Abstract Body

Despite significant progress in addressing HIV/AIDS, children, adolescents and young adults living with HIV (CAYHIV) continue to have poorer outcomes than adults aged ?25 years. Loss to follow up (LTF) in antiretroviral therapy (ART) programs is a major challenge to achieving optimal treatment outcomes and the UNAIDS 2030 target. We aimed to compare site-reported mortality with mortality among patients confirmed LTF who were traced.

We included routine observational data on patients initiating ART between 2004-2017 who were not traced, and data obtained from tracing studies for those who were traced from five countries (Lesotho, Malawi, Mozambique, Zambia and Zimbabwe) in Southern Africa. We used Mantel Haensel rates to estimate the crude survival function of mortality and rate ratios. We used random effects Weibull distribution models to examine factors associated with mortality among CAYHIV confirmed LTF and among routinely observed patients. We define LTF as having no recorded visit in the facility for ?182 days and not known to be deceased or transferred out.

We included 106,927 patients from the routine observational cohort and 680 patients who were selected through stratified random sampling and successfully traced (vital status ascertained). There were 2,506/106,927 (2.3%) deaths observed in the routine data and 62/680 (9.1%) additional deaths from tracing. The overall mortality rate in the first five years on ART was 1.25 (95% confidence interval (CI): 1.20, 1.30) per 100 person-years (py) among routine cohort patients and 6.70 (95% CI: 5.18, 8.67) per 100 py among traced patients. Higher mortality (Table 1) in the observational data was associated with being male, younger age at ART start, shorter duration on ART, and initiating ART after 2006. In the tracing data, younger age at ART start compared to 20-24 years and shorter duration on ART were associated with higher mortality.

We observed far higher mortality among CAYHIV who were lost and traced compared to mortality reported at the facility. Program-level mortality estimates among CAYHIV are likely to be underestimated if there is no additional outcome ascertainment done among those considered LTF.