Background: Systematic reviews have described high rates of attrition in patients receiving antiretroviral therapy (ART). However, migration and clinical transfer may lead to overestimation of attrition. Using a newly linked national laboratory database in South Africa, we assessed system-wide retention in care.
Methods: South Africa''s National Health Laboratory Service maintains a database of all public sector CD4 count and viral load (VL) test results since 2004. We developed an algorithm to link individual lab results using probabilistic matching techniques, creating a national cohort of HIV patients. We analyzed data on all patients initiating ART in 2004 and 2005 (during which time VL were collected at ART initiation) and who had at least one subsequent lab result. We assessed retention in care as time to a patient''s most recent lab result (CD4 or VL), following patients through March 2015. Patients were identified as still in care if their last lab test occurred April 2013 - March 2015. We assessed two retention concepts: (a) system-wide retention including all lab results regardless of testing facility and (b) retention at the initiating clinic, in which lab tests at other facilities were ignored. These two concepts mirror the information available on patient histories from clinic-based and health system-wide perspectives.
Results: We followed 53,880 patients who initiated ART in 2004 and 2005. Eight-year retention at the initiating clinic was 13.1% (95%CI: 12.9-13.4). After allowing for transfers, system-wide eight-year retention increased to 47.3% (95%CI 46.9-47.7)(Figure).

Figure 1 - Effect of patient transfer on retention estimates
[Figure 1 - Effect of patient transfer on retention estimates]

Conclusions: Patient migration and transfer are common throughout sub-Saharan Africa. Whereas prior cohort studies have tracked patients through resource-intensive follow-up, we show the utility of a national laboratory database for passive tracking of patients regardless of where they seek care. These findings have implications not just for measurement, but also potential to improve continuity of patient care in migration populations.