Effects of individual and neighborhood socioeconomic status on outcomes following a colorectal cancer diagnosis

Evidence indicates inequalities in socioeconomic status (SES), especially those that result in differential access to early detection as well as to high-quality medical and supportive care, as the primary drivers of persistent disparities in outcomes following colorectal cancer (CRC) diagnosis. Yet research is limited on the independent and joint effects of individual SES (iSES) characteristics and neighborhood SES (nSES) on the likelihood of survival and health-related quality of life (HRQoL) following a CRC diagnosis. Further, no prior study of CRC outcomes has attempted to disentangle the effects of area-level SES from the effects of neighborhood contextual factors (e.g., population density, level of high-intensity development). We used data from 3949 incident CRC cases (51% women) diagnosed between 1997-2018, who lived in the 13-county catchment area of the population-based Western Washington Surveillance, Epidemiology and End Results (SEER) cancer registry, and who participated in the Puget Sound Colorectal Cancer Cohort (PSCCC). We first focused on assessing the independent and joint effects of two self-reported iSES factors (i.e., educational attainment and household income) and census block group-level nSES on all-cause and disease-specific survival following a CRC diagnosis. We performed both overall and gender-specific analyses. In the same study population, we also evaluated nSES effects as well as the effects of two other neighborhood factors, population density and high-intensity development level, when all three factors were included in the same analytic model. Finally, using a subset of this study population – cases diagnosed between 2016-2018 – we investigated the possible independent and joint effects of three iSES characteristics (i.e., self-reported educational attainment and household income, SEER-reported insurance type) and nSES on overall HRQoL in individuals recently diagnosed with CRC. After a median 4.0 years of follow-up, 1591 cases died (844 due to CRC). After adjusting for iSES factors and nSES, lower household income was associated with poorer all-cause survival (p-trend: 0.04), especially in women (p-trend: <0.01); lower income effects in women were largest given the context of living in a low nSES neighborhood (HR: 1.61, 95% CI: 1.27-2.04). With respect to CRC-specific survival, lower educational attainment was related to higher mortality in men (p-trend 0.03) and living in a low nSES neighborhood was modestly associated with poorer survival in the overall study population (HR: 1.36, 95% CI: 1.02-1.80). After additionally adjusting estimates for neighborhood population density and the neighborhood level of high-intensity development, living in a high nSES neighborhood remained marginally related to better CRC-specific survival (p-trend: 0.05). In addition, living in a densely populated neighborhood was associated with lower disease-specific mortality (p-trend: 0.02). When we stratified estimates by iSES factors, living in a high nSES neighborhood was modestly associated with survival only in cases reporting higher amounts of educational attainment; living in an intensely developed neighborhood was related to poorer survival only in cases reporting lower household income. After adjusting for iSES factors and nSES, reporting a lower household income (p-trend:<0.01) and using Medicaid insurance versus private or other government-based insurance (p-value: <0.01) were each statistically and clinically significantly related to lower overall HRQoL after a CRC diagnosis. Low income effects were largest in the context of low nSES (B: -12.57, 95% CI: -16.90, -8.24), while Medicaid effects were roughly equivalent in high nSES (B: -9.81, 95% CI: -15.38, -4.25) and low nSES neighborhoods (B: -9.79, 95% CI: -15.66, -3.92). Our findings indicate that the impacts of SES on outcomes following a CRC diagnosis are likely a mixture of iSES and nSES effects, and that there is likely something uniquely harmful about living in a low nSES neighborhood even after adjusting for other neighborhood contexts. Future research should investigate these associations in other study populations and geographic locations and also work to formally identify mediating mechanisms.