New England Section of the American Urological Association

Back to 2018 Program


Specialist Density and Surgery vs. Radiation Therapy for Prostate Cancer: Does Supply Induce Demand?
Lael Reinstatler, MD, MPH1; Elias S. Hyams, MD2
1Dartmouth Hitchcock Medical Center, Lebanon, NH; 2Columbia University, New York, NY

Background: There is substantial regional variation in rates of surgery and radiation therapy (RT) for prostate cancer (CaP), and lack
of objective explanations for why this occurs. Workforce supply (i.e. of urologists (URO) and radiation oncologists [RADONC]) is one potential explanation for
treatment variation between regions. In this study, we evaluated whether population density of URO and RADONC was associated with higher rates of surgical
and radiation treatment, respectively.
Materials & Methods: Regional rates of radical prostatectomy (RP) and RT (external beam therapy and brachytherapy) among male Medicare beneficiaries with
CaP were obtained for 2007-2012 from the Dartmouth Atlas (www.dartmouthatlas.org). These rates were adjusted for age and race. Workforce data for URO and RADONC were obtained from the American Medical Association for 2012, and the population density of specialists for 306 hospital referral
regions (HRR) was calculated. We calculated Pearson correlations for specialist density and rates of surgery and radiation therapy, and compared rates of treatment for highest and lowest 20 HRRs based on specialist density.
Results: Mean regional URO density was 97 per 100,000 male Medicare beneficiaries (median 82, range 20-270). Mean regional RADONC density
was 40 (median 32, range 10-410). There was no significant correlation between URO density and regional rate of RP (r=-0.12, p=0.11), nor for RADONC density and regional rate of RT (r=-0.02 p=0.74). Comparing the highest and lowest 20 regions for URO density, there was no significant difference in average rate of RP (109 vs. 126 per 1,000 male beneficiaries with prostate cancer, p=0.13), though there was slightly higher RP volume in low density regions. Comparing highest and lowest 20 regions for RADONC density, there was no significant difference in RT, though there was a higher rate of RT in high RADONC density regions (259 vs. 232, p=0.11).
Conclusions: There is no clear relationship between specialist density and rates of definitive treatment for CaP. These data suggest that it is not merely the presence of specialists, but other cultural and systemic factors that are contributing to the use of certain therapies. As patient decisions should not depend on availability of specialists but rather personal and medical criteria, these data are tentatively reassuring. Further research in drivers of treatment decisions, with goals of ensuring shared decision-making, are needed.


Back to 2018 Program