Trends in Medical and Surgical Admission Length of Stay by Race/Ethnicity and Socioeconomic Status: A Time Series Analysis

• Main Authors: Arnab K. Ghosh, Mark A. Unruh, Orysya Soroka, Martin Shapiro
• Format: Article
• Language: English
• Published: SAGE Publishing 2021-07-01
• Series: Health Services Research & Managerial Epidemiology
• Online Access: https://doi.org/10.1177/23333928211035581

Background: Length of stay (LOS), a metric of hospital efficiency, differs by race/ethnicity and socioeconomic status (SES) and longer LOS is associated with adverse health outcomes. Historically, projects to improve LOS efficiency have yielded LOS reductions by 0.3 to 0.7 days per admission. Objective: To assess differences in average adjusted length of stay (aALOS) over time by race/ethnicity, and SES stratified by discharge destination (home or non-home). Method: Data were obtained from 2009-2014 Healthcare Cost and Utilization Project State Inpatient Datasets for New York, New Jersey, and Florida. Multivariate generalized linear models were used to examine trends in aALOS differences by race/ethnicity, and by high vs low SES patients (defined first vs fourth quartile of median income by zip code) controlling for patient, disease and hospital characteristics. Results: For those discharged home, racial/ethnic and SES aALOS differences remained stable from 2009 to 2014. However, among those discharged to non-home destinations, Black vs White aALOS differences increased from 0.21 days in Q1 2009, (95% confidence interval (CI): 0.13 to 0.30) to 0.32 days in Q3 2013, (95% CI: 0.23 to 0.40), and for low vs high SES patients from 0.03 days in Q1 2009 (95% CI: -0.04 to 0.1) to 0.26 days, (95% CI: 0.19 to 0.34). Notably, for patients not discharged home, racial/ethnic and SES aALOS differences increased and persisted after Q3 2011, coinciding with the introduction of the Affordable Care Act (ACA). Conclusion: Further research to understand the ACA’s policy impact on hospital efficiencies, and relationship to racial/ethnic and SES differences in LOS is warranted.