Kaplan-Meier Survival Analysis

Kaplan-Meier survival analysis is a nonparametric method that estimates the probability of surviving beyond each time point from time-to-event data, accounting for censored observations.

Kaplan-Meier 생존분석은 중도절단된 관측치를 고려하여 생존시간 자료로부터 각 시점 이후까지 생존할 확률을 추정하는 비모수 방법이다.

How do I estimate and compare survival curves when some subjects have not yet had the event?

일부 대상자가 아직 사건을 겪지 않았을 때 생존곡선을 어떻게 추정하고 비교하나요?

The Kaplan-Meier estimator calculates the survival probability at each observed event time while correctly handling censored cases, and the curves of different groups are usually compared with the log-rank test. It fits time-to-event outcomes such as death, relapse or churn, where follow-up ends before the event for some subjects.

When to use Kaplan-Meier Survival Analysis

• Time-to-event data with right-censored observations
• Estimating median survival time and survival at fixed time points
• Comparing survival curves between treatment or exposure groups

When to use something else

• You need to adjust for several covariates: use Cox proportional hazards regression
• Several mutually exclusive event types compete: use a competing risks (cumulative incidence) analysis

Assumptions and requirements

• Non-informative (independent) censoring
• Survival probabilities do not depend on calendar time of entry
• Event times recorded accurately
• Independent subjects

What the analysis reports

• Survival curves with confidence bands
• Median survival time with confidence interval
• Survival probabilities at fixed time points
• Number-at-risk table
• Log-rank test

How to report it

Median survival was [value] months (95% CI [[lower], [upper]]) in [group A] and [value] months in [group B], and the difference was significant, log-rank χ²([df]) = [value], p = [value].

Related methods

• Log-Rank Test
• Cox Proportional Hazards Regression
• Restricted Mean Survival Time

Run Kaplan-Meier Survival Analysis on Quorant