Background: While much attention has focused on CD4 count predicting clinical outcomes of antiretroviral therapy (ART), less attention has been paid to its role in predicting loss from HIV care. Baseline CD4 count has been mildly associated with attrition, but we hypothesized time updated CD4 count would be a better predictor of dropping out of ART care. Methods: We used data from Themba Lethu, a large ART clinic in Johannesburg, South Africa (>25,000 patients ever initiated). We included all ART-naïve adult patients initiating treatment between April 2004 and December 2011. We defined loss as being ≥3 months late for a scheduled appointment with no subsequent visit. Current CD4 count was defined in 6 (+/- 2) month intervals from ART initiation. Hazard ratios (HR) and 95% confidence intervals (CI) were estimated using Cox proportional hazards regression. Models were adjusted for sex, age, and baseline predictors of loss (p<0.2). Results: 17,177 patients were included in our analysis. Patients were predominantly female (61.4%) with low median (IQR) baseline CD4 count (94; 34-168). 25.1% (4305/17177) of patients were lost in a median of 14.0 person-months (rate: 6.96/1000 person-months). In both unadjusted and adjusted models using baseline CD4, those with a CD4 <50 at ART initiation were at no increased risk of attrition vs. those with a CD4 >200 (aHR 0.95; 95%CI: 0.84-1.08)), even after adjusting for predictors of attrition. However, those with a time updated CD4 count <50 (aHR 2.68; 95%CI: 2.17, 3.31), 50-100 (aHR 1.84; 95%CI: 1.46, 2.31) and 100-200 (aHR 1.55; 95%CI: 1.29-1.88) were at substantially increased risk of loss vs. >200. Conclusions: While baseline CD4 count was a poor predictor of loss from ART programs, time updated CD4 count was a strong predictor. Monitoring patients who continue to have low CD4 counts over time on ART could help target patients at high risk for loss.
Conference: SA AIDS 2013, Durban, SA