Empiric gram-negative antibiotics usage among critically ill patients across 13 ICUs of Nepal: A retrospective cohort study

Abstract

Background This study aims to investigate the associations between patient characteristics and the receipt of empiric broad-spectrum gram-negative antimicrobials among adult patients admitted to ICUs in Nepal. Methods A retrospective cohort study was conducted in 13 ICUs in Nepal between January 1st, 2020, and December 31st, 2022. Empiric antibiotic use was defined as receiving gram-negative antibiotics within the first two days of ICU admission without evidence of infection from culture. Patient characteristics between patients receiving narrow-spectrum antibiotics and broad-spectrum gram-negative antibiotics were compared. Continuous variables were compared using mean and standard deviation, with mean differences and 95% confidence intervals calculated. Categorical variables were compared using the chi-square test. A multi-level multivariable regression model assessed the relationship between selected variables and broad-spectrum antibiotic usage, treating admitting ICU as a random factor. Results Among 12,349 eligible patients, 8720 (70.6%) received Empiric gram-negative antibiotics. Of those, 3240 (37.15%) received empiric narrow-spectrum gram-negative antibiotics, and 5480 (62.8%) received Empiric broad-spectrum gram-negative antibiotics. Patients receiving broad-spectrum gram-negative antibiotics were older (57.4 vs 53.8 yrs); had higher mean heart rates (97.3 vs 89.5 bpm), respiratory rates (24 vs 21.9), a fraction of inspired oxygen ( 50% vs 40%), blood sugar levels (163.4 vs 153.9 mg/dl), total leucocyte counts (12,685 vs 10,647 cells/cu mm) and serum creatinine (2.06 vs 1.03 umol/l) compared to narrow spectrum antibiotics. Similarly, patients receiving broad-spectrum gram-negative antibiotics had lower systolic blood pressure (122.6 vs 126.5 mm Hg), partial pressure of oxygen (96.7 vs 105.3), and pH ( 7.33 vs 7.36) compared to narrow-spectrum antibiotics. On multivariable regression; patients with readmission, cardiac support, and mechanical ventilation were more likely to receive broad-spectrum antibiotics. Conclusion The rate of empiric broad-spectrum gram-negative antibiotics usage in intensive care units is significantly high and associated with features of increased severity of illness.