Spatial Analysis of Antiretroviral Therapy Attrition Among Adults in Zimbabwe HIV: Geo-Additive Bayesian Survival Models

This study aims to describe the spatial heterogeneity of antiretroviral therapy (ART) attrition and identify the hot spots for ART attrition and its correlates before the “treat all” strategy to come up with interventions that strengthen ART retention in Zimbabwe. Secondary data analysis was conducted using individual-level data. A Bayesian geo-additive survival model was utilised, which simultaneously models the non-linear functions of numeric covariates and the baseline time with the spatial effects at the district level adjusting for the fixed effects. The percentage of ART attrition was 30.6% (n = 114,022). The risk of attrition increases with an increase in the number of years on ART. Those enrolled at a provincial/referral (risk ratio (RR) = 2.25; 95% credible interval (CrI): 2.211 to 2.301) or district/mission (RR = 2.5; 95%CrI: 2.394 to 2.63) (reference:primary health care) and tuberculosis-infected patients (RR = 3.589; 95%CrI: 3.291 to 3.911) had an increased risk of ART attrition. The 20-year-olds had the highest risk of ART attrition. There was a distinct structured spatial variation in ART attrition along the Beitbridge-Harare band. Differentiated adherence counselling for adolescents, implementation of strategies for managing patients with tuberculosis coinfection and efficient tracking of LTFU clients are crucial to minimise ART attrition; hence, optimising the benefits of the HIV “treat all” strategy.