Kaplan-Meier Survival Curve Generator
Product-limit survival estimation with right-censoring, hazard ratios and the Mantel-Cox log-rank test.
How it works
Survival is estimated by the product-limit method: at each event time the probability of surviving that interval is (1 − dᵢ/nᵢ), and overall survival is the running product of those terms. Subjects lost to follow-up are right-censored — they contribute to the risk set until censoring and then leave it without counting as events. Groups are compared with the Mantel-Cox log-rank test.
Frequently asked questions
What does censoring mean in survival analysis?
A censored observation is one where the event of interest has not been observed by the end of follow-up — the subject was still alive at last contact, withdrew from the study, or the study ended. Censored subjects still contribute information: they remain in the at-risk denominator up to the moment they leave. Discarding them instead would bias survival estimates downward.
How do you compare two Kaplan-Meier survival curves?
Use the log-rank (Mantel-Cox) test. It compares the number of events observed in each group against the number expected if survival were identical, summed across all event times, and returns a chi-square statistic and p-value. It weights all time points equally, so it is most sensitive when the hazard ratio between groups is roughly constant over time.
Method reference
Kaplan EL & Meier P (1958) J Am Stat Assoc 53(282):457-481.