CHANGING PATTERNS IN COMPETING CAUSES OF DEATH IN MEN WITH PROSTATE CANCER: A POPULATION BASED STUDY

Abstract

We examined trends in hospitalization and death in men with prostate cancer to determine whether outcomes have changed with time in men diagnosed and treated for this disorder. A population based cohort study of 180973 patients with prostate cancer in the 1979 to 1996 Surveillance, Epidemiology and End Results cancer registry and 450448 admissions in the 1987 to 1996 Surveillance, Epidemiology and End Results-Medicare linked database were analyzed. ORs derived from logistic regression were used to assess time trends in mortality and hospitalization. Multinominal logistic regression was used to obtain the adjusted proportions of deaths due to various causes in different years. In men with prostate cancer the risk of death from cancer was 39.7% (OR = 0.61, 95% CI = 0.56 to 0.66), which was lower in 1995 to 1996 than in 1979 to 1980. Decreases in prostate cancer death were greater than those in cardiovascular disorders (OR = 0.85, 95% CI = 0.78 to 0.92) and evident even in men with nonlocalized disease. Overall nonprostate cancer causes of mortality increased (OR = 1.65, 95% CI = 1.52 to 1.79) and ultimately exceeded that due to prostate cancer. By 1995 to 1996 the proportion of prostate cancer deaths was similar to that of cardiovascular disorders (27.7% and 26.6%, respectively) and substantially less than that of all other sources combined (45.7%). Similar effects were observed for prostate cancer (OR = 0.40, 95% CI = 0.37 to 0.42) and nonprostate cancer (OR = 2.51, 95% CI = 2.36 to 2.68) hospitalizations. In men with prostate cancer decreases in prostate cancer hospitalization and mortality have been greater than those in competing diseases with time. Most deaths in patients with prostate cancer, including those with nonlocalized disease, are now due to nonprostate cancer causes.