• T N Jang, B I Kuo, S H Shen, C P Fung, S H Lee, T L Yang, and C S Huang.
• Department of Medicine, Shin Kong Wu Ho-Su Memorial Hospital, Taipei, Taiwan.
• J Formos Med Assoc. 1999 Jul 1;98(7):465-73.

AbstractAlthough gram-positive organisms are the most common causes of nosocomial bloodstream infections, gram-negative bacteremia carries higher risks of severe sepsis, septic shock, and death among critically ill patients in intensive care units (ICUs). We performed a prospective epidemiologic analysis of nosocomial gram-negative bacteremia episodes among ICU patients and sought to identify risk factors for mortality among these patients. All episodes of nosocomial gram-negative bacteremia documented in five ICU wards of our hospital during a 2-year period were included. There were 147 episodes (124 patients) of gram-negative bacteremia documented during the study period. The overall mortality rate was 36.1%, and 77.4% of all deaths were directly related to the bloodstream infection. Gram-negative bacteremia was associated with prolonged ICU stay (45.7 d vs 6.1 d for all ICU patients). The most common isolate was Acinetobacter baumannii, followed by Burkholderia cepacia and Enterobacter cloacae. The most frequent source of infection was the lower respiratory tract (32.0%). Of the agents tested, ciprofloxacin, imipenem, and ceftazidime were the most active against the clinical isolates. Multivariate logistic regression analysis identified the presence of septic shock (odds ratio, OR = 17.66, p < 0.001) and rapidly fatal and ultimately fatal underlying conditions (OR = 3.47, p = 0.032) as being independent risk factors for mortality. Early appropriate antibiotic treatment did not result in significant improvement in survival. These findings suggest that prevention of lower respiratory tract colonization and nosocomial pneumonia are crucial for reducing the incidence of nosocomial gram-negative bacteremia in the ICU. Serious underlying illnesses and septic shock were the most important risk factors for death in these patients.

24 keywords...

hide…