Survival outcomes by insurance coverage in non-Hodgkin lymphoma (NHL): Experience from two academic centers and a safety-net health system.

e19058 Background: The impact of insurance coverage on survival in NHL has been described, but it remains unclear how this relationship varies by care setting. We compared overall survival (OS) among NHL patients treated at two academic medical centers participating in the Lymphoma Epidemiology of Outcomes (LEO) Cohort Study (Emory University Hospital, GA; MD Anderson Cancer Center, TX) and at the Harris Health System (HHS), a safety-net system providing charity care to low-income uninsured and underinsured residents of Harris County, TX. Methods: We analyzed patients diagnosed between 2015 and 2025 at two LEO centers (N=2,864) and at HHS (N=726). We examined 5-year OS by care setting, insurance type, and NHL subtype (indolent: follicular lymphoma, marginal zone lymphoma; aggressive: diffuse large B-cell lymphoma, Burkitt/lymphoblastic lymphoma). We conducted separate Cox proportional hazards models for academic centers and HHS to assess associations between insurance coverage and OS. Covariates included age, sex, race/ethnicity, year of diagnosis, ZIP-level income, and NHL subtype. Results: At academic centers, 8% of patients were Black and 18% were Hispanic; at HHS, 18% were Black and 68% were Hispanic. Mean ZIP-level income was higher at academic centers ($106,916) than at HHS ($74,458). Across NHL subtypes and insurance categories, 5-year OS was higher among patients treated at academic centers than HHS for both indolent NHL (92% vs. 90%) and aggressive NHL (77% vs. 63%). At academic centers, OS was highest among privately insured patients (indolent: 96%; aggressive: 84%), followed by uninsured patients (89%; 78%) and Medicare beneficiaries (89%; 73%), and lowest among Medicaid patients (not reportable due to few observations; 62%). At HHS, OS was highest among privately insured patients (not reportable; 67%), followed by uninsured patients (93%; 66%), and lowest among Medicare beneficiaries (74%; 39%) and Medicaid patients (72%; 59%). In adjusted Cox models, Medicaid coverage was associated with poorer OS compared with private insurance at both academic centers (HR 4.03, p=0.025) and HHS (HR 2.23, p=0.009). Uninsured status was associated with poorer OS compared with private insurance at academic centers (HR 2.78, p=0.018) but not at HHS. Medicare coverage was associated with better OS than private insurance at academic centers (HR 0.51, p=0.001) but not at HHS. Conclusions: Insurance-related survival differences in NHL varied by care setting. Medicaid coverage was associated with poorer OS across both settings, identifying Medicaid patients as a persistently high-risk group. In contrast, uninsured patients had poorer OS at academic centers but not at HHS, suggesting that safety-net charity care may attenuate survival disparities among uninsured patients.