Advanced imaging and hospice use in end-of-life cancer care

Introduction: Advanced imaging can inform prognosis and may be a mechanism to de-escalate unnecessary end-of-life care in patients with cancer. Associations between greater use of advanced imaging and less-aggressive end-of-life care in real-world practice has not been examined. Methods: We conducted a retrospective analysis of SEER-Medicare data on patients who died from breast, lung, colorectal, or prostate cancer between 2002 and 2007. Hospital referral region (HRR)-level use of computerized tomography (CT), magnetic resonance imaging, and positron emission tomography was categorized by tertile of imaging use and correlated with hospice enrollment overall and late hospice enrollment using multivariable logistic regression. Results: A total of 55,058 patients met study criteria. Hospice use ranged from 50.8% (colorectal cancer) to 62.1% (prostate cancer). In multivariable analyses, hospital referral regions (HRRs) with high rates of CT imaging were associated with lower odds of hospice enrollment (odds ratio, 0.80; 95% CI, 0.70–0.90) and late enrollment among those who did enroll (odds ratio, 1.49; 95% CI, 1.26–1.76). HRRs with the highest rates of CT use were predominantly located in the Midwest and Northeast and associated with higher percentage population of black patients (14.5 vs 5.6%), greater comorbidity (28.4 vs 23.7%), metropolitan residence (93.9 vs 78.5%), and less than high school education (26.4 vs 19.3%). Conclusion: In this population-based retrospective study, we did not observe evidence that overall and timely hospice are associated with higher rates of imaging near the end of life. An observed association between higher rates of imaging, particularly CT, may be explained in part by HRR-level differences in practice patterns and patient demographic characteristics. Further research is warranted to explore the ability of oncologic imaging to appropriately de-escalate care.