Date of Award

January 2020

Document Type

Open Access Thesis

Degree Name

Master of Public Health (MPH)


School of Public Health

First Advisor

Chima D. Ndumele


Objective: Although racial and gender disparities in healthcare treatment have been widely studied, relatively few studies have focused on the critically ill population and have had conflicting results. In this study, we sought to assess the associations between a patient’s race and gender and their ICU mortality, length of stay, and any observed delays in the discharge process.

Methods: We utilize a large single-center database called MIMIC-III to analyze the association be- tween a patient’s race and gender and the outcomes of hospital mortality, ICU length of stay, and delays in discharge process. All analysis is conducted using R. Multivariable logistic regression is used for the outcome of hospital mortality and multivariable linear regression is used for the outcomes of ICU length of stay and delays in discharge. The statistical level of significance is set at alpha = 0.05.

Results: Without adjusting for any covariates, compared to White Males, Asian Females had a lower ICU length of stay, White Females had a higher odds of hospital mortality, Hispanic/Latino Males and Females had a lower odds of hospital mortality, and White and Black/African-American Females had a longer delay in discharge. Upon adjusting for several covariates, White Females and Black/African- American Females had an ICU length of stay that was on average lower than White Males and Black/African-American Females had on average a lower odds of hospital mortality than White Males. The significant association seen for Asian Females for ICU length of stay and White Females and Hispanic/Latino Males and Females for hospital mortality disappeared. There was also no significant racial and gender difference for delays in discharge when controlled for other covariates.

Conclusion: The racial and gender differences for the outcomes of ICU length of stay and hospital mortality may indicate difference in treatment patterns or could also be attributed to other factors such as difference in treatment preferences. However, this study has assessed the associations between each race and gender category for three different healthcare outcomes in a classically understudied population. The significant results obtained here can serve as an important cornerstone to motivate future research on health disparities in the critically ill population.


This is an Open Access Thesis.

Open Access

This Article is Open Access