Contextual Analysis of the Effect of Prostate-Specific Antigen Testing on Prostate Cancer Outcomes among Elderly Men
Authors
Advisor
Date
Embargo until
Language
Book title
Publisher
Peer Reviewed
Type
Research Area
Jurisdiction
Other Titles
See at
Abstract
Background: Prostate cancer (PCa) is the most common non-cutaneous cancer diagnosed in American men, with a median age of diagnosis of 66 years. There is limited information regarding the impact of pre-diagnosis annual prostate-specific antigen (PSA) testing on PCa outcomes considering the role of contextual characteristics. This study examined geographic variation in the effect of pre-diagnosis annual PSA testing on PCa outcomes, and county-level factors underlying such variation among Medicare-eligible older men. Methods: A retrospective cohort study was designed to analyze men aged 65+ from the linked Surveillance, Epidemiology, and End Results (SEER)-Medicare database. Pre-diagnosis annual PSA testing was stratified as receipt of 0-1 (reference), 2-3 or ≥4 PSA tests 5-years before PCa diagnosis. Composite indices for county-level socioeconomic and health services supply (HSS) characteristics were created using factor analysis. Cluster-specific logistic regression models were used to examine the effect of pre-diagnosis PSA testing on stage at diagnosis and the likelihood of undergoing expectant management (EM) for early-stage PCa. Results: Among 37,760 men, 6% were diagnosed with incident distant PCa and 20% underwent EM within the first two years of PCa diagnosis. In the adjusted analyses, greater intensity of PSA testing was associated with a lower likelihood of incident distant PCa diagnosis (AOR for ≥4 PSA tests: 0.20, 95%CI: 0.17-0.22), and a lower likelihood of undergoing EM (AOR: 0.48, 95%CI: 0.43-0.54). Decreasing HSS characteristics were associated with a higher likelihood of distant PCa diagnosis (AOR: 1.44, 95%CI: 1.08-1.92), and higher likelihood of undergoing EM (AOR: 1.32, 95%CI: 1.09-1.59). While, the effect of pre-diagnosis PSA testing on stage at diagnosis did not vary across counties (p-value=0.21), the effect on undergoing EM varied across counties (p-value<0.01). The fully-adjusted predicted proportions ranged from 3-15% (SD: 7%) for distant PCa diagnosis, and from 12-39% for EM (SD: 4%) across counties. Conclusions: Stage at diagnosis and treatment for early-stage PCa among older Medicare beneficiaries were influenced by individual and county characteristics. Geographic variation in PCa outcomes highlights the variation in health care needs across the U.S. The findings suggest that PCa screening guidelines should be informed by both patient and contextual characteristics.