Utilizing pharmacy refill data to predict loss to follow-up among people living with HIV in Manyara region of Tanzania

Abstract

Abstract Background Achieving optimal adherence and retention in HIV care is essential for sustaining viral suppression. Pharmacy refill records offer an objective approach to assessing adherence in settings where routine viral load testing is limited. This study evaluated pharmacy refill adherence, loss to follow-up (LTFU), and their predictors among people living with HIV (PLHIV) in the Manyara region of Tanzania. Methods We conducted a retrospective cohort analysis of 22,650 PLHIV across five districts using the National CTC-2 database. LTFU was defined as no clinic visit for 180 days or more. We also analyzed a cross-sectional final status field updated by health trackers to distinguish research-defined LTFU from confirmed clinical outcomes. Predictors were evaluated using multivariable logistic regression, and spatial mapping identified geographic disparities. Results The mean pharmacy refill adherence was 84.1%, with 57.9% achieving good adherence (>=85%). In the longitudinal analysis, 32.8% of patients met the research definition for LTFU (>=180-day absence) at some point during the study period. Cumulative LTFU was significantly higher in earlier initiation cohorts (2017-2019) compared to the 2021 cohort (aOR 1.89; 95% CI 1.76-2.02). However, cross-sectional system records, which health trackers update, showed that only 2.9% remained truly lost to care; 65.3% were active at their original clinic, 23.1% had eventually transferred to other facilities, and 6.7% were deceased. In multivariable regression, poor pharmacy adherence was the strongest behavioral predictor of disengagement (aOR 2.04; 95% CI 1.77-2.35). Significant geographic variation was observed, with residence in Simanjiro independently associated with the highest odds of LTFU (aOR 3.60; 95% CI 2.67-4.85). Spatial mapping confirmed a clustering of poor outcomes in districts characterized by nomadic pastoralist livelihoods. Conclusion Pharmacy refill adherence is a potent predictor of disengagement and a practical early-warning indicator. The high rate of silent transfers and district-level disparities, particularly in nomadic hotspots, highlight the need for a national unique patient identifier and mobility-friendly retention strategies. Integrating automated refill alerts into the 90-day tracking window is essential to achieve 95-95-95 targets.