Is there a method out there that can compare the the KM estimate (i.e. survival proportion) at a specific time point (e.g. 5-year) across more than 2 groups? For example, a clinician would like to know if the survival probability at 5 years is different between groups. For groups = 2, the situation is simple in that I can assume each KM estimate is asymptotically normal and then apply Delta theorem to compute a z-score from the estimated difference and standard errors in KM estimates between 2 groups. Does anyone know of a method that will produce a statistic for >2 groups?

Right now I'm considering some sort of joint multivariate normal distribution with independent covariance matrix, but perhaps this has been derived before in the literature.

  • $\begingroup$ Now I am interested in this question. Could you explain in detail what you do for two groups? $\endgroup$
    – Viktor
    Commented Oct 27, 2018 at 8:40


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.