The epidemiology of bloodstream infection contributing to mortality: the difference between community-acquired, healthcare-associated, and hospital-acquired infections

Abstract

Abstract Background The epidemiology of bloodstream infection (BSI) is well-established; however, little is known about the contribution of different pathogens to mortality. To understand true burden of BSI, pathogens contributing to mortality were investigated and compared according to where the BSI was acquired. Methods Data from deceased patients in two teaching hospitals in the Republic of Korea were collected. BSI contributing mortality was defined as BSI within 2-weeks before death. Cases were grouped by acquisition sites: community-acquired (CA)-, healthcare-associated (HCA)-, and hospital-acquired (HA)-BSI. Drug resistance, BSI focus, and appropriateness of empirical antimicrobial therapy were also compared. Results Among 1849 deceased patients in the hospitals, 280 (15.1%) patients experienced BSI within 2-weeks before death. In all, 71, 53, and 156 patients in the CA-, HCA-, and HA-BSI groups, respectively, with 316 total isolated pathogens were analyzed. The three most common pathogens were Klebsiella pneumoniae (17.1%), Escherichia coli (16.4%), and Staphylococcus aureus (11.4%). While K. pneumoniae and E. coli were the most common pathogens in CA- and HCA-BSI, Acinetobacter baumannii and Candida species were in HA-BSI. 26.3% (41/156) of patients experienced breakthrough HCA-BSI during administration of carbapenem and/or vancomycin. The proportion of central venous catheter-related infection (0%, 3.4% and 28.3%), carbapenem resistant-Gram negative bacilli (0%, 6.9% and 21.9%), and inappropriate empirical antimicrobial therapy (21.1%, 37.7% and 51.9%; all P < 0.001) were more frequently observed in HA-BSI. Conclusion The epidemiology of BSI related to mortality had unique characteristics according to the acquisition site. Given the epidemiology of HA-BSI, infection control and antibiotics stewardship programs should be emphasized.