Paper detail

The Effect of Treatment-Related Deaths and "Sticky" Diagnoses on Recorded Prostate Cancer Mortality

Background: Although recorded cancer mortality should include both deaths from cancer and deaths from cancer treatment, there is evidence suggesting that the measure may be incomplete. To investigate the completeness of recorded prostate cancer mortality, we compared other-cause (non-prostate cancer) mortality in men found and not found to have prostate cancer following a needle biopsy. Methods: We linked Medicare claims data to SEER data to analyze survival in the population of men aged 65+ enrolled in Medicare who resided in a SEER area and received a needle biopsy in 1993-2001. We compared other-cause mortality in men found to have prostate cancer (n=53,462) to that in men not found to have prostate cancer (n=103,659). Results: The age-race adjusted other-cause mortality rate was 471 per 10,000 person-years in men found to have prostate cancer vs. 468 per 10,000 in men not found to have prostate cancer (RR = 1.01;95% CI:0.98-1.03). The effect was modified, however, by age. The RR declined in a stepwise fashion from 1.08 (95% CI:1.03-1.14) in men age 65-69 to 0.89 (95% CI:0.83-0.95) in men age 85 and older. If the excess (or deficit) in other-cause mortality were added to the recorded prostate cancer mortality, prostate cancer mortality would rise 23% in the youngest age group (from 90 to 111 per 10,000) and would fall 30% in the oldest age group (from 551 to 388 per 10,000). Conclusion: Although recorded prostate cancer mortality appears to be an accurate measure overall, it systematically underestimates the mortality associated with prostate cancer diagnosis and treatment in younger men and overestimates it in the very old. We surmise that in younger men treatment-related deaths are incompletely captured in recorded prostate cancer mortality, while in older men the diagnosis "sticks"-- once diagnosed, they are more likely to be said to have died from the disease.