Using a life table to quantify survival-related selection bias in studies of racial health disparities

Selection bias is one of the major threats to validity in epidemiologic studies. Survival-related selection bias occurs when people who died before the study cannot be recruited for participation in the study. Previous studies used real or simulated data to quantify survival-related selection bias in racial disparity analyses but reported various findings. Using a life table of Black, Hispanic, and white people born alive in the United States, 2022, and cumulative mortality rate ratio as an approximate of annual mortality rate ratio, we estimated that survival-related selection bias caused a 10.2% underestimation in Black-white all-cause mortality rate ratio and a 6.7% underestimation in Hispanic-white all-cause mortality rate ratio. We demonstrate that life tables can be used to quantify survival-related selection bias in racial disparity analyses of all-cause mortality. This method may be adaptable for other outcomes (e.g., diabetes) and in other disparity analyses (e.g., sex).