Introduction of bias by merging patients with initial surgery and reoperations in analysis of survival.