Fig. 2

Establishing convergent validity by comparing the diagnosis date obtained from the administrative data to the diagnosis date abstracted by registrars. The histogram shows the percentage of patients that have a given difference between diagnosis dates. The diagnosis date derived from the administrative data (“algorithm date”) was the same as the tumor registry date among 27% of patients. Just 2.9% of patients did not have newly diagnosed bladder cancer, because their registry diagnosis date preceded the date obtained from administrative data by more than 90 days