Kaplan-Meier Survival Analysis

The goal of the Kaplan-Meier procedure is to create an estimator of the survival function based on empirical data, taking censoring into account.

Topics

For those with a calculus background, you can also see the proofs of some of the properties described on the above web pages at

References

Wikipedia (2015) Kaplan-Meier estimator
https://en.wikipedia.org/wiki/Kaplan%E2%80%93Meier_estimator

Rich, J. T. et al. (2014) A practical guide to understanding Kaplan-Meier curves
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3932959/

MedCalc (2023) Kaplan-Meier survival analysis
https://www.medcalc.org/manual/kaplan-meier.php

9 thoughts on “Kaplan-Meier Survival Analysis”

  1. Hi Charles,

    I would like to know if you can help me about this issue.

    I have a trial where I have follow up time until 120 months, but the last death occurred before 24 months, where more than 80% of the patients survived until the end. However, when I created the survival curve on excel, the x axis stops at 24 months (last death). Is there a way to show in the graph that the patients survived until the end ?

    Thank you very much in advance. By the way, great site and software. It is helping me a lot.
    All the best,

    Rodrigo

    Reply

  2. There is a topic I’ve been struggling with for some time around the fatigue testing of components and structures and the determination of the constant amplitude fatigue limit (CAFL). The CAFL is a limiting stress below which, the fatigue life is infinite. We’ve been working on a particular welded structural detail and have roughly 42 results, 16 of which are runouts. In most cases, the CAFL has been determined by a staircase method – choosing a test parameter, say a stress of 10 ksi, running the test and see if it fails or does not fail by a specified life. If it fails, the next test is run at a lower stress, say 9.5 ksi, until you reach a point where no specimens fail. In our case the life we’ve been using is 20,000,000 cycles. I’ve been looking for a different method for analyzing the data. Some use an MLE methodology, but as the scatter tends to increase as the fatigue life increases, this often results in estimates of a lower bound for design that are very low. We’ve attempted to apply the K-M method and had a result that looked at survivability as a function of the applied stress range. The curve goes nearly vertical near what would seem to be the CAFL. Are you aware of anyone trying to apply the method to a survivability issue like this and does this seem reasonable? Thanks

    Reply

Leave a Comment