topic Re: How to perform Pairwise comparison on Kaplan-Meier survival curve? in Discussions

How to perform Pairwise comparison on Kaplan-Meier survival curve?

CorrelationDeer — Fri, 09 Jun 2023 01:01:17 GMT

Hello,

I am currently conducting a survival experiment where I am investigating the impact of 8 different diets on the survival of my subjects. There are 8 total groups, each consisting of approximately 20 individuals, that are exposed to a single diet from day 1 until they pass away. Every day, I record the number of deceased individuals.

After conducting the Kaplan-Meier curve using the survival analysis feature in JMP, I discovered significant differences among the groups. My next step is to compare the 8 diets and determine whether there are any significant differences between pairs of diets.

However, I am having difficulty finding the appropriate test in JMP, whether it be post-hoc or pairwise comparison. I have searched the community discussions, but have not found a post that addresses my specific needs. Could anyone offer any assistance or suggestions on how to proceed with my analysis?

Thank you in advance for any help you can provide.

Re: How to perform Pairwise comparison on Kaplan-Meier survival curve?

peng_liu — Sun, 30 Apr 2023 18:08:53 GMT

I think that you at least need the Local Data Filter feature in JMP to help you to get pair-wised comparisons. Here is a screenshot to illustrate. There are 8 groups in the data, I can pick any two groups to compare using the Local Data Filter.

If you are interested in multiple-comparisons, then these two resources are relevant: Multiple Comparisons Problem and Bonferroni Correction . In a summary, you need to adjust your confidence level based on how many simultaneous hypotheses you have. And your hypotheses are those pair-wised tests.

Re: How to perform Pairwise comparison on Kaplan-Meier survival curve?

DensityOfSpaces — Thu, 08 Jan 2026 20:50:08 GMT

In SAS Enterprise one can enter the following for multiple comparisons using the Sidak correction. Is there an equivalent approach in JMP?

Strata 'Groups'/ test=logrank adjust=sidak;