Evaluating methods to quantify the effect of deprescribing a medication on patient clinical outcomes using a simulation study

In a Monte-Carlo simulation of deprescribing with time-dependent confounding and competing risks, IPTW Cox regression and IPTW sequential Cox regression were the only methods achieving nominal coverage and unbiasedness.

• Why it matters: Deprescribing questions using electronic health records can yield biased estimates when time-varying confounding, early adverse events, and time-dependent effects are not handled.
• What they did: The study simulated synthetic datasets under multiple data-generating mechanisms and evaluated seven estimation approaches, including Cox regression, IPTW Cox regression, sequential Cox variants, extended Cox regression, and the parametric g-formula.
• The result: Only IPTW Cox regression and IPTW sequential Cox regression met nominal coverage and unbiasedness across the scenarios tested, supporting their use for target trial emulation in this setting.

AI summary of the abstract · may contain errors · not medical advice · read the paper.