I am a student in a hospital department and looking back through ~5 years of testing I have found ~70 patients who received a test. I found a threshold which is the lowest result compatible with survival; i.e. every patient who had a result below this threshold passed away. A test result above this threshold was not specific to survival, however. So I can say that based on this sample, a low test result was 100% specific to non-survival. However, my problem is that only 12 patients survived, so I'm worried that if I had a larger sample including more patients that survive, some of those patients might have had lower test results and the threshold would be lower. How would I go about working out how many patients who survive I would need to test to be confident in my threshold? Can I use a sample size calculation?