Restricted mean survival time
Restricted mean survival time is average time alive or event-free up to a chosen horizon, measured as area under a survival curve.
Fit
When to use it, and when not to
Use it when
- You want a survival contrast in days, months or years through a clinically useful horizon.
- A constant hazard ratio is doubtful, but survival can be estimated through the chosen horizon.
Not when
- You need lifetime survival beyond observed follow-up. A justified parametric survival model needs separate extrapolation checks. Read: Choosing a survival measure beyond the hazard ratio
- Death competes with a nonfatal outcome. Do not integrate a Kaplan-Meier curve that censors deaths as ordinary losses. Read: Choosing an analysis for competing risks
Before you start
What you need
- A horizon chosen in advanceUse the same horizon for both groups, supported by follow-up in each.
- The event defining survivaloutcomeTypeState whether the time is alive, relapse-free or free of another defined first event.
- Survival estimates and uncertaintyKeep event times, censoring times and the comparison direction.
- Within each reported group, loss to follow-up is unrelated to future event times.How to check: Record why follow-up ended. Assess how results change if those lost have a worse outlook.
- Different people supply independent observations.How to check: For patients grouped within clinics or families, use inference that accounts for those groups.
- Follow-up supports estimation through the chosen horizon.How to check: Show risk sets near the horizon. A few remaining patients cannot support a precise long-term claim.
- An observational group difference can be confounded.How to check: Use a justified adjustment method if the target is an adjusted or causal contrast.
Plain language
How it works
Imagine cutting the survival curve at a fixed time and adding the area beneath it. For a step curve, each rectangle is its width in time times its survival height. The sum is restricted mean survival time. Subtract one group area from the other to get a difference in the same time units.
The measure does not require proportional hazards. It still needs valid survival curves and a horizon with adequate support. Royston and Parmar develop estimation and trial comparisons using this measure. Their later paper studies design and power; a reported difference needs its own standard error and confidence interval. It is not the mean among only those who died, and it is not a median.
Worked example
A worked example
- Setting
- Two groups are followed through month 6.
- Question
- How much time is spent alive, on average, during those six months?
- Approach
- Group A survival is 1.0 from months 0 to 2, 0.8 from 2 to 4, and 0.5 from 4 to 6. Group B survival stays 1.0.
- Result
- Restricted mean survival is 4.6 months for A and 6.0 months for B. The A-minus-B difference is -1.4 months.
- What it means
- The difference concerns the first 6 months only. It says nothing about the difference in lifetime survival.
In practice
Recent studies PubMed search · not vetted
Watch out
Common mistakes
The one to remember
Restricted mean survival time depends on a stated horizon with adequate follow-up. It does not estimate lifetime survival. A Kaplan-Meier curve that censors a competing death or another event overstates cumulative incidence when read as one minus survival.
- Choosing the horizon because it gives the strongest result.Instead: Prespecify the horizon and label other horizons as sensitivity analyses.
- Calling the area a median survival time.Instead: The restricted mean averages time through a horizon; the median marks when survival reaches one half.
Inside Mallard
How Mallard uses this
Learn search lists Restricted mean survival time. The method registry links this method to its single-method article and code.
restricted-mean-survival-time).analysis.rmst, the id a plan cue records.Related Learn pages
Related designs and methods
Evidence
Sources & further reading
- The use of restricted mean survival time to estimate the treatment effect in randomized clinical trials when the proportional hazards assumption is in doubt. (opens in a new tab) · Stat Med · doi 10.1002/sim.4274
- Restricted mean survival time: an alternative to the hazard ratio for the design and analysis of randomized trials with a time-to-event outcome. (opens in a new tab) · BMC Med Res Methodol · doi 10.1186/1471-2288-13-152