Objective: To derive and internally validate a clinical prediction rule for virologic response based on CD4 cell count increase after initiation of HAART in a resource-limited setting.
Design and methods: A retrospective cohort study at two HIV care clinics in Gaborone, Botswana. The participants were previously treatment-naive HIV-1-infected individuals initiating HAART. The main outcome measure was a plasma HIV-1 RNA level (viral load) < or = 400 copies/ml (i.e. undetectable) 6 months after initiating HAART.
Results: The ability of CD4 cell count increase to predict an undetectable viral load was significantly better in those with baseline CD4 cell counts < or = 100 cells/microl [area under the ROC curve (AUC), 0.78; 95% confidence interval (CI), 0.67-0.89; versus AUC, 0.60; 95% CI, 0.48-0.71; P = 0.018]. The sensitivity, specificity, and positive and negative predictive values of a CD4 cell count increase of > or = 50 cells/microl for an undetectable viral load in those with baseline CD4 cell counts < or = 100 cells/microl were 93.1, 61.3, 92.5 and 63.3%, respectively. Alternatively, these values were 47.8, 87.1, 95.0 and 24.5%, respectively, if a increase in CD4 cell count of > or = 150 cells/microl was used.
Conclusions: CD4 cell count increase after initiating HAART has only moderate discriminative ability in identifying patients with an undetectable viral load, and the predictive ability is higher [corrected] in patients with lower baseline CD4 cell counts. Although HIV treatment programs in resource-constrained settings could consider the use of CD4 cell count increases to triage viral load testing, more accurate approaches to monitoring virologic failure are urgently needed.