National Healthcare Safety Network (NHSN) criteria are used to identify reportable central line-associated blood stream infections (CLABSIs) in the United States. When normal skin flora are implicated, designation of a bloodstream infection as a CLABSI requires the recovery of the same organism from 2 or more blood specimens drawn on separate occasions. 1 , 2 This NHSN criterion is greatly dependent on the accuracy of the organism identification method used.
Matrix-associated laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) has emerged as an efficient and more accurate identification method than traditional biochemical testing.Reference Patel 3 We investigated the impact of MALDI-TOF MS on the assignment of CLABSI status when commensal organisms are involved.
METHODS
We queried the TheraDoc system at The Children’s Hospital of Philadelphia to identify CLABSIs meeting NHSN criteria from June 1, 2013, to June 30, 2015. Only CLABSIs involving organisms on the NHSN “Common Commensals” list were included. 2 Cases were deidentified and randomly assigned case identification numbers.
For each CLABSI, all isolates from the “infection window period” were included and retrieved from frozen storage. 2 Each isolate underwent identification by MALDI-TOF MS using Vitek MS (bioMerieux, Durham, NC) and the RUO MALDI Biotyper with software version 3.0 (Bruker Daltonics, Billerica, MA). The Vitek 2 GP identification card (bioMérieux), the BD Phoenix PID panel (Becton Dickinson Diagnostics, Sparks, MD), and the MicroScan Pos Breakpoint Combo 20 panel (Siemens, Princeton, NJ) were performed for biochemical identification. Confidence scores >80% for Vitek MS (50/50 for Streptococcus mitis/S. oralis); ≥2.0 for the Biotyper; and ≥90% for BD Phoenix, Vitek 2, and MicroScan were considered acceptable. All suspected S. viridans group streptococci (VGS) underwent optochin susceptibility and bile solubility testing, PCR to amplify a 761-base-pair segment of the tuf gene, followed by Sanger sequencing (see Online Supplementary Material).Reference York, Traylor, Hardy and Henry 4 , Reference Picard, Ke and Boudreau 5
RESULTS
In all 13 cases involving coagulase-negative Staphylococcus (CoNS), the identification of the isolate from the first positive blood culture matched that from subsequently positive blood cultures by all 5 identification methods. Overall, 11 cases were identified as Staphylococcus epidermidis by all methods. Isolates from case #7 were identified as S. hominis by all methods except MicroScan. Isolates from case #8, were identified as S. epidermidis by all methods except the BD Phoenix method.
Overall, 8 cases involved VGS (Table 1). Gold standard tuf sequencing yielded >99% homology with Streptococcus mitis and/or S. pneumoniae and/or S. pseudopneumoniae in all isolates. All were optochin resistant and bile insoluble, making S. pneumoniae unlikely. Vitek MS reported all identifications as “S. mitis/S. oralis.” In contrast, the Biotyper yielded reported S. mitis and S. oralis separately. Only 4 cases would have met CLABSI criteria as 3 isolates were erroneously reported as S. oralis. The Vitek 2 method would have yielded 5 CLABSIs; it failed to identify isolate 17B, and there were 2 errors. Moreover, 5 cases would have met CLABSI criteria with BD Phoenix, but there were 10 errors, including 2 reports of S. pneumoniae. It did not identify isolates 16A and16B. MicroScan also erroneously reported S. pneumoniae in 5 isolates. Confidence scores in 6 isolates were <90%. Furthermore, 2 cases of CLABSI attributed to commensals would have been reported with MicroScan. Also, 5 additional cases of CLABSI attributed to S. pneumoniae might have been reported. Finally, 1 case yielded 3 positive blood cultures; all 3 isolates were identified as Kocuria kristinae by all 5 methods.
TABLE 1 Bacterial Identification of Central-Line–Associated Blood Stream Infections Involving Streptococcal-Like Organisms by Testing Method
NOTE. MS, mass spectroscopy; NG, no growth; BD, Becton-Dickinson; CLABSI, central-line–associated bloodstream infection.
DISCUSSION
To our knowledge, this is the only study in the literature to investigate the relationship between identification method and CLABSI designation.
In the 13 CLABSIs involving CoNS, identifications did not agree across the 5 identification methods but all within-case identifications were concordant. Published data suggest that the accuracy of MALDI-TOF MS for CoNS is superior to automated biochemical methods and that errors with the latter are particularly prevalent in species other than S. epidermidis.Reference Loonen, Jansz, Stalpers, Wolffs and van den Brule 6 , Reference Dupont, Sivadon-Tardy and Bille 7 A larger study that includes more species other than S. epidermidis may yield more errors and discrepancies between initial and subsequent identifications when biochemical methods are used.
All 8 CLABSI cases involving VGS yielded tuf sequences suggesting an identification of S. mitis and/or S. pseudopneumoniae. While MALDI TOF-MS yielded S. mitis and/or S. oralis identifications consistently, the biochemical methods repeatedly produced identifications that were incorrect or were associated with low confidence scores. Vitek 2 performed best, followed by BD Phoenix, then MicroScan. Problems with biochemical identification of VGS have been well documented in the medical literature, but published literature have also suggested that Vitek MS and the Biotyper can misidentify VGS.Reference Alatoom, Cunningham, Ihde, Mandrekar and Patel 8 – Reference Branda, Markham, Garner, Rychert and Ferraro 10
Identification systems failed to identify isolates in cases #16 and #17. In these situations, CLABSI status would have depended on the exact ID reported by the laboratory. If “viridans group Streptococcus” was reported, both may have qualified as CLABSI. 2 However, if a more conservative approach had been taken (eg, by reporting “Gram-positive cocci” only) the cases might not have met CLABSI criteria.
This study has several limitations. First, the small sample size was due to the relative infrequency of CLABSI events at a single institution. Second, the FDA-cleared Bruker CA system is now available, but its performance has not been examined. Third, our sequencing procedures were unable to differentiate S. mitis from S. pseudopneumoniae. A more complex, multiplex PCR approach is necessary to definitively differentiate the 2 species.
In conclusion, this study showed that the method of organism identification used at an institution may impact upon whether a patient with a BSI meets criteria for a reportable CLABSI. MALDI-TOF MS platforms appear to yield consistently accurate results, which may translate to more cases of reported CLABSI.
Acknowledgments
We thank the technologists in the clinical microbiology laboratory at The Children’s Hospital of Philadelphia for their contributions. We also thank Ms Karen Bendig from Temple University Hospital for assisting with this manuscript.
Financial support: No financial support was provided relevant to this article.
Potential conflicts of interest: All authors report no conflicts of interest relevant to this article.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit https://doi.org/10.1017/ice.2017.112
