median value of kmplot

Question

[yueli7]

Hello,

Anyone draw Kaplan-Meier plotter by using

[1]kmplot.com/analysis/?

In my understanding, the median value of split patients should be the higher 50%, and the lower 50%.

If there is 3951 patient, the lower should be 1976, and the higher should be 1975. But when I input ABCC2 gene in breast cancer, the lower is 2020, and the higher is 1931.

My question is why the lower is not 1976, the higher is not 1975?

The result is:

Number at risk

low 2020

high 1931

Thanks in advance for great help.

[Kevin Blighe]

There are certain circumstances in which the median will not always divide a cohort into equal halves. Also, the coding behind this program may be doing some other filtering which, if occurring, should be listed in the published manuscript or documentation on the website.

[yueli7]

Thank you so much for answering my question.

I think this can solve my problem.

determining cutoff for Kaplan Meier

Thanks!

[Kevin Blighe]

Yes, I was going to say that splitting the cohort by the median is not really a good idea. It should be at least split into thirds, and then comparing the upper and lower tertile.

Best of luck, my friend.

[yueli7]

Thank you so much for your answering.

I should read more paper on this issue.

Thanks