Race Does Not Impact Sepsis Outcomes When Considering Socioeconomic Factors in Multilevel Modeling

Research output: Contribution to journalArticlepeer-review

Abstract

OBJECTIVES: To determine whether race is a major determinant of sepsis outcomes when controlling for socioeconomic factors. DESIGN: Retrospective cohort study. SETTING: Barnes-Jewish Hospital a 1,350 bed academic medical center. PATIENTS: Eleven-thousand four-hundred thirty-two patients hospitalized between January 2010 and April 2017 with sepsis and septic shock. INTERVENTIONS: Multilevel random effects modeling was employed whereby patients were nested within ZIP codes. Individual patient characteristics and socioeconomic variables aggregated at the ZIP code level (education, employment status, income, poverty level, access to healthcare) were included in the model. MEASUREMENTS AND MAIN RESULTS: In hospital mortality, length of stay, need for vasopressors, and mechanical ventilation were the main endpoints. Black patients had more comorbidities than White patients except for cirrhosis and malignancy. In unadjusted comparisons, White individuals were more likely to require mechanical ventilation and had higher mortality rates and longer hospital stays for both low- and high-income groups. When nesting within ZIP codes and accounting for socioeconomic variables, race did not have a significant effect on mortality. Non-White races had lower odds ratio for mechanical ventilation. CONCLUSIONS: Our study demonstrates that race is not an independent risk factor for sepsis mortality, as well as sepsis-related length of stay. We should expand our inquiry into determinants of sepsis outcomes by including socioeconomic variables.

Original languageEnglish
Pages (from-to)410-417
Number of pages8
JournalCritical care medicine
Volume50
Issue number3
DOIs
StatePublished - Mar 1 2022

Keywords

  • mortality
  • multilevel model
  • outcomes
  • race
  • sepsis

Fingerprint

Dive into the research topics of 'Race Does Not Impact Sepsis Outcomes When Considering Socioeconomic Factors in Multilevel Modeling'. Together they form a unique fingerprint.

Cite this