Sensitivity Analyses
We re-analyzed our survival models evaluating the relationship between
race/ethnicity and survival to evaluate whether the relationship varied
when limited to 5-year hazard of death. The Cox proportional hazard
model to determine the 5-year hazard of death (Supplemental Table S1)
showed only Hispanic patients having a significantly higher hazard of
death than non-Hispanic white patients in both the unadjusted model
(p<0.0001) and the adjusted model (p=0.0003). To a large
extent, 5- year hazard of death associated with histology type and
surgery modality did not differ significantly within each group while
significant differences were largely noted in the remaining covariates
in both the unadjusted and adjusted models except for hazard of death
associated with radiation, which was also non-significant. Additionally,
we found no significant interaction between race and histology type in
the survival model, indicating there is no differential impact of race
on survival by histology type.