The Hispanic Paradox in endometrial cancer: A National Cancer Database study

Objective. To compare the overall survival of non-Hispanic white and Hispanic women with endometrial cancer. Methods. We performed an observational retrospective cohort study of Hispanic and non-Hispanic women with endometrial cancer from the 2004-2014 National Cancer Database. Baseline characteristics were compared with the Chi-squared test for categorical variables or the Mann-Whitney U test for ordinal or continuous variables. The Kaplan-Meier method was used to estimate unadjusted survival times, which were compared with the log-rank test. Missing data was imputed using multiple imputation with chained equations. A multivariable parametric accelerated failure time model for survival was used. Sensitivity analyses were performed using matched cohort analyses of the overall cohort, and of subgroups based on stage or type. Results. 112,574 non-Hispanic and 6313 Hispanic women met inclusion criteria. Five-year survival was slightly higher for Hispanic women (83.1% (82.1-84.3%) versus 81.4% (81.2-81.7%), P = 0.002). Hispanic women were younger, treated at lower volume hospitals, and more often diagnosed with a type II histology and stage II-IV disease compared to non-Hispanic women (all P < 0.001). With multivariable adjustment for measured confounders, Hispanic women lived 8% longer than non-Hispanic women (time-ratio (95% CI) 1.08 (1.02-1.14), P = 0.01). When bias-reducing matched cohort analyses were used for sensitivity analyses, Hispanic women did not have significantly different survival than non-Hispanic women. Conclusion. Hispanic ethnicity was not associated with a clinically meaningful difference in survival among women with endometrial cancer. (C) 2017 Elsevier Inc. All rights reserved.