Regional variation in survival after metastatic prostate cancer diagnosis.

Abstract

4666 Background: Survival after diagnosis of metastatic prostate cancer (mPC) averages 2-3 years. Substantial regional differences in survival have been documented for PC prior to 2000. We investigated the extent to which regional variation exists in survival for mPC patients during 2000-2007. Methods: Using the Surveillance, Epidemiology, and End Results (SEER) database linked to Medicare claims, we identified men (mean age, 77.6 years) continuously enrolled in Medicare Parts A/B who were diagnosed with mPC between 2000 and 2007 and not diagnosed with another malignancy. The base-case model was limited to hospital service areas with >50 patients. A Cox proportional hazards model was used to estimate hazard ratios (HR) for overall survival (OS), adjusting for year of diagnosis, age, marital status, poverty and education covariates, Gleason score, comorbidities, and region covariates. Sensitivity of results was tested by limiting the analysis to patients surviving at least 6 months after diagnosis and by removing regional sample size limits. We report HRs for covariates, results of a Wald test of joint significance for region effects, and percentage difference from the mean for each region’s HR for death. Results: A total of 2696 patients with mPC met the inclusion criteria. OS was 37% at 3 years and mean HR for death was 4.8 (sd 2.1). HRs for death were positively correlated with age (HR=1.96, 95% CI: 1.6-2.3 for >80 years), PC-specific comorbidity index (HR=1.2, 95% CI: 1.1-1.3), and Gleason score (HR=6.6, 95% CI: 2.4-17.8 for poorly differentiated). Year of diagnosis, race, and socioeconomic status were not significantly associated with mortality. Wald test of joint significance for survival across regions of P=.019 indicated significant differences in survival across regions and was robust in sensitivity analyses. Patients living in regions with the worst survival rates had HRs for death that were 20% higher than the mean (HR=5.8) and those living in regions with the best survival rates had HRs 30% lower than the mean (HR=3.4). Conclusions: There are significant regional differences in survival for mPC patients in the 2000s. Further research is needed to determine if treatment differences play a role in this disparity.