345 CONTEMPORARY NATIONALLY-REPRESENTATIVE TRENDS IN TREATMENT FOR LOCALIZED PROSTATE CANCER

Abstract

INTRODUCTION AND OBJECTIVES: Primary treatment trends for localized prostate cancer, including substantial local variation, have previously been documented among selected community-based U.S. practices. We aimed to determine whether these trends are confirmed in larger, more broadly representative databases. METHODS: We analyzed treatments in two national administrative datasets including men covered by Medicare in 1998–2006 and by private insurance (i3 database) in 2002–2006. Treatment distributions were determined in each year among watchful waiting/active surveillance (WW), radical prostatectomy (RP), radiation therapy (RT), cryotherapy, and primary androgen deprivation therapy (PADT). Trends in specific modalities were further examined within RP and RT. County-level variation was assessed, and treatment predictors were determined using multinomial regression. RESULTS: Among 54,322 Medicare patients (mean age at diagnosis 75), WW use fell from 30% in 1998 to 22% in 2002 and remained low. PADT use peaked at 24% in 2002 and fell to 17% by 2006. RP use varied between 11% and 13% during the study period, while RT use rose from 35 to 44%. Laparoscopic/robot-assisted RP (LRRP) and intensity-modulated radiation therapy (IMRT) were both first reported in 1999, and rose to 35% of RP and 53% of RT cases, respectively, by 2006. Among 16,161 privately insured patients (mean age 65), WW and PADT fell steadily from 25% and 12% in 2002, respectively, to 12% and 7% in 2006. RT use varied between 30 and 32%, while RP use rose from 33 to 48%. LRRP and IMRT use rose from 1% and 15% of RP and RT cases, respectively, in 2002 to 41% and 48% in 2006. Use of neoadjuvant androgen deprivation therapy (NADT) declined among RP and brachytherapy patients, to 6 and 25%, respectively, among Medicare patients and to 2% and 22% among private insurance patients by 2006. Use of NADT rose to 41% and fell to 37% among external-beam RT patients with Medicare and private insurance, respectively. There was considerable geographic variation in treatment selection. Across counties, use of WW, PADT, RP, and RT ranged, respectively, from 3 to 55%, 0 to 48%, 0 to 50%, and 15 to 71%. Comorbidity, age, income, year of diagnosis, and county-level sociodemographic variables predicted treatment. CONCLUSIONS: These trends echo those documented in prior studies in smaller datasets, and these data confirm extensive geographic variation. In both databases, there is rapid, ongoing adoption of high-cost technologies among both RP and RT patients. These findings underscore the urgent need for high-quality cost-effectiveness research comparing these treatments.