0
$\begingroup$

Let's assume we have a 5-year follow-up for survival for two groups, and I report their crude mortality rates for years 1, 2, 3, 4 and 5. After that, would it be reasonable to test the mortality difference once, at the end of the follow-up, or should I do the log-rank test for each year?

For example:

The groups' mortality was similar during the five-year follow up (p = 0.5).

Does the above also mean that their mortality rates were similar at year 1 or 3?

$\endgroup$

1 Answer 1

0
$\begingroup$

This depends on what your prior belief/hypothesis is. If you are wanting to test whether the mortality rates are the same over the 5 years, then you should simply run the test on all the information you have available, the 5 years.

If it were the case that you had some reason (prior to seeing the data) to believe there might have been a change in the mortality difference for different periods, and you wanted to know if the mortality was the same in each period, you should run the test on the different periods. But you should really make this decision before seeing the data.

$\endgroup$

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.