Back to Annual Meeting Program

Objective: Examine association between insurance and pancreatic cancer outcomes
Design: Retrospective medical record review from tumor registry data. Univariate and multivariate analyses performed by chi-square and logistic regression; survival calculated using Kaplan-Meier (SAS 9.3)
Setting: Single institution
Patients: Pancreatic cancer patients 2001-2011
Outcomes: Stage at diagnosis, surgery with therapeutic intent, chemotherapy, radiation therapy, survival
Results: 1,244 patients identified: 51.5% Medicare, 3.4% Medicaid, 43.9% private insurance, 1.3% uninsured. Rates of uninsurance and private insurance dropped (1.8% to 0.7%; 48.2% to 41.3%.) Medicare increased (46.3% to 55.0%); Medicaid remained stable (3%.) More Medicaid and uninsured were unresectable (59.5% and 75%), compared to Medicare and private (46.3% and 48.5%.) Insurance was associated with receipt of surgery, chemotherapy, radiation and survival; uninsured had a median survival of 107 days, <50% of Medicare/Medicaid patients and <25% of private. Resectable patients with public/no insurance had lower rates of chemotherapy, radiation, and surgery. After multivariate modeling for sex, age, race, marital status, insurance, and stage, insurance predicted receipt of chemotherapy and radiation; receipt of surgery lost significance. Uninsured patients were still significantly less likely to undergo chemotherapy (OR 0.3; 95% CI 0.1-0.9) and Medicare patients significantly less likely to undergo radiation (OR 0.6; 95% CI 0.4-0.8.)
Conclusion: The impact of insurance on cancer outcomes is controversial and affects regional and national policy. We find that insurance status is associated with both: 1) cancer stage at presentation and 2) likelihood of treatment. Subsequently, uninsured patients demonstrate vastly inferior survival. Particularly in such an aggressive disease, ensuring that patients are diagnosed earlier and that patients receive all indicated care is essential. Further investigation of the drivers of these insurance-associated disparities is warranted.