A Primer on Restricted Mean Survival Time in Surgical Research

Time-to-event outcomes, such as postoperative complications, disease recurrence, and mortality, are central to surgical research. Historically, these outcomes have been summarized using hazard ratios from Cox proportional hazards models. However, in many surgical trials, the proportional hazards assumption is often violated, particularly when early postoperative risks differ from long-term outcomes, limiting the interpretability of the hazard ratio.

We conducted a narrative review to present a conceptual framework for using restricted mean survival time (RMST) as an alternative analytical approach.

RMST quantifies the average event-free survival over a prespecified, clinically meaningful follow-up period, offering an absolute measure of treatment effect that can be easily understood by clinicians, patients, and policymakers. Unlike hazard ratios, RMST remains valid under nonproportional hazards, accommodates competing risks, and can incorporate covariate adjustment. Drawing on contemporary applications in cardiac surgery, oncology, and other surgical fields, we illustrate how RMST clarifies complex temporal patterns of risk, complements conventional survival metrics, and supports patient-centered decision-making.

By incorporating RMST into trial design, analysis, and reporting, researchers can enhance the interpretability of findings, facilitate cross-study comparisons, and provide a more transparent assessment of treatment benefits. This conceptual review highlights the practical value of RMST and advocates for its broader adoption to improve the rigor, clarity, and clinical relevance of survival analyses in surgical research.