Abstract
Combined antiretroviral therapy (cART) durability and time to modification are important quality indicators in HIV/AIDs treatment programs. This analysis describes the incidence, patterns, and factors associated with cART modifications in HIV patients enrolled in four treatment centers in Asmara, Eritrea from 2005 to 2021. Retrospective cohort study combining data from 5020 [males, 1943 (38.7%) vs. females, 3077 (61.3%)] patients were utilized. Data on multiple demographic and clinical variables were abstracted from patient's charts and cART program registry. Independent predictors of modification and time to specified events were evaluated using a multi-variable Cox-proportional hazards model and Kaplan-Meier analysis. The median (±IQR) age, CD4(+) T-cell count, and proportion of patients with WHO Clinical stage III/IV were 48 (IQR 41-55) years; 160 (IQR 80-271) cells/µL; and 2667 (53.25%), respectively. The cumulative frequency of all cause cART modification was 3223 (64%): 2956 (58.8%) substitutions; 37 (0.7%) switches; and both, 230 (4.5%). Following 241,194 person-months (PMFU) of follow-up, incidence rate of cART substitution and switch were 12.3 (95% CI 11.9-12.8) per 1000 PMFU and 3.9 (95% CI 3.2-4.8) per 10,000 PMFU, respectively. Prominent reasons for cART substitution included toxicity/intolerance, drug-shortage, new drug availability, treatment failure, tuberculosis and pregnancy. The most common adverse event (AEs) associated with cART modification included lipodystrophy, anemia and peripheral neuropathy, among others. In the adjusted multivariate Cox regression model, Organisation (Hospital B: aHR = 1.293, 95% CI 1.162-1.439, p value < 0.001) (Hospital D: aHR = 1.799, 95% CI 1.571-2.060, p value < 0.001); Initial WHO clinical stage (Stage III: aHR = 1.116, 95% CI 1.116-1.220, p value < 0.001); NRTI backbone (D4T-based: aHR = 1.849, 95% CI 1.449-2.360, p value < 0.001) were associated with increased cumulative hazard of treatment modification. Baseline weight (aHR = 0.996, 95% CI 0.993-0.999, p value = 0.013); address within Maekel (aHR = 0.854, 95% CI 0.774-0.942, p value = 0.002); AZT-based backbones (aHR = 0.654, 95% CI 0.515-0.830, p value < 0.001); TDF-based backbones: aHR = 0.068, 95% CI 0.051-0.091, p value < 0.001), NVP-based anchors (aHR = 0.889, 95% CI 0.806-0.980, p value = 0.018) were associated with lower cumulative hazards of attrition. The minimal number of switching suggests inadequate VL testing. However, the large number of toxicity/intolerance and drug-shortage driven substitutions highlight important problems in this setting. Consequently, the need to advocate for both sustainable access to safer ARVs in SSA and improvements in local supply chains is warranted.