Abstract
Uganda implemented the antiretroviral therapy (ART) optimization program in July 2019, based on an eligibility rule. ART optimization targeted individuals with prior viral load (VL) < 1000 copies/mL, while those with VL ≥ 1000 copies/mL continued with non-optimized regimens. We assessed the effectiveness of ART optimization on VL suppression among children and adolescents with HIV (CAWH) in Uganda. We also assessed the compliance of human immunodeficiency virus (HIV) clinics with the eligibility rule and its effect on ART optimization. Therefore, we designed a quasi-experimental study using data from 21 urban and rural HIV clinics. The exposure was ART optimization, defined as the initiation or transition of CAWH on dolutegravir or a protease inhibitor (boosted lopinavir). Children and adolescents with HIV on an optimized ART regimen formed the exposed group, while those on a non-optimized ART regimen comprised the nonexposed group. The primary outcome was VL suppression, defined by VL < 1000 copies/mL after ≥6 months of ART optimization. We assessed the effectiveness of ART optimization on VL suppression using 2-stage least squares instrumental variable regression due to imperfect compliance with the eligibility rule across the clinics. We also established the effectiveness of the eligibility rule on ART optimization for individuals just below and just above the cutoff. Sensitivity analysis was performed using a noncausal approach. We analyzed data from 2999 CAWH aged ≤19 years and found an overall VL suppression of 76.1% (2282/2999). We found that ART optimization showed a trend toward improved VL suppression (risk ratio [RR] 1.81, 95% CI: 0.79-4.14). However, compliance with the rule was only for 2.6% of the participants, and the rule did not improve ART optimization (RR 0.96, 95% CI: 0.88-1.05). Overall, ART optimization, guided by an eligibility rule, did not achieve the target of ≥95% VL suppression among CAWH across the 21 public HIV clinics in Uganda, partly due to low compliance with the rule, although it showed a trend toward improvement. Addressing context-specific biological, behavioral, social, and structural barriers is needed to optimize VL outcomes.