Non-Hodgkin's lymphoma survival by neighborhood disinvestment and socioeconomic composition

Background: Socioeconomic factors have been associated with adverse survival from non-Hodgkin's Lymphoma (NHL), but few studies have considered the role of neighborhood disinvestment. This study investigated associations among neighborhood socioeconomic composition, disinvestment (e.g., disrepair and deterioration), stage at diagnosis, and NHL survival. Methods: Age, sex, marital status, race-ethnicity, diagnosis date, residence at diagnosis, health insurance, stage at diagnosis (localized, regional, distant), and histologic subtype (Diffuse large B-Cell lymphoma, follicular, Chronic lymphocytic leukemia / small lymphocytic lymphoma, other B-cell, and T-cell) were from 1209 Franklin County, Ohio residents diagnosed with non-Hodgkins lymphoma between 2010 and 2019 as recorded in the Ohio Cancer Incidence Surveillance System. Neighborhood disinvestment was from a virtual neighborhood audit of 6 indicators - garbage, graffiti, dumpsters, building conditions, yard conditions, abandoned buildings - within 5,000+ Franklin County Google Streetview location-dates estimated at residential addresses and diagnosis dates using spatio-temporal models. Census tract-level, neighborhood socioeconomic composition was measured by the Yost index - a validated composite of seven socioeconomic factors annually assessed from the US American Community Survey - and linked to each NHL case by residential census tract and year of diagnosis. We fitted accelerated failure time models of all-cause and NHL-specific survival time as functions of neighborhood socioeconomics, disinvestment, and stage, adjusted for covariates. Survival time was right-censored at the earliest of death from other causes, loss to follow-up, or end of follow-up (12/31/2020). Results: Across 5308 person years, there were 362 all-cause and 163 NHL-specific deaths. Disinvestment and stage statistically interacted in models of all-cause but not NHL-specific deaths. Among those diagnosed regional stage, all-cause survival time decreased by 35% (95% CI: 7%, 55%), with each standard deviation increase in disinvestment after adjustment for covariates. There was no evidence of associations between disinvestment and all-cause survival among localized and distant stages, disinvestment and NHL-specific survival, and socioeconomic composition and NHL-specific survival. Conclusion: Both neighborhood socioeconomic composition and disinvestment among those diagnosed at regional stage are independently associated with shorter all-cause, but not NHL-specific, survival among NHL cases within a Midwestern Metropolis.