More expected events than group size in survival analysis

Question

[nimrodrap]

Hi, I'm doing survival analysis on TCGA data using the R survival package. After using the survdiff function, I get that one group of patients has a higher expected value (Expected) than its group size (N):

Has anyone encountered this behaviour, or can suggest why it happens?

Answer 1

[Kevin Blighe]

Hello, yes, I have seen this before. The expected values are weighted, which is why they can sometimes be more than the observed and total number of events. This page goes over the calculations: https://mcardle.wisc.edu/mstat/help/help/Notes-07.html

Kevin