Geographic Density of Linear Accelerators and Receipt of Radiation Therapy for Prostate Cancer

Abstract

Radiation therapy (RT) is one among several definitive treatments for patients with prostate cancer. Recent analyses have suggested distance from a radiation oncologist influences the likelihood that a patient is to receive RT for their prostate cancer. We aimed to assess if linear accelerator density within healthcare referral regions (HRRs) impacts practice patterns regarding the treatment of prostate cancer patients.Along with the American Society for Radiation Oncology (ASTRO), we identified, and quality checked the location of linear accelerators in the United States as of July 2020, derived from the International Atomic Energy Agency Directory of Radiotherapy Centers (DIRAC) database. The address to each practice was geocoded using the Google Maps API within QGIS 3.12. The number of linear accelerators was determined in each HRR as defined by the Dartmouth Atlas; HRRs represent "regional healthcare markets for tertiary medical care." In 2014, the Dartmouth Atlas published data regarding the management of prostate cancer by HRR based on Medicare data from 2007 - 2011. Linear regression tested the association of linear accelerator density with receipt of prostate cancer radiation therapy.We identified 3,521 linear accelerators in the United States serving a population (as of the most recent US Census in 2010) of 309 million people; this represents approximately 88,000 people per linear accelerator, or 11.4 linear accelerators per million people. Among the 306 HRRs, the mean density of linear accelerators is 11.7 per million (standard deviation 5.2); 5 HRRs do not have a linear accelerator. For 227 HRRs, data regarding the proportion of Medicare beneficiaries over the age of 75 with prostate cancer who received RT was available. For every 1 additional linear accelerator per million, an additional 2.1 prostate cancer patients per 1,000 received RT (95% CI: 0.08 - 4.2; P = 0.04). For 234 HRRs reporting the number of prostatectomies per 1,000 Medicare beneficiaries under 75 with prostate cancer, there was no relationship between prostatectomy rate and number of linear accelerators (P = 0.42).Increasing HRR linear accelerator density is associated with increasing rate of receipt of radiotherapy for prostate cancer in elderly men. These findings need to be explored across other disease sites and age ranges to ascertain the magnitude of this effect and optimize the distribution of linear accelerators to best ensure access to care across the geographic spectrum.