Primary Submission Category: Target Trial emulation / Application of causal inference
Emulation of a target trial to estimate the per-protocol effect of switching from triple to dual antiretroviral therapy on clinical outcomes in people with HIV
Authors: Sophia Rein,
Presenting Author: Sophia Rein*
Background
Intention-to-treat analyses of randomized trials found similar rates of virologic failure in people with HIV switching from triple to dual antiretroviral therapy (ART) compared with those continuing on triple ART. We emulated a target trial to estimate the per-protocol effect of switching from triple to dual ART on virologic failure.
Methods
To emulate the target trial, we identified eligible individuals in 10 observational cohorts between January 2015 and April 2023. Using a sequential emulation with a potential time zero for each individual during each month, we estimated the 5-year risk of virologic failure via pooled logistic regression with censoring if individuals were nonadherent to their initial strategies. We used non-stabilized inverse probability weights to adjust for baseline and time-varying confounding, and nonparametric bootstrapping with 500 samples to calculate 95% confidence intervals (CIs).
Results
During follow-up, adherence to the initial regimen was about 80% in 2,921 included dual therapy initiators and 76% in 79,565 non-initiators. The estimated 5-year risks (95%CI) of virologic failure were 4.38% (3.06-5.68) in dual therapy initiators and 4.19% (3.95-4.47) in non-initiators (risk ratio: 1.04 (0.74-1.35)).
Conclusion
We estimated that switching from triple to dual ART had little impact on virologic failure. Analyses on differences in the risk of clinical events and death, where evidence from randomized trials is lacking, are ongoing.