Fig 1.

Flow diagram showing patient and control participation and follow-up. *Lost to follow-up includes study participants who did not participate in the interview for that year but could have participated in subsequent interview years; controls who were diagnosed with prostate cancer after baseline and who were censored from date of diagnosis (n=35). Lost to follow-up frequencies not available for controls in the time periods preceding years 3 and 10 because controls were not interviewed in those years