Racial disparities exist in end-of-life lung cancer care, which could potentially lead to considerable racial differences in end-of-life care costs. This study for the first time estimates the racial differences in end-of-life care costs among lung cancer patients, and identifies and quantifies factors that contribute the most to these differences using a statistical decomposition method.
This is a retrospective analysis of patients 66 years and older, diagnosed with stage I-IV lung cancer, who died on or before December 31, 2013, using the Surveillance Epidemiology and End Result-Medicare data from 1991 to 2013. Ordinary least square regression of logarithmically transformed cost was used to estimate racial differences in end-of-life care costs among lung cancer patients. Blinder-Oaxaca decomposition was used to identify and quantify factors that contributed the most to these differences.
Non-Hispanic blacks had 10% to 13% higher end-of-life care costs as compared with non-Hispanic whites. Geographic variations, baseline comorbidity indices and stage at diagnosis contributed the most to explaining the racial differences in costs, with geographic variation explaining most of the differences. However, the observed factors could only explain 25% to 32% of the racial differences in end-of-life care costs.
Geographic differences in access to timely and appropriate care, and provider practice patterns, should be examined to understand the reasons behind geographic variations in racial disparity. Provider-level educational interventions to reduce small area practice variations and differential management of patients by race, as well as racially sensitive patient-level educational and navigational interventions might be critical in improving quality of care and reducing costs during end-of-life.
*Department of Management, Policy and Community Health, School of Public Health, The University of Texas Health Science Center at Houston, Houston
†Division of Pulmonary Critical Care & Sleep Medicine, The University of Texas Medical Branch, Galveston, TX
This study used the linked SEER-Medicare database. The interpretation and reporting of these data are the sole responsibility of the authors.
G.S. serves on the scientific advisory board for Sunovion, Astra Zeneca, and Mylan. The remaining authors declare no conflicts of interest.
Reprints: Suja S. Rajan, MHA, MS, PhD, Department of Management, Policy and Community Health, School of Public Health, The University of Texas Health Science Center at Houston, RAS E-319, 1200 Pressler St., Houston, TX 77030. E-mail: email@example.com.