Predicting Other Cause Mortality Risk for Older Men with Localized Prostate Cancer: A Dissertation

摘要: Background: Overtreatment of localized prostate cancer (PCa) is a concern as many men die other causes prior to experiencing treatment benefit. This dissertation characterizes the need for assessing cause mortality (OCM) risk in older with PCa and informs efforts identify patients most likely benefit from definitive treatment. Methods: Using linked Surveillance Epidemiology End Results-Medicare Health Outcomes Survey database, 2,931 (mean age=75) newly diagnosed clinical stage T1a-T3a 1998-2009 were identified. Survival analysis methods used compare observed 10-year OCM by primary type. Age health factors predictive type assessed multinomial logistic regression. Predicted estimates Social Security life tables (recommended expectancy evaluation) two estimation tools compared rates. An improved prediction model was developed fitting Fine Gray competing risks models age, sociodemographic, comorbidity, activities daily living, patient-reported data predictors. The tools’ ability discriminate between who died those did not evaluated Harrell’s c-index (range 0.5-1), which also guided new selection. Results: Fifty-four percent underwent radiotherapy while 13% prostatectomy. Twenty-three treated 12% undergoing prostatectomy experienced within 10 years thus considered overtreated. indicative shorter (increased worse physical health, smoking) had little no association assignment but significantly related reductions likelihood overestimated discriminated poorly over (c-index=0.59). Existing less overestimate rates limited discrimination (c-index=0.64). A self-reported Charlson comorbidity index score, overall (excellent-good/fair/poor), smoking, marital status predictors (c-index=0.70). Conclusions: primarily attributable may be reduced pretreatment assessment mortality-related factors. provides prognostic utilizes set five characteristics that better diagnosis than age comorbidity-based assessments alone.