Concept Architecture
Concept
Theoretically, a Kaplan-Meier Curve is the graphical representation of the Kaplan-Meier survival estimator, showing the estimated probability of surviving beyond successive time points. The curve provides a non-parametric description of survival experience while accounting for right-censored observations. In health economics, Kaplan-Meier curves are fundamental for visualising clinical trial survival data, comparing treatment groups and informing the selection and validation of survival models used in economic evaluations.
Mathematically, the Kaplan-Meier curve is constructed by plotting the product-limit estimates of the survival function at each observed event time. The resulting step function decreases only when an event occurs and remains constant between events. Vertical drops correspond to observed failures, while censored observations reduce the number at risk without causing a decrease in survival probability.
In practice, Kaplan-Meier curves are generated using statistical software from patient-level time-to-event data. They are routinely presented in clinical trials, observational studies and health technology assessments to describe overall survival, progression-free survival and other time-to-event outcomes. The empirical survival curve is frequently used to assess goodness-of-fit when selecting parametric models for long-term extrapolation in health economic analyses.
Purpose
Used to display empirical survival over time, compare treatment groups visually, assess censoring patterns and provide the benchmark for fitting and validating parametric survival models in health economic evaluation.
Mathematical Formulae
Primary Formula
?(t) = ???�? (1 ? d? / n?)
where:
- ?(t) = estimated survival probability
- d? = number of events at time t?
- n? = number at risk immediately before time t?
Supporting Formulae
Conditional survival probability at each event time:
p? = 1 ? d? / n?
Overall survival:
?(t) = ?p?
Related Mathematical Methods
- Kaplan-Meier estimator
- Log-rank test
- Cox proportional hazards model
- Nelson-Aalen estimator
- Parametric survival modelling
- Survival analysis
Example
A randomised oncology trial follows patients for five years. At each observed death, the Kaplan-Meier survival estimate decreases, producing a stepwise survival curve. Separate curves for the intervention and comparator groups demonstrate improved survival with the intervention and provide the empirical benchmark against which Weibull, Gompertz and generalised gamma models are evaluated for long-term extrapolation.
Excel Implementation
| Function | Example Formula | Health Economics Application |
|---|---|---|
| PRODUCT | =PRODUCT(E$2:E2) | Calculate cumulative Kaplan-Meier survival probabilities from conditional survival estimates. |
| IF | =1-(C2/B2) | Calculate conditional survival probability at each event time. |
| COUNTIFS | =COUNTIFS(TimeRange,">="&A2) | Calculate the number of patients at risk immediately before each event time. |
| Scatter Chart | Step chart with horizontal connectors | Plot the Kaplan-Meier survival curve for visual comparison of treatment groups. |
VBA (Optional)
Automate construction of Kaplan-Meier survival curves, update survival estimates following new event data and export publication-ready survival plots.
Sources
- Kaplan EL, Meier P. Nonparametric Estimation from Incomplete Observations.
- Klein JP, Moeschberger ML. Survival Analysis: Techniques for Censored and Truncated Data.
- Collett D. Modelling Survival Data in Medical Research.
- Briggs A, Claxton K, Sculpher M. Decision Modelling for Health Economic Evaluation.
- NICE. Health Technology Evaluation Manual.
Related Concepts (2)
Library
Publications
1
NICE DSU Technical Support Document 21: Flexible methods for survival analysis — Rutherford, Lambert, Sweeting, Pennington, Crowther, Abrams & Latimer, TSD 21 ed., 2020 (NICE Decision Support Unit (University of Sheffield))
Guidance extending standard survival analysis to flexible parametric methods — spline-based models, fractional polynomials, mixture and cure models, and relative-survival approaches — for capturing complex hazard functions in economic evaluation.
Frequently Asked Questions (6)
What is a Kaplan-Meier curve?
A graphical display of the Kaplan-Meier survival estimate, shown as a step function that decreases at each observed event over time.
Source: Kaplan & Meier 1958
What does the height of a Kaplan-Meier curve show at a given time?
At any point along the horizontal time axis, the height of a Kaplan-Meier curve gives the estimated proportion of the group still event-free by that time. Reading up from a chosen time to the curve and across to the vertical axis gives the survival probability then. The curve starts at one, since everyone is event-free at the outset, and falls toward zero as events accumulate. Its height is therefore a direct estimate of survival at each moment. Collett (2015) explains this reading.
Source: Collett 2015
Why does the Kaplan-Meier curve step down?
The Kaplan-Meier curve steps down at each observed event because the estimated survival probability is updated only when an event occurs, dropping by an amount reflecting the proportion of those at risk who have the event at that time, and remaining constant between events. Censored observations do not cause a step but reduce the number at risk, affecting later steps. This step pattern arises from the product-limit calculation, which revises survival at each event time, giving the curve its characteristic staircase shape.
Source: Kaplan & Meier 1958
How is a Kaplan-Meier curve interpreted?
A Kaplan-Meier curve is interpreted by reading the estimated survival probability at each time from the height of the curve, with steeper or more frequent steps indicating higher event rates and a flatter curve indicating better survival. Median survival can be read where the curve crosses fifty per cent. Comparing curves for different groups shows differences in survival, with a lower curve indicating worse survival. The curve conveys the survival experience over time, though estimates become less reliable at later times as the number at risk falls.
Source: Collett 2015
What do censoring marks on a Kaplan-Meier curve show?
Censoring marks on a Kaplan-Meier curve, often shown as ticks, indicate the times at which individuals were censored, that is, lost to follow-up or still event-free at the end of observation, without experiencing the event. These marks do not cause the curve to step down but signal that those individuals leave the risk set at that time, reducing the number at risk for subsequent events. Showing censoring marks conveys how much follow-up remains over time and how heavily later parts of the curve rest on few individuals.
Source: Kaplan & Meier 1958
What are the limitations of a Kaplan-Meier curve?
A Kaplan-Meier curve becomes less reliable at later times, where few individuals remain at risk, so the steps are large and estimates uncertain, and it cannot extend beyond the observed follow-up, so it does not provide long-term extrapolation. It shows the observed survival but does not model covariate effects or the hazard's form. It also assumes non-informative censoring. These limitations mean the curve is read with caution at its tail and complemented by regression models and parametric extrapolation where needed.
Source: Collett 2015
Trust Record
Verified by Dr Darrin Baines
British health economist
Professional identity: darrinbaines.org
Verification date: 21 Oct 2025
Content version: 1.0.0
Canonical Identity
- Persistent URI
- https://healtheconomics.wiki/concept/kaplan-meier-curve
- Term code
- HE-EM-SM-037
Stable URI · Machine-readable · Resolvable · CC BY 4.0