A Comparison of Time to Sepsis Alert and Antibiotics in Emergency Department Patients by Language Preference

Abstract

Background

Prior studies have found increases in sepsis-related mortality among patients with non-English language preference (NELP) vs. English language preference (ELP) in the inpatient setting, even after controlling for demographic, illness severity, and clinical encounter variables. It is unclear if disparities in sepsis care extend to the emergency department.

Objective

We compared time to sepsis alert and antibiotics between patients with NELP vs. ELP overall and by emergency severity index (ESI) subgroup.

Methods

In this retrospective observational cohort study, the independent variable was preferred language, and the dependent variables were time to sepsis alert (clinician-initiated vs. automatic) and antibiotics. We developed multivariable adjusted plus inverse probability of treatment weight (IPTW) Cox proportional hazard models.

Results

There were no overall differences in the time to sepsis alerts or time to antibiotic administration. In the ESI 2 subgroup, NELP patients had faster time to sepsis alert than ELP patients in clinician-initiated alerts (clinician-initiated: absolute difference: 17 min; covariate +IPTW: HR: 1.76, 99.44% CI 1.22, 2.54, p < 0.001). Within the ESI 3 to 5 subgroup, among patients with a clinician-initiated alert, NELP patients had slower time to sepsis alert than ELP patients (absolute difference: 17 min; covariate +IPTW: HR: 0.64, 99.44% CI 0.44, 0.94, p = 0.012).

Conclusion

While we found no overall differences in time to sepsis alert, there were significant differences in subgroups, suggesting that language preference may be associated with triage score assignment and sepsis recognition. The lack of differences observed in time to antibiotics underscores the potential for thoughtfully designed best practice alerts to support equitable clinical practice.