Bloodstream infections (BSIs) are severe infections caused by the presence of microbes in blood [1]. For BSIs, the standard method for bacterial identification is blood culture (BC), but there are several practical problems. Most significantly, BC takes about half a day to several days to positivity and more time to identify microbes. In addition, obtaining BC after starting antibiotics reduces the yield of positive culture [2,3].

Various methods for identifying bacteria have been investigated to compensate for the problems with BC. Mass spectrometry and DNA-microarray-based methods to analyze colonies obtained from a positive BC yield a sensitivity of 81%–100 % and a specificity of 81%–100 % with BC as the standard, but they require recovery of the organism from the BC [4,5]. Nucleic-acid amplification tests (non-culture diagnosis) have identification times of 3–12 h with a sensitivity of 37%–90 % and a specificity of 66%–92 % [5]. However, they are limited in the number of identifiable microbes, and extensibility for novel strains.

The melting temperature (Tm) mapping method (TM) was reported as a novel method for microbial identification [6]. In TM, bacterial DNA directly extracted from whole blood is used for identification of bacteria by the intrinsic patterns of Tm values in the bacterial-specific 16S ribosomal RNA gene (16S rDNA). A previous study for adults with BSI has shown that TM has high match rates with BC (positive-match rate, 85 % and negative-match rate, 98 %) and may be more sensitive than BC [6]. TM was also shown to be rapid and expandable in detectable bacterial species [6]. However, data on TM in children with BSI are limited. In this study, to examine potential clinical application of TM in children with BSI, we investigated the match rates of pathogenic bacteria identified by TM and BC, the identification times, and the amount of bacterial DNA in blood samples using TM at a children's hospital.