Outcomes of patients lost to follow-up after antiretroviral therapy initiation in rural north-eastern South Africa.

Objective The vital status of patients lost to follow-up often remains unknown in antiretroviral therapy (ART) programmes in sub-Saharan Africa because medical records are no longer updated once the patient disengages from care. Thus, we aimed to assess the outcomes of patients lost to follow-up after ART initiation in north-eastern South Africa. Methods Using data from a rural area in north-eastern South Africa, we estimated the cumulative incidence of patient outcomes (i) after treatment initiation using clinical records, and (ii) after loss to follow-up (LTFU) using data from clients that have been individually linked to Agincourt Health and Demographic Surveillance System (AHDSS) database. Aside from LTFU, we considered mortality, re-engagement and migration out of the study site. Cox proportional hazards regression was used to identify covariates of these patient outcomes. Results Between April 2014 and July 2017, 3700 patients initiated ART and contributed a total of 6818 person-years of follow-up time. Three years after ART initiation, clinical record-based estimates of LTFU, mortality and documented transfers were 41.0% (95% CI: 38.5-43.4%), 1.9% (95% CI 1.0-3.2%) and 0.1% (95% CI 0.0-0.9%), respectively. Among those who were LTFU, the cumulative incidence of re-engagement, out-migration and mortality at 3 years were 38.1% (95% CI 33.1-43.0%), 49.4% (95% CI 43.1-55.3%) and 4.7% (95% CI 3.5-6.2%), respectively. Pregnant or breastfeeding women, foreigners and those who initiated ART most recently were at an increased risk of LTFU. Conclusion LTFU among patients starting ART in north-eastern South Africa is relatively high and has increased in recent years as more asymptomatic patients have initiated treatment. Even though this tendency is of concern in light of the prevention of onwards transmission, we also found that re-engagement in care is common and mortality among persons LTFU relatively low.