TrialCurveRestricted mean survival time (RMST) calculator
RMST is the area under the Kaplan–Meier curve up to a chosen time τ: the average event-free time patients experience over that horizon. The difference in RMST between arms ("patients on drug lived on average 2.1 months longer over 3 years") holds even when hazards are not proportional (Royston & Parmar 2013).
From patient-level or reconstructed data
Have reconstructed or real IPD already? Paste CSV rows as arm,time,event (event 1 = event, 0 = censored).
From a published figure
No IPD? Upload the KM plot. The reconstructed data come back with RMST already computed, and you can download the CSV to recompute at any τ.
Choosing τ
- Choose τ before looking at the results, based on clinical relevance: a 2-, 3- or 5-year horizon, for example.
- τ must not exceed the follow-up of the arm with the shortest follow-up. By default the calculator uses the smallest maximum follow-up across arms.
- Report the RMST difference with its 95% CI alongside the HR, especially when the curves cross.
RMST vs hazard ratio vs median
The HR summarises relative risk and assumes proportional hazards. The median describes one point on the curve and is "not reached" when survival stays above 50%. RMST uses the whole curve up to τ and is expressed in time units that are easy to communicate. HTA bodies increasingly ask for it when hazards are non-proportional.
FAQ
What does 'median not reached' mean?
The KM curve never fell to 50% during follow-up, so the median survival time cannot be estimated. RMST is still defined up to any τ within follow-up.
How is the RMST confidence interval computed?
Here, with the Klein–Moeschberger variance of the area under the KM curve; the difference between arms assumes independent samples.
Reconstruct IPD from your own figure. Upload a Kaplan–Meier plot and get curves, numbers at risk, pseudo-IPD, hazard ratios and a validation pack. Three figures a month are free.
Try the live demo Browse trial data API docs