RESUMO
BACKGROUND: Hispanic patients with esophageal cancer (EC) have racially disparate survival outcomes compared with white patients. OBJECTIVES: We explored the impact on survival of racial differences in socioeconomic factors, tumor characteristics, and rates of surgical utilization in patients with EC. METHOD: Using the SEER (Surveillance, Epidemiology, and End Results) registry, we identified 22,531 cases of EC in Hispanic and white patients between the ages of 18 and 65 years in 2003-2014. Of these, 6,250 cases had locoregional EC. Patients were categorized according to age, gender, education, tumor grade, histology, primary tumor site, and surgical status. Postdiagnosis survival was examined over time and compared by race and stratified by surgical status. RESULTS: Compared with whites, Hispanics with EC had significantly higher unadjusted mortality (hazard ratio [HR] 1.11; 95% confidence interval [CI] 1.06-1.17; p < 0.001) as did Hispanics with locoregional EC (HR 1.15; 95% CI 1.03-1.29; p = 0.01). In the multivariate analysis, several socioeconomic and tumor factors were found to be independently associated with survival by race, including county of residence income and prevalence of smoking, tumor grade, stage, and primary site, and surgical utilization. After adjusting for demographic and tumor characteristics, surgical utilization in patients with locoregional EC had a significant interaction with race on overall mortality (p = 0.01). Hispanics with locoregional EC were significantly less likely to receive surgery than whites (46 vs. 60%; p < 0.001) and not receiving surgery was associated with a significantly lower overall survival (HR 2.84; 95% CI 2.65-3.04; p < 0.001). CONCLUSIONS: A lower rate of surgery among Hispanics with potentially resectable esophageal cancer was associated with a decreased survival rate when compared to whites, even when adjusting for relevant socioeconomic and tumor factors. These data support the need to better address patient barriers to surgical treatment and the systemic biases present in medical care.