Immune Virologic and Adherence Measures as Time-Dependent Predictors of Loss to Follow-Up

Background: For programs supporting large numbers of patients on ART, loss to follow-up (LTFU) and inadequate adherence pose major obstacles in achieving successful outcomes. In preliminary analyses examining patterns and predictors of attrition, we found that baseline CD4+ cell counts, baseline viral loads and early adherence patterns were associated with overall LTFU. In this evaluation, we further characterized the association of CD4+ cell counts, viral loads (VL), and treatment gaps on LTFU in a time-dependent fashion. Methodology: From 2004-2012, Harvard/APIN received PEPFAR funds (HRSA #U51HA02522) to support and develop a sustainable HIV prevention, care and treatment program across multiple clinical centers in Nigeria. All patient ART-related data were collected and stored electronically. LTFU was defined as ≥60 days since the last scheduled refill date. For the analyses on CD4 counts and VL, we examined LTFU in 6-month increments, evaluating the CD4+ cell counts and VL taken at the prior visit as a predictor of loss in a time-dependent fashion. For the evaluation on large early treatment no LTFU, the gap was defined as a contiguous ≥90 days between refill dates during the first 6 months of treatment (i.e., “+90 gap”). Logistic regression methods were used to evaluate the association between +90 gap and LTFU. Results: Between 2004-February 2011, the Harvard/APIN PEPFAR program enrolled 64,372 previously ARVnaive ART patients. Of those, 35% were eventually LTFU. In evaluating the association of CD4+ cell counts and VL suppression rates in a prior visit for months 6, 12, 18, 24, 30, 36, 42, and 48, we found that CD4+ cell counts and VL at prior visits remained significant predictors of attrition (p<0.05; Figure 1). Additionally, patients with the +90 gap were more likely to become LTFU beyond 6 months than those that did not have the gap (p<0.05). Patients with the +90 gap were less likely to have VL suppression and had a lower CD4 count increase by month 12 versus those without the gap (p<0.05). Conclusions: The majority of analyses on LTFU focus on baseline measures as predictors of attrition. In this evaluation, we were able to show that timedependent CD4+ cell counts and viral loads are continuous predictors of loss at 6-month time points starting at ART initiation extending to month 48 and that early large gaps in treatment were associated with LTFU within the first year of ART initiation. These data highlight additional indicators that should trigger counseling to ensure low rates of attrition.