Oh, whether short vs long is associated with cancer. I guess that should have been clear from De Vliegende Hollander's query.

On a different note, that second graph looks like nonsense to me. The cancer group holds steady, while the control group bobs up and down. The effect is in the control group, not in the cancer group. This graph tells nothing about cancer. 95% it's pure nonsense, but maybe it tells something about the control group - it's predicting their censoring by death. Which is why it's crazy to use time to censoring as the metric for the control group.