Iterative time distribution matching for target trial emulation

Purpose of review Immortal time bias (ITB) is a persistent threat to validity in observational studies, especially when treatment initiation occurs after cohort entry. Although widely recognized, available analytic and design-based approaches differ markedly in assumptions, feasibility, and vulnerability to residual bias or information loss. With the growing reliance on target trial emulation of real-world evidence to inform clinical and policy decisions, a concise synthesis of existing methods and their limitations is timely, together with evaluation of newer approaches that correct ITB without compromising cohort integrity. Recent findings The recent literature demonstrates marked heterogeneity in the performance of methods addressing ITB. Naïve baseline assignment consistently biases treatment effects, while exclusion-based designs and landmark analyses produce unstable, design-dependent estimates. Time-dependent Cox models and clone-censor-weighting approaches generally attenuate bias but depend on strong modelling assumptions. Prescription Time Distribution Matching (PTDM) improves temporal alignment but excludes untreated individuals when pseudo-treatment times exceed observed follow-up. Iterative Time Distribution Matching (ITDM), an extension of PTDM, resolves this limitation by repeatedly sampling and constraining pseudo-treatment times within each individual's valid risk period, retaining a fuller cohort and recovering additional events. We demonstrate these differences by applying each method to the Stanford Heart Transplant dataset, illustrating how methodological choice can entirely flip the perceived efficacy of a life-saving intervention. Summary ITB is best addressed through design-level corrections aligned with target trial principles rather than post hoc modelling alone. ITDM provides a practical, robust alternative that preserves cohort completeness and enhances causal validity, supporting more reliable effect estimation in clinical research using real-world evidence.