RMST Restricted Mean Survival Time
RMST is the area under the Kaplan-Meier curve over [0, τ] and can be read as "the mean survival time up to time τ". When the proportional-hazards assumption fails and the HR is hard to interpret, the RMST difference is a more intuitive and robust between-group measure. Computed locally in your browser; data are not uploaded.
① Input
Each row: survival time status (1=event/0=censored) [group]. Group is optional; with two groups, the RMST difference is compared automatically.
How to use & methodology
Why is RMST better than the HR?
The HR assumes the two groups' hazard ratio is constant (proportional hazards); when survival curves cross or the hazard ratio changes over time, the HR loses meaning. The RMST difference does not rely on this assumption and reads directly as 'how much the mean survival time within τ differs', which is more intuitive clinically.
How do I choose τ?
Pre-specify it by the clinical question before looking at the data, often the common observable follow-up time (e.g. the smaller of the groups' maximum follow-up times, or a clinical milestone like 5 years). τ cannot exceed any group's longest observed time.
What are the units of RMST?
The same as survival time (months, years, etc.). For example RMST(5 years)=4.2 years means mean survival of 4.2 years within 5 years. A difference of 0.6 years means about 7 more months of life on average.
How does it work with Kaplan-Meier and Cox?
First look at the KM curves and log-rank; if proportional hazards holds, use Cox to report the HR; if the curves cross or PH fails, use the RMST difference as the primary or supplementary between-group measure.