Abstract Body

Retention in care (RIC) and sustained viral suppression are key program indicators and prerequisites to achieving good health for people with HIV. Estimates of these indicators are susceptible to bias from poor record keeping and transfers from care. We examined RIC among adults aged ?15 years in uMkhanyakude District who initiated antiretroviral therapy (ART) in 2013-2017.

Individuals were considered retained if they had a clinic visit recorded in the electronic ART database (TIER.net) in the past 6 months, or were recorded as having transferred out. We used Kaplan Meier methods to estimate RIC based on TIER.net alone. We then corrected estimates of RIC to account for poor record keeping or clinic transfers using information from two data sources: drug levels and viral load on dried blood spots from an annual HIV serosurvey conducted in the area and viral load data from a community-based multimorbidity study.

3202 (79% female) individuals who initiated ART during 2013-2017 were included in the analysis. Based on TIER.Net data alone, RIC at 5 years was 60.4% (95%CI=57.8-62.9%), and slightly lower in women than men (59.5%, 95%CI 56.5-62.3% vs 63.9%, 95%CI 58.4-68.8%). After correcting for community-based data collection, RIC at 5 years was 89.5% (CI=87.4-91.6%). Risk of death was 2.2%, based on TIER.net alone, vs 5.2% using mortality data from community-based data.

TIER.net registers appear to vastly underestimate RIC and underestimate mortality for people with HIV on ART. Development and implementation of a National Electronic Health Record System and/or upgrading the TIER.net system to include cross-clinic communication about patient transfers and/or consolidating individual health records is likely to improve patient care and promote more valid estimates of key clinical indicators to optimize allocation of resources.