In the largest reported analysis of perioperative mortality evaluating the 15 most common surgically treated malignancies, those with Medicaid coverage or without insurance were more likely to die within 30 days of surgery.
Arya Amini, Norman Yeh, Bernard Jones, Yevgeniy Vinogradskiy, Edward Bedrick, Chad G. Rusthoven, Ava Amini, William T. Purcell, Brian D. Kavanagh, Sana D. Karam, Christine M. Fisher; University of Colorado; Northwestern University
PURPOSE: Cancer survival is known to vary based on socioeconomic factors, including insurance status. The purpose of this study was to evaluate predictors for perioperative mortality (death within 30 d of cancer-directed surgery) for the 15 most common surgically treated cancers in the United States.
PATIENT AND METHODS: The Surveillance, Epidemiology, and End Results (SEER) database was examined for the 15 most common surgically resected cancers. The database was queried from 2007 to 2011, with a total of 506,722 patients included in the analysis. Binomial logistic regression was used to assess the effect of patient and tumor characteristics on perioperative mortality under multivariable analysis.
RESULTS: The insurance status for all patients was as follows: non-Medicaid insurance (83%), any Medicaid (10%), uninsured (4%), and unknown (3%). Under univariable analysis, predictors for perioperative mortality included Medicaid or uninsured status (P < .001), older age (≥ 60 yr) (P = .015), nonwhite race (P < .001), being unmarried (P < .001), urban and rural residence (vs metropolitan) (P = .002), higher percent of county below the federal poverty level (P < .001), and lower median household income (P < .001). Perioperative mortality was also associated with more advanced disease, including higher tumor stage (P < .001) and metastasis (P < .001). After adjusting for age, race, sex, marital status, residence (urban or rural), extent of disease (in situ, local, regional, distant), and percentage of county below federal poverty level, patients with either Medicaid insurance (odds ratio [OR] = 1.22; 95% confidence interval [CI], 1.15–1.29; P < .001) or uninsured status (OR = 1.75; 95% CI, 1.61–1.87; P < .001) were more likely to die within 30 days of surgery compared with patients with non-Medicaid insurance. Additional statistically significant predictors for perioperative mortality under multivariable analysis included rural residence (OR = 1.07), race-including being African-American (OR = 1.07), Hispanic, (OR = 1.27), and Asian or Pacific Islander (OR = 1.19)-and being unmarried (OR = 1.08).
CONCLUSION: In the largest reported analysis of perioperative mortality evaluating the 15 most common surgically treated malignancies, those with Medicaid coverage or without insurance were more likely to die within 30 days of surgery.
Proceedings of the 97th Annual Meeting of the American Radium Society - americanradiumsociety.org