Multi-level Social Determinant Assessments of Skull Base Squamous Cell Carcinoma Treatment Disparities in the United States

Objectives: Utilizing a national patient population of skull-base squamous cell carcinoma patients (SBSC) and multilevel social determinant of health (SDoH) models comprising individual- and community-level factors, including the Yost-SES-Index, to observe how multilevel social determinant factors associate with SBSC-treatment disparities across the US. Design, Setting, Participants: Retrospective cohort study assessed SBSC-patients between 2010-2018 from SEER were analyzed by age-adjusted multivariable, multilevel logistic regression models utilizing covariates of individual-level sex & race/ethnicity and census-level rurality-urbanicity and Yost-SES-Index (composite measure of poverty, education, income, housing) were logistically regressed, stratified across non-/nasopharyngeal-SBSC types Main Outcome Measures: Surgical treatment, radiation therapy, chemotherapy, and delay-in-treatment-initiation (3-or-more months after diagnosis) outcomes Results Across 8852 patients, lower Yost-SES-Index (OR 0.73, 95%CI 0.63-0.85) was a negative predictor of non-nasopharyngeal-SBSC-surgery. For radiation therapy receipt, female sex (1.33, 1.16-1.54) was a non-nasopharyngeal-SBSC positive predictor while lower Yost-SES-Index (0.75, 0.64-0.87) was a nasopharyngeal-SBSC negative predictor. For chemotherapy, minority race/ethnicity was a non-nasopharyngeal-SBSC positive predictor (1.25, 1.07-1.46) whereas lower Yost-SES-Index (0.85, 0.73-0.98) was a nasopharyngeal-SBSC negative predictor. Finally, female sex (1.40, 1.17-1.69), minority race/ethnicity (1.30, 1.08-1.57), and lower Yost-SES-Index (1.49, 1.24-1.79) were markedly positive predictors for delays-in-treatment. Conclusions For SBSC, minority race/ethnicity, female sex, and community-level SES were the greatest independent factor associated with treatment disparities. Rurality-urbanicity was not significantly associated with treatment disparities. Multilevel modeling objectively assessed the associative effects of each factor for conferring treatment disparities.