Secondary Logo

29

THE EPIDEMIOLOGY OF HOSPITAL-ONSET SEPSIS USING CLINICAL DATA FROM 136 U.S. HOSPITALS

Rhee, Chanu1; Wang, Rui2; Zhang, Zilu3; Kadri, Sameer4; Fram, David5; Jin, Robert2; Klompas, Michael6

doi: 10.1097/01.ccm.0000550821.37919.f5
Star Research Presentations: Sepsis
Free

1Harvard Medical School/Harvard Pilgrim Health Care Institute, Boston, MA

2Harvard Pilgrim Health Care Institute, Boston, MA

3Harvard Medical School, Boston, MA

4N/A, Bethesda, MD

5Commonwealth Informatics, Waltham, MA

6N/A, Boston, MA

Learning Objectives: Relatively little is known about hospital-onset (HO) sepsis. Most prior epidemiologic studies of sepsis have used administrative data, which have variable accuracy, do not reliably differentiate between community-onset (CO) vs HO disease, and lack granular clinical information.

Methods: We identified sepsis among adult patients hospitalized in 2009–2015 by applying CDC Adult Sepsis Event criteria (which requires clinical indicators of presumed infection and concurrent organ dysfunction) to detailed electronic health record data from 136 U.S. hospitals in the Cerner HealthFacts dataset. We used logistic and Cox regression models with time varying covariates to compare the risk of in-hospital death for HO-sepsis (defined by blood cultures, antibiotics, and organ dysfunction on hospital day ≥3) vs patients with CO-sepsis (blood cultures, antibiotics, or organ dysfunction on hospital day <3) and without sepsis, controlling for demographics, comorbidities, infectious diagnoses, and severity-of-illness at sepsis onset or on admission.

Results: The cohort included 2.3 million patients. Of these, 97,352 had sepsis of which 11,782 (12.1%) was HO-sepsis. Compared to CO-sepsis, HO-sepsis patients were younger (median age 66 vs 68) but had more comorbidities (median Elixhauser score 14 vs 11) including heart failure (26 vs 22%), renal disease (23 vs 20%), and cancer (17 vs 11%). HO-sepsis patients had higher SOFA scores at sepsis onset (median 4 vs 3), higher rates of intra-abdominal infections (20 vs 15%), more positive blood cultures (26 vs 21%), longer hospital length-of-stay (median 19 vs 8 days) and ICU length-of-stay (median 6 vs 4 days), and higher in-hospital mortality (34 vs 17%) (p<0.001 for all). On multivariate analysis, HO-sepsis was associated with higher risk of hospital death vs CO-sepsis (odds ratio 2.10, 95% CI 2.08–2.12) and vs patients without sepsis (hazard ratio 3.02, 95% CI 2.99–3.04).

Conclusions: HO-sepsis accounts for 1 in 8 sepsis cases, increases the risk of death 3-fold, and carries mortality rates twice as high as CO-sepsis. Even after controlling for comorbidity burden and severity-of-illness, HO-sepsis is associated with a higher risk of death than CO-sepsis. This may be due to residual confounding or differences in quality of care. Further studies are needed to elucidate these factors.

Copyright © by 2019 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.