Identifying Predictors of Long Length of Stay and Mortality for General Surgery Patients in Surgical Intensive Care Units: A Comparative Study

Abstract

Objective: Approximately 5 million patients are admitted to intensive care units (ICUs) annually in the United States, with 500,000 of these patients dying per year during their admission. We aim to provide unadjusted rates of complications and mortality in a cohort of surgical ICU patients and to compare outcomes between a large, urban, academic referral center and community hospitals using a risk-adjusted model. Methods: We constructed a retrospective cohort of patients who underwent general surgery procedures and stayed in the surgical ICUs at Northwestern Memorial Hospital (NMH), Central DuPage Hospital (CDH), and Lake Forest Hospital (LFH) from March 2018-April 2021. Patients at the latter suburban community hospital sites were grouped. Patients treated at NMH versus CDH and LFH were compared using univariate analyses of complications and mortality and multivariable analyses adjusted for patient clinical and sociodemographic characteristics. Results: Univariate analysis demonstrated that an ICU stay at NMH was associated with higher rates of hyperglycemia (45.8% vs. 30.7%, p <0.001), malnutrition (33.3% vs. 20.5%, p<0.001), and wound infections (8.9% vs 4.9%, p = 0.03), but lower mortality rates (4.7% vs 8.8%, p = 0.006). BMI, ASA score, surgery type, and having subsequent procedures were the characteristics most commonly associated with adverse events. Multivariate analysis demonstrated that hospital site was not significantly associated with long length of stay (OR = 1.1 [0.60-2.03]) but being treated at CDH/LFH was associated with higher mortality (OR = 2.34 [1.16-4.73]). Discussion: These data provide baseline rates of adverse outcomes in a large cohort of surgical ICU patients, and show that mortality, but not length of stay, may differ between the surgical ICUs studied. Our results provide a foundation for future study of surgical ICU outcomes within the Northwestern Healthcare network and may be hypothesis-generating for additional research within this population.

Files

Additional details