Virological failure on antiretroviral therapy is routinely defined by a viral load threshold, yet the assays used to measure viral load differ in their lower limits of detection, and those differences propagate into the switching decisions that determine subsequent regimen exposure. the present analysis formalises the switching problem as a target trial and derives the conditions under which a resistance-guided switching rule estimated from observational cohorts identifies the same counterfactual quantity as the trial it emulates. The analysis distinguishes assay-limit-induced censoring from informative monitoring and shows that naive complete-case estimation of the switch effect is biased whenever the probability of an unquantifiable viral load depends on prior regimen history. A target-trial emulation estimator is specified with inverse-probability-of-censoring weights, and its operating characteristics are characterised under a declared variance structure with an intraclass correlation of 0.64, between-cluster standard deviation of 6.0 and residual standard deviation of 4.5. Design calculations indicate that 54 clusters are required to detect a 5 percentage-point effect at 80 per cent power, and power is reported across cluster counts of 8, 16, 24 and 32 and effect sizes of 3, 5 and 8 percentage points over 2000 replications per design cell. The contribution is a formal account of when artificial-intelligence resistance-guided switching rules can be interpreted causally, and a design bound on the cluster numbers such evaluations require.
No takes yet. Share an insight, caveat, or question.
Ndiaye et al. (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: