A cross-sectional study examines all of the participants at one time. So for a cross-sectional study of cancer survivors, we would gather some sample of former patients and find out how long they have been in remission.
The problem with that cross-sectional design is we would be missing the people whose cancer had returned and caused them to die before our experiment started; we would have introduced a problem called "survivor bias" by only counting people who were STILL in remission and still alive.
A better experiment will use a longitudinal design of enrolling people whose cancer has just gone into remission, and following them to see how long they stay healthy.
