Treatment patterns of empirical antibiotic therapies for critically ill patients with sepsis: a cross-sectional study in Vietnam

Abstract

Background: A systematic description of patterns of empirical antibiotic therapies (EAT) and the factors driving these in critically ill patients with sepsis is lacking in resource-limited settings. We aimed to address this knowledge gap by exploring the EAT patterns and risk factors for deviations from the guidelines in EAT in a Vietnamese hospital.

Methods: We conducted a simple random-sampling cross-sectional study at the intensive care unit (ICU) using medical records of critically ill patients with sepsis. The outcomes were guideline-compliant and guideline-deviant EAT (GcEAT and GdEAT). GcEAT was full compliance (choice and dosage) with local protocol for sepsis management, while GdEAT was any treatment that was not GcEAT. We used descriptive statistics to present data. Risk factors for GdEAT were reported with adjusted odds ratio, 95% confidence interval (OR and 95% CI; using multivariable logistic regression) and E-value (for factors with significant associations).

Results: Among 93 medical records included for analysis (median age of 68, 41.9% being male, 24.7% being overweight-to-obese). Initiation through the concurrent administration of three antibiotic agents was reported in 37.6% of all cases (96.8% with pseudomonal coverage, 91.4% with anaerobic coverage, 34.4% with MRSA coverage, 49.5% with enterococcal coverage). GcEAT was detected in 62 records (66.7%, 95% CI 56.6% to 75.4%), while GdEAT was in 31 records (33.3%, 95% CI 24.6% to 43.4%). The specific patterns in the GdEAT group were: (1) broader/narrower antibacterial coverage (87.1%) and overlapping targets (12.9%) (for choice of EAT), (2) higher/lower doses (45.2%), longer/shorter dosing intervals (38.7%), and without therapeutic drug monitoring (16.1%) (for dosage of EAT). Acute kidney injury before ICU admission was identified as a potential risk factor for GdEAT (adjusted OR 3.41, 95% CI 1.16 to 10.01; E-value 3.10, lower bound of CI 1.37), which was mainly driven by deviations in the dosage of EAT (adjusted OR 3.45, 95% 1.18 to 10.08; E-value 3.12, lower bound of CI 1.39).

Conclusion: GdEAT was moderately prevalent in sepsis management of critically ill patients at a Vietnamese healthcare setting, primarily due to deviations in choice and dosage of the EAT. Further research and updated guidelines should address the optimal EAT in patients with acute kidney injury to avoid unstandardised deviations and noncompliances with antimicrobial protocols.

Clinical trial number: Not applicable.

Keywords: Critical illness; Empirical antibiotic; Perioperative care; Risk factors; Sepsis.

Fig. 1

Flowchart of the study subjects. Abbreviation: GcEAT, guideline-compliant empirical antibiotic therapy; GdEAT, guideline-deviant empirical antibiotic therapy. Some percentages may not equate to 100 due to rounding

Fig. 2

Suspected risk factors for GdEAT. Abbreviations: CI, confidence interval; ICU, intensive care unit; GcEAT, guideline-compliant empirical antibiotic therapy; GdEAT, guideline-deviant empirical antibiotic therapy; OR, odds ratio. We estimated the adjusted ORs using multivariable logistic regression. Variables other than sex and age were included based on the insights of the physicians, which were factors that could complicate the choice and dosage of antibiotics. The cut-offs of laboratory variables were derived from conventional practices, per local protocol. We chose the 2-day cut-off for categorising the length of stay before ICU admission to focus on hospital-acquired infections, i.e., within 48 h of hospitalisation. Underweight and overweight-to-obese statuses, per classification for Asians, were having body mass index < 18.5 kg/m² and ≥ 23 kg/m² [16], respectively. Acute kidney injury was detected using the definition of KDIGO [18]