTa.

aResults are from a Poisson regression model adjusted for the length of follow-up. Unadjusted results are from the bivariate models with the indicated VL parameter as the predictor and AIDS development as the outcome. All VL parameters were dichotomized based on their respective medians. The RRs are for the association of slow decay (as indicated by less than median decay constants and slope) and high (greater than median) cumulative viral load with AIDS development. Adjusted models are multivariate models that included the following covariates: baseline and nadir CD4 count, pre-HAART VL, time to HAART, age at HAART initiation (per 10 years) and time to VL suppression were included as continuous variables.