Geographic Access to Oncology Services in the United States (US): Travel Disparities May Affect Granulocyte-Colony Stimulating Factor (G-CSF) Administration

Introduction: Cancer patients receiving myelosuppressive chemotherapy often have suppressed white blood cell counts and are at risk for febrile neutropenia (FN). Treatment with granulocyte-colony stimulating factors (G-CSFs) after administration of chemotherapy is indicated for patients with a clinically significant risk of FN. In the US most doses of G-CSFs are administered in the clinic during separate visits after the chemotherapy session, to ensure on-label administration. Patients typically must return to the clinic between 1 and 7 times per chemotherapy cycle for G-CSF injections. Travel time and distance to specialized cancer care can be significant, and was previously reported to average 26 minutes round-trip for residents of the continental US, and considerably longer for some sub-populations and geographic areas (Onega et al. Cancer 2008). Travel distance has also been linked to reduced access to needed care. This study examined the travel burden specifically for a national sample of US cancer patients who are often indicated for G-CSF therapy.

Methods: We estimated the driving distance and time to oncologist offices and hospitals for a national cohort of 3,606,270 Medicare fee-for-service beneficiaries with breast, lung, colorectal or prostate cancer. Medicare beneficiary counts by age cohort (<65, 65+), gender and ZIP code from the 2013 American Community Survey (ACS) were combined with 2014 Medicare cancer prevalence rates by age cohort, gender and county from the Medicare Disparities Mapping database to estimate the number of cancer patients by age and gender in each ZIP code. Oncologist practice locations and hospital addresses were obtained from the Medicare Physician Compare and Hospital Compare websites and geocoded using Google Maps. ZIP code population-weighted centroids were calculated from US Census Bureau population and geospatial files using PostGIS. Driving distance and time to the nearest oncologist office or acute care hospital from each ZIP code's centroid were calculated using Open Street Maps road networks. Geographic and socio-economic characteristics of each ZIP code from the ACS were used to stratify and analyze the travel estimates.

Results: Cancer patient prevalence in 31,305 ZIP Codes ranged from 1 to 3531 patients. Patients would travel to 6753 unique oncologist or hospital addresses as the nearest possible locations for G-CSF therapy. The mean (median) round-trip driving distance to the nearest oncology provider was 10.6 (6.6) miles and the mean (median) round-trip driving time was 18.4 (12.7) minutes. When stratified by population density, the mean (median) round-trip travel time varied from 5.8 (5.2) minutes in Very Dense Urban areas to 50.2 (47.8) minutes in Super Rural areas. Variation by Census Region ranged from 15.2 (10.2) minutes in the Northeast to 21.0 (15.6) minutes in the South.

About 90% of patients would have round-trip travel times of <40 minutes. However, patients in areas with above average concentrations of elderly, poor or disabled residents were more likely to experience longer travel. Patients living in ZIP codes with an above average (above the median) concentration of residents aged 65+ were almost twice as likely to need to drive >40 minutes round-trip to the nearest oncology provider (13.5% vs 6.9% of patients in ZIP codes with a lower percent of elderly). Mean drive times were 20.2 minutes in high elderly ZIP codes vs 16.6 minutes in low elderly ZIPs. Patients living in ZIPs with above average concentrations of residents living below the Federal Poverty level were slightly more likely to drive 40+ minutes to care (10.6% vs 9.7% in low poverty areas). Patients living in ZIPs with above average concentrations of disabled residents were more than twice as likely to drive 40+ minutes to care (14.1% vs 6.3% in low disability areas). Mean travel time was almost 25% higher in the ZIPs with higher percentages of disabled residents.

Conclusions: Cancer patients needing G-CSFs following myelosuppressive chemotherapy appear to have generally good geographic access to oncology services in the US, but travel disparities exist for some vulnerable populations. In this modeling study, travel may be under-estimated due to inclusion of all acute care hospitals as potential providers of oncology services. To address this limitation, our future research will link patients to their actual sites of care for G-CSFs.