Impact of Repeating Blood Cultures on Detecting Unpredicted Causative Microorganisms

Background:  Blood cultures (BCs) are important for diagnosis and management of patients with infection. However, the impact of repeating BCs have not elucidated yet. Our objective was to assess the difference between initial and repeated BCs results over the past decade.

Methods:  We retrospectively analyzed all consecutive BCs performed episodes at the Mie University Hospital from January 2006 till December 2015. All patients with repeating BCs were recruited to this study. Any episode occurred 31 or more days after previous BC was considered as a separate condition. The timings of repeating BCs after initial episode (Day 0) were categorized as follows: Day 1-2, Day 3-5, and Day 6 or later. The following microorganisms were considered as probable contaminants: CNS, Bacillus species, Corynebacterium species, Propionibacterium species, and Micrococcus species. The frequencies of BCs’ positivity and distribution of detected microorganisms in the initial and repeated BCs were analyzed.

Results: Among a total of 12,593 cases with initial BCs, repeating BCs were performed in 4,136 cases. True positive rate (excluding probable contaminants) at Day 0, Day 1-2, Day 3-5, and Day 6 or later were 8.8%, 7.8%, 4.7%, and 7.2%, respectively. The pseudo-positive rates among each group were between 2.7-4.0%. The frequently detected microorganisms at initial BCs were as follows: CNS (25.0%), E. coli (13.0%), MSSA (9.4%), Streptococcus sp. (7.8%), Klebsiella sp. (7.1%),　MRSA (6.6%), and E. faecalis (5.4%). However, at Day 3-5, increasing frequencies of MSSA, MRSA, and E. faecalis detection and decreasing of E. coli, Streptococcus sp., Klebsiella sp. were observed. At day 6 or later, Candida sp. (6.0%) and E. faecium (5.7%) came up to the top 5 organisms following CNS (26.2%), MRSA (11.9%) and E. coli (7.9%). The results of repeated BCs among initial positive cases were same pathogen (21.5%) and new pathogen (11.1%). The positive rates of repeating BCs among initial pseudo-positive and negative cases were 10.2% and 9.8%, respectively.

Conclusion:  A certain number of repeating BCs cases showed different microorganisms from initial one. There were several changes in the distribution of detected microorganisms from initial BCs to repeating ones with chronologically increased frequencies of drug resistant organisms. These results suggest repeating BCs are needed in cases of unfavorable clinical course.