Abstract
BACKGROUND: Early detection and treatment of Gram-negative bacteria (GNB), major causative pathogens of sepsis (a potentially fatal condition caused by the body's response to an infection), may benefit a patient's outcome, since the mortality rate increases by 5-10% for each hour of delayed therapy. Unfortunately, GNB diagnosis is based on bacterial culture, which is time consuming. Therefore, an economic and effective GNB (defined as a positive blood, sputum, or urine culture) infection detection tool in the emergency department (ED) is warranted. METHODS: We conducted a retrospective cohort study in the ED of a university-affiliated medical center between January 01, 2014 and December 31, 2017. The inclusion criteria were as follows: (1) age ≥ 18; (2) clinical suspicion of bacterial infection; (3) bacterial culture from blood, sputum, or urine ordered and obtained in the ED. Descriptive statistics was performed on patient demographic characteristics, vital signs, laboratory data, infection sites, cultured microorganisms, and clinical outcomes. The accuracy of vital signs to predict GNB infection was identified via univariate logistic regression and receiver operating characteristic (ROC) curve analysis. RESULTS: A total of 797 patients were included in this study; the mean age was 71.8 years and 51.3% were male. The odds ratios of patients with body temperature ≥ 38.5 °C, heart rate ≥ 110 beats per minute, respiratory rate ≥ 20 breaths per minute, and Glasgow coma scale (GCS) < 14, in predicting GNB infection were found to be 2.3, 1.4, 1.9, and 1.6, respectively. The area under the curve values for ROC analysis of these measures were 0.70, 0.68, 0.69, and 0.67, respectively. CONCLUSION: The four physiological parameters were rapid and reliable independent predictors for detection of GNB infection.