An antibiotic decision-making tool for patients with pneumonia admitted to a medical intensive care unit.

Pneumonia is a leading cause of death in medical intensive care units (MICUs). Delayed or inappropriate antibiotic therapy largely increases morbidity and mortality. Multidrug-resistant (MDR) micro-organisms are major reasons for inappropriate antibiotic use. Currently there is no good antibiotic decision-making tool designed for critically ill patients. The objective of this study was to develop a convenient MDR prediction scoring system for patients admitted to MICUs with pneumonia. A retrospective cohort study was conducted using databases and chart reviews of pneumonia patients admitted to a 30-bed MICU from 2012 to 2013. Forward logistic regression was applied to identify independent MDR risk factors for prediction tool development. A total of 283 pneumonia episodes from 263 patients with positive cultures from blood or respiratory secretions were recruited, of which 154 (54.4%) were MDR episodes. Long-term ventilation (OR = 11.09; P = 0.026), residence in a long-term care facility (OR = 2.50; P = 0.005), MDR infection/colonisation during the preceding 90 days (OR = 2.08; P = 0.041), current hospitalisation ≥2 days (OR = 1.98; P = 0.019) and stroke (OR = 1.81; P = 0.035) were identified as independent predictors for MDR pneumonia. The area under the ROC curve of this prediction tool was much higher than that of ATS/IDSA classification (0.69 vs. 0.54; P <0.001). The prediction accuracy of this tool with risk score ≥1 for MDR infections was 63.7%. This simple five-item, one-step scoring tool for critically ill patients admitted to the MICU could help physicians provide timely appropriate empirical antibiotics.