Rachel Kohn, MD, MSCE
Assistant Professor of Medicine
University of Pennsylvania
Philadelphia, Pennsylvania, United States
Disclosure information not submitted.
Jennifer Ginestra, MD
Hospital of the University of Pennsylvania
Philadelphia, Pennsylvania
Disclosure information not submitted.
Rebecca Hubbard, PhD
Professor of Biostatistics
University of Pennsylvania, United States
Disclosure information not submitted.
Catherine Auriemma, MD, MSHP
Fellow
University of Pennsylvania, United States
Disclosure information not submitted.
Mitesh Patel, MD, MBA, MS
VP, Clinical Transformation
Ascension Health, United States
Disclosure information not submitted.
George Anesi, MD
Assistant Professor of Medicine, University of Pennsylvania
University of Pennsylvania
Philadelphia, Pennsylvania
Disclosure information not submitted.
Andrew Crane-Droesch, PhD
Data Scientist
University of Pennsylvania, United States
Disclosure information not submitted.
Scott Halpern, MD, PhD
Professor of Biostatistics
University of Pennsylvania, Pennsylvania, United States
Disclosure information not submitted.
Meeta Kerlin, MD, MS
MD
Perelman School of Medicine, University of Pennsylvania, United States
Disclosure information not submitted.
Gary Weissman, MD, MSHP
Assistant Professor of Medicine
University of Pennsylvania, United States
Disclosure information not submitted.
Title: Association of time of day with antimicrobial start for ward patients with hospital-acquired sepsis
Introduction:
Sepsis is a top cause of mortality among hospitalized patients. Sepsis treatment, the timely initiation of broad-spectrum antimicrobials, is often delayed, especially among patients with hospital-acquired sepsis (HAS). Time of day is associated with suboptimal outpatient medical care, but its association with treatment initiation in HAS is unknown. We aimed to test the association of time of day with antimicrobial initiation among ward patients with HAS. We hypothesized that handoffs at shift changes and time from shift start would be associated with antimicrobial initiation delays.
Methods:
Retrospective cohort study of HAS episodes among patients admitted 7/2017-12/2019 in 5 hospitals. The exposure was time of day (reference: 7AM, workday start). We used discrete-time time-to-event models to estimate probability of antimicrobial initiation at each hour among patients with HAS, defined by CDC Adult Sepsis Event (ASE) criteria. We adjusted for hospital, admission year, quarter, age, gender, race, ethnicity, admission diagnosis category, and medical vs surgical service. In a secondary analysis, we used quantile regression to test the association of time of day of HAS onset with hours to antimicrobial initiation.
Results:
Among 222,102 hospitalizations, 1,672 (0.8%) had HAS. Patients with HAS were median 63 years old (IQR 54-71), 910 (54%) were male, and 1,372 (82%) were on medical services. Median time to antimicrobial initiation was 4.1 hours (IQR 0.4-22.3). Marginally-adjusted hourly probabilities of antimicrobial initiation ranged from 2.1% (95% CI 1.3% to 3.0%) at 7AM to 10.2% (95% CI 8.4% to 12.1%) at 6PM, with nadirs at 7AM and 7PM. Probabilities also declined as the night shift progressed from 9.8% (95% CI 8.0% to 11.7%) at 9PM to 2.3% (95% CI 1.4% to 3.2%) at 6AM. Time to antimicrobial initiation was median 2.7 hours (IQR 0-14.0) during the day shift, and 8.3 hours (IQR 1.5-34.3) during the night shift.
Conclusions:
Antimicrobial initiation declined at shift changes and overnight among ward patients with HAS. These findings show that the time of day is associated with treatment patterns for ward patients with HAS. Future work should validate these findings in other settings and elucidate the staffing, cognitive, and organizational mechanisms in order to develop quality-enhancing interventions.