OBJECTIVES: To examine geographic variation in the effect of pre-diagnosis annual prostate specific antigen (PSA) testing on the likelihood of undergoing expectant management among Medicare-eligible older men.

METHODS: We analyzed patients aged 65 years or older from the linked Surveillance, Epidemiology, and End Results (SEER)-Medicare database. The sample included men diagnosed with localized/regional (early-stage) prostate cancer from 2004-2007, with continuous fee-for-service Medicare coverage in the 60 months prior to prostate cancer diagnosis. Annual PSA testing was stratified as receipt of 0-1, 2-3 and 4 or more tests during the 5 years before prostate cancer diagnosis. Cluster-adjusted logistic regression models were used to quantify the effect of annual PSA testing on treatment receipt controlling for patient and county characteristics. We assessed geographic variation in prostate cancer treatment outcomes using 1) random intercept/slope models, 2) variance partition coefficients, and 3) caterpillar plots of predicted proportions of men undergoing expectant management across counties.

RESULTS: Among 35,362 older Medicare beneficiaries, 20% underwent expectant management within the first two years of early-stage prostate cancer diagnosis. Multivariate analyses showed that greater intensity of annual PSA testing was associated with a statistically significant lower likelihood of undergoing expectant management compared to receiving zero or one PSA test prior to diagnosis (4 or more times OR: 0.48, 95% CI 0.43-0.54; 2-3 times OR:0.64, 95% CI: 0.58-0.70). County of residence accounted for 2.5% of the variation in treatment receipt. The effect of annual PSA-testing on expectant management varied across counties (p<0.003). The predicted proportions of men undergoing expectant management after adjusting for individual and county-level characteristics ranged from 11.65% to 38.66% (Mean: 20.25%, Median: 20.56%, SD: 3.96%) across 158 SEER covered counties.

CONCLUSIONS: There was variation in the use of expectant management among Medicare beneficiaries based on patient demographic characteristics and geographic location.