Restricted Mean Survival Time (RMST)

Restricted mean survival time (RMST) is the expected event-free time up to a pre-specified time horizon, equal to the area under the survival curve up to that horizon.

제한평균생존시간(RMST)은 미리 정한 시점까지의 기대 무사건 생존시간으로, 해당 시점까지 생존곡선 아래 면적과 같다.

How do I compare survival between groups when the proportional hazards assumption does not hold?

비례위험 가정이 성립하지 않을 때 집단 간 생존을 어떻게 비교하나요?

Restricted mean survival time compares groups by the average event-free time up to a chosen horizon, reported as a difference or ratio in time units. It fits when hazards are not proportional, curves cross, or an absolute and clinically interpretable summary is preferred over a hazard ratio.

When to use Restricted Mean Survival Time (RMST)

• Survival curves cross or the proportional hazards assumption is violated
• Reporting treatment benefit in time units, such as months of event-free life gained
• Comparing two groups up to a clinically meaningful follow-up time

When to use something else

• Hazards are proportional and a relative effect is wanted: use the Cox proportional hazards model
• Covariate adjustment of many predictors is the focus: use RMST regression or a Cox model

Assumptions and requirements

• The time horizon (tau) is pre-specified and within the follow-up of both groups
• Independent (non-informative) censoring
• Enough subjects remain at risk near the horizon

What the analysis reports

• RMST for each group up to tau with confidence intervals
• Difference and ratio in RMST with confidence intervals and p-value
• Kaplan-Meier curves with the area up to tau
• Restricted mean time lost (optional)

How to report it

Up to [tau] months, restricted mean survival time was [value] months in [group 1] and [value] months in [group 2], difference = [value] months, 95% CI [[value], [value]], p = [value].

Related methods

• Kaplan-Meier Estimator
• Log-Rank Test
• Cox Proportional Hazards Model

Run Restricted Mean Survival Time (RMST) on Quorant