Early Empiric Antibiotic Use in Patients Hospitalized With COVID-19: A Retrospective Cohort Study*
Authors' conclusion
Although treatment of COVID-19 patients with empiric antibiotics has declined during the pandemic, the frequency of use remains high. There is significant inter-center variation in antibiotic prescribing practices and evidence of potential harm.
Study context
- Population
- Patients hospitalized with COVID-19, 66 US health systems in the N3C database
- Participants
- 322,867 first hospitalizations
- Exposure
- Early empiric antibiotic use vs no early antibiotic use
- Early empiric antibiotic use (EEAU)n = 43,089
- Outcome
- Early empiric antibiotic use over time
Methodological detailspropensity-matched, multicentre (66 health systems)
- Propensity-matched
- Multicentre (66 health systems)
Main finding
As reported in the abstract
- Antibiotic use in March 2020
- 23%
- Antibiotic use in June 2022
- 9.6%
- Significant inter-center variation in EEAU (deviance explained 7.33% vs 20.0%)
- p < 0.001
Observational data: associations, not effects.
Further findings
Factors associated with EEAU
| OR (95% CI) | ||
|---|---|---|
| Mechanical ventilation before day 2 | 3.57 (3.42–3.72) | |
| ECMO before day 2 | 2.14 (1.75–2.61) | |
| Early vasopressor use | 1.85 (1.78–1.93) | |
| 0.20.5125Odds ratio (log scale) |
| Outcome | OR (95% CI) | |
|---|---|---|
| In-hospital mortalityAfter propensity matching | 1.27 (1.23–1.33) | |
| Prolonged mechanical ventilationAfter propensity matching | 1.65 (1.50–1.82) | |
| 0.512Odds ratio (log scale) |
Factors associated with EEAU
- Region of residence not associated with EEAU
Safety
| Outcome | OR (95% CI) | |
|---|---|---|
| Late broad-spectrum antibiotic exposureAfter propensity matching | 3.24 (2.99–3.52) | |
| Late Clostridium difficile infectionAfter propensity matching | 1.60 (1.37–1.87) | |
| 0.20.5125Odds ratio (log scale) |
Original publication
Early Empiric Antibiotic Use in Patients Hospitalized With COVID-19: A Retrospective Cohort Study*. Crit Care Med. Published 1 May 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →