Repeated measures has never been my strong suit. However, I have a small dataset of 10 patients. Each had a lab value (e.g. tumor marker) measured during the treatment course. Some patients who survived longer had a more values than others. Each lab value has a specific date, but these are different for all patients. I guess I could go back to each patient and record the start of treatment as week = 0. Then calculate week number to correspond to the date of the lab measurement. That should work, albeit time consuming.

Thoughts?