Introduction: There have been conflicting data regarding the relationship between sepsis-bundle adherence and mortality. Moreover, little is known about how this relationship may be moderated by the anatomic source of infection or the location of sepsis declaration.
Methods: This was a multi-center, retrospective, observational study of adult patients with a hospital discharge diagnosis of severe sepsis or septic shock. The study included patients who presented to one of three Los Angeles County Department of Health Services (DHS) full-service hospitals January 2012 to December 2014. The primary outcome of interest was the association between sepsis-bundle adherence and in-hospital mortality. Secondary outcome measures included in-hospital mortality by source of infection, and the location of sepsis declaration.
Results: Among the 4,582 patients identified with sepsis, overall mortality was lower among those who received bundle-adherent care compared to those who did not (17.9% vs. 20.4%; p=0.035). Seventy-five percent (n=3,459) of patients first met sepsis criteria in the ED, 9.6% (n=444) in the intensive care unit (ICU) and 14.8% (n=678) on the ward. Bundle adherence was associated with lower mortality for those declaring in the ICU (23.0% adherent [95% confidence interval {CI} {16.8-30.5}] vs. 31.4% nonadherent [95% CI {26.4-37.0}]; p=0.063), but not for those declaring in the ED (17.2% adherent [95%CI {15.8-18.7}] vs. 15.1% non-adherent [95% CI {13.0-17.5}]; p=0.133) or on the ward (24.8% adherent [95% CI {18.6-32.4}] vs. 24.4% non-adherent [95% CI {20.9-28.3}]; p=0.908). Pneumonia was the mostcommon source of sepsis (32.6%), and patients with pneumonia had the highest mortality of all other subsets receiving bundle non-adherent care (28.9%; 95% CI [25.3-32.9]). Although overall mortality was lower among those who received bundle-adherent care compared to those who did not, when divided into subgroups by suspected source of infection, a statistically significant mortality benefit to bundle-adherent sepsis care was only seen in patients with pneumonia.
Conclusion: In a large public healthcare system, adherence with severe sepsis/septic shock management bundles was found to be associated with improved survival. Bundle adherence seems to be most beneficial for patients with pneumonia. The overall improved survival in patients who received bundle-adherent care was driven by patients declaring in the ICU. Adherence was not associated with lower mortality in the large subset of patients who declared in the ED, nor in the smaller subset of patients who declared in the ward.
