Rapid identification of bacteria from positive blood culture bottles by MALDI-TOF MS following short-term incubation on solid media

Transcription

1 Journal of Medical Microbiology (2015), 64, DOI /jmm Rapid identification of bacteria from positive blood culture bottles by MALDI-TOF MS following short-term incubation on solid media Osman Altun, Silvia Botero-Kleiven, Sarah Carlsson, Måns Ullberg and Volkan Özenci Correspondence Volkan Özenci Received 15 May 2015 Accepted 8 September 2015 Division of Clinical Microbiology F72, Karolinska Institute, Karolinska University Hospital, Huddinge, SE Stockholm, Sweden Rapid identification of bacteria from blood cultures enables early initiation of appropriate antibiotic treatment in patients with bloodstream infections (BSI). The objective of the present study was to evaluate the use of matrix-associated laser desorption ionization-time of flight (MALDI-TOF) MS after a short incubation on solid media for rapid identification of bacteria from positive blood culture bottles. MALDI-TOF MS was performed after 2.5 and 5.5 h plate incubation of samples from positive blood cultures. Identification scores with values $1.7 were accepted as successful identification if the results were confirmed by conventional methods. Conventional methods included MALDI-TOF MS, Vitek 2, and diverse biochemical and agglutination tests after overnight culture. In total, 515 positive blood cultures with monomicrobial bacterial growth representing one blood culture per patient were included in the study. There were 229/515 (44.5 %) and 286/515 (55.5 %) blood culture bottles with Gramnegative bacteria (GNB) and Gram-positive bacteria (GPB), respectively. MALDI-TOF MS following short-term culture could accurately identify 300/515 (58.3 %) isolates at 2.5 h, GNB being identified in greater proportion (180/229; 78.6 %) than GPB (120/286; 42.0 %). In an additional 124/515 bottles (24.1 %), identification was successful at 5.5 h, leading to accurate identification of bacteria from 424/515 (82.3 %) blood cultures after short-term culture. Interestingly, 11/24 of the isolated anaerobic bacteria could be identified after 5.5 h. The present study demonstrates, in a large number of clinical samples, that MALDI-TOF MS following short-term culture on solid medium is a reliable and rapid method for identification of bacteria from blood culture bottles with monomicrobial bacterial growth. INTRODUCTION Mortality rates due to bloodstream infections (BSI) remain unacceptably high. The case-fatality rates of nosocomial BSI range from 12 to 32 % (Goto & Al-Hasan, 2013). Early initiation of appropriate antibiotic therapy significantly reduces mortality, morbidity and hospital costs (Beekmann et al., 2003; Judd et al., 2014). Rapid and accurate identification of micro-organisms from BSI is an obvious target to enable early appropriate antimicrobial therapy. In addition, rapid species identification may lead to decreased use of broad-spectrum antibiotics, the widespread use of which is associated with greater adverse patient effects and an increased risk of selection of resistant bacteria (Kang et al., 2005; Kollef, 2008). Abbreviations: AST, antibiotic susceptibility testing; BSI, bloodstream infections; GNB, Gram-negative bacteria; GPB, Gram-positive bacteria; MALDI-TOF, matrix-associated laser desorption ionization-time of flight. This understanding has led to the development of rapid methods for identification of micro-organisms causing BSI. Several rapid molecular-based species identification methods such as broad-range PCR followed by sequencing, microarray, fluorescent in situ hybridization (FISH) and pathogen-specific real-time PCR have been developed (Peters et al., 2006; Dark et al., 2015). Many of these methods allow a sensitive and specific microbial species identification significantly faster than conventional bacteriological agar-based culture and phenotypic identification methods. However, these methods are still labour intensive and expensive, rely on special equipment, and in most instances also exhibit an unsatisfactorily narrow diagnostic spectrum. Methods focused on identification of single pathogens such as Staphylococcus aureus (Chapin & Musgnug, 2003) or Streptococcus pneumoniae (Petti et al., 2005) have also been described. These methods are relatively easy to perform and inexpensive. However, as they detect only a single pathogen in each reaction, they possess a limited usefulness G 2015 The Authors Printed in Great Britain

2 Rapid culture and MALDI-TOF MS Matrix-associated laser desorption ionization-time of flight (MALDI-TOF) MS analysis of bacterial colonies grown on solid medium allows fast and reliable identification of micro-organisms within minutes. Recently, several groups have reported promising results for MALDI-TOF MS in identification of micro-organisms directly from positive blood culture bottles (Wüppenhorst et al., 2012; Fothergill et al., 2013). However, these protocols are time consuming, require skilled laboratory personnel and are relatively difficult to implement in the workflow of routine clinical practice. Previously, we presented a novel method for rapid and reliable identification of micro-organisms from blood culture bottles based on simple routine diagnostic tests following a decreased incubation time on solid medium (Özenci et al., 2008). Subsequently, the method was integrated in our routine clinical practice. The introduction of MALDI-TOF MS in our laboratory spurred us to investigate the identification capabilities of this method. The aim of this study was to evaluate rapid bacterial species identification from positive blood culture bottles subsequent to short-term subculture on solid medium. METHODS Blood culture bottles and system. The blood cultures were collected at the clinical wards and then transferred to the laboratory according to Karolinska University Hospital guidelines. The standard blood volume for inoculation in BacT/Alert FA, FA Plus, FN and FN Plus blood culture bottles (biomérieux) was 8 10 ml. BacT/Alert PF bottles were inoculated with ml blood as recommended. The bottles were incubated in the BacT/Alert 3D automated blood culture system (biomérieux) until positivity or for a maximum incubation time of 5 days. The performance of the method was analysed for positive blood cultures with monomicrobial bacterial growth. Bottles yielding polymicrobial or yeast growth were excluded. Only one positive bottle per patient was included in the study. Rapid culture method. Broth from bottles that signalled positive was first evaluated by Gram stain. The subculture process was done by transferring several drops of broth with an inoculation needle onto dual blood/cystine lactose electrolyte-deficient (CLED) agar plates (Alpha Biosciences and Lab M, respectively) (3 drops per side) and chocolate agar plates (BD) (5 drops); the volume per drop was approximately 30 ml. The sample was then streaked on the agar surface in two directions with a 10 ml bacteriological loop. Plates were incubated at 35 uc (blood/cled agar plates in air and chocolate agar in 5 % CO 2 atmosphere) for 2.5 and 5.5 h before analysis by MALDI- TOF MS. These two time points were chosen according to our previous experience with this rapid culture method, which has been in routine use in our laboratory since 2007 (Özenci et al., 2008). If a satisfactory MALDI-TOF identification score (i1.7) was obtained after the 2.5 h incubation, the sample was not tested again after 5.5 h. All isolates were also identified by conventional methods after overnight incubation of plates. When anaerobic bacteria were suspected from Gram-stain morphology, similar amounts of broth were subcultured in addition on blood agar plates and incubated at 35 ucinan anaerobic environment for the same time periods. Conventional identification methods. All positive bottles were subjected to standard culture and identification methods. Initially, Gram stain was performed. The results of the Gram stains directed further subculturing of broth onto relevant agar plates including blood, CLED, chocolate and anaerobic blood agar. Plates were incubated overnight in the appropriate environment and individual bacterial colonies were used for species identification by MALDI-TOF MS (Bruker Daltonik) or the Vitek 2 XL system (biomérieux). A high confidence score (w2.0) for direct colony testing by MALDI-TOF MS and w93 % probability results by Vitek 2 GN, GP or ANC ID cards were considered identification at the species level as recommended by the manufacturers. When the criteria for identification at the species level by MALDI-TOF MS or Vitek 2 were not met the following tests were used: Gram staining, catalase, oxidase, DNase, indole spot and L-pyrrolidonyl-b-naphthylamide (Remel), agglutination tests for group A, B, C, D, G streptococci and Streptococcus pneumoniae (both from Oxoid), and Salmonella spp. (Reagensia). In addition, optochin discs on gentian violet blood agar were used for the differentiation of a-streptococci and pneumococci. MALDI-TOF MS. MALDI-TOF MS was performed directly from bacterial material on plates as described by Croxatto et al. (2012). Briefly, a very thin layer from visible bacterial growth after 2.5 and 5.5 h incubation was scraped with a 1 ml bacterial loop and smeared on 1 spot of a 96-spot stainless steel target plate (Bruker Daltonik). On-plate extraction was performed by overlaying each spot with 1 ml 70 % formic acid solution (Fluka) and then allowing the samples to air dry for 2 min. One microlitre of CHCA matrix (a-cyano-4- hydroxycinnamic acid) in 50 % acetonitrile and 2.5 % trifluoroacetic acid (Bruker Daltonik) was then applied on each sample and air-dried before analysis with MALDI-TOF MS. The Bruker Biotyper 3.1 software and library (version 4613; Bruker Daltonik) was used for spectra analysis. Following the manufacturer s instructions, scores of i1.7 were considered as low-confidence identification but interpreted as a successful identification if confirmed by conventional methods obtained from overnight cultures. Scores of v1.7 were considered as failed identification. RESULTS Identification by conventional culture methods In total, 579 positive blood cultures were studied. There were 12/579 (2.1 %) blood cultures with polymicrobial and 52/579 (9.0 %) bottles with yeast growth. Therefore, 515 positive clinical blood cultures were included during the study period, comprising 286/515 (55.5 %) bottles with Gram-positive bacteria (GPB) and 229/515 (44.5 %) with Gram-negative bacteria (GNB). Conventional methods could identify 267/286 (93.4 %) GPB and 223/ 229 (97.4 %) GNB to the species level after overnight culture (Tables 1 and 2). Identification by MALDI-TOF after rapid culture After 2.5 h. Accurate bacterial identification after a 2.5 h incubation was obtained in 300/515 (58.3 %) samples. At this time point, 120/286 (42.0 %) GPB and 180/229 (78.6 %) GNB could be correctly identified with the presented method. Interestingly, the majority of Enterobacteriaceae and Staphylococcus aureus were identified in 2.5 h, 173/209 (82.8 %) and 44/61 (72.1 %), respectively. In addition, 16/36 (44.4 %) Enterococcus spp., 3/6 (50 %) aerobic non-enterobacteriaceae GNB and 4/14 (28.6 %) anaerobic GNB could be identified (Tables 1 and 2)

3 O. Altun and others Table 1. Identification of GPB by rapid culture followed by MALDI-TOF MS and by conventional methods CoNS, Coagulase-negative staphylococci. Organism Rapid culture followed by MALDI-TOF Conventional methods 2.5 h 5.5 h No ID.24 h Staphylococcus spp. 96 (49.7 %) 61 (31.6 %) 36 (18.7 %) 193 (100 %) Staphylococcus aureus Staphylococcus capitis Staphylococcus epidermidis Staphylococcus haemolyticus Staphylococcus hominis Staphylococcus pettenkoferi Staphylococcus saprophyticus Other CoNS* Enterococcus spp. 16 (44.4 %) 17 (47.3 %) 3 (8.3 %) 36 (100 %) Enterococcus faecalis Enterococcus faecium Enterococcus casseliflavus Streptococcus spp. 5 (13.5 %) 8 (21.6 %) 24 (64.9 %) 37 (100 %) Streptococcus agalactiae Streptococcus dysgalactiae Streptococcus pyogenes Streptococcus anginosus Streptococcus gallolyticus Streptococcus parasanguinis Streptococcus pneumoniae Streptococcus salivarius Other a-streptococcid Other GPB Gemella sanguinis Rothia mucilaginosa Micrococcus spp Listeria monocytogenes Bacillus thuringiensis Corynebacterium spp Anaerobic GPB Clostridium halophilum Clostridium perfringens Clostridium spp Propionibacterium acnes Propionibacterium avidum Eggerthella lentad Gram-positive anaerobic rod Total 120 (42 %) 90 (31.5 %) 76 (26.5 %) 286 (100 %) *Could not be identified by MALDI-TOF MS or Vitek 2; both isolates were DNase negative. D These isolates could not be identified by MALDI-TOF MS; they were classified as a-streptococci based on optochin resistance and negative agglutination for group A, B, C, D, G streptococci and Streptococcus pneumoniae. Vitek 2 was not performed in these cases. d Identified by Vitek 2, with a high identification score of 99 %. After 5.5 h. Identificationofanadditional90GPBand 34 GNB could be achieved by the presented method after 5.5 h growth on agar plates. In total, 424/515 (82.3 %) of all isolates included in the study were identified at this time point, the cumulative rates of identification after 5.5 h being 73.4 % for GPB and 93.4 % for GNB. The majority of Staphylococcus aureus (56/61, 91.8 %), enterococci (33/36, 91.7 %) and b-haemolytic streptococci group A, B and G (8/10, 80 %) were correctly identified at the end of the short-term culture period. Identification of non-staphylococcus aureus staphylococci was less successful as only 101/132 (76.5 %) of the isolates could be characterized to the species level. Major limitations were seen in 1348 Journal of Medical Microbiology 64

4 Rapid culture and MALDI-TOF MS Table 2. Identification of GNB by rapid culture followed by MALDI-TOF MS and by conventional methods Organism Rapid culture followed by MALDI-TOF Conventional methods 2.5 h 5.5 h No ID.24 h Enterobacteriaceae 173 (82.8 %) 28 (13.4 %) 8 (3.8 %) 209 (100 %) Escherichia coli Enterobacter asburiae Enterobacter aerogenes Enterobacter cloacae Enterobacter kobei Klebsiella pneumoniae Klebsiella oxytoca Raoultella planticola Citrobacter freundii Citrobacter koseri Morganella morganii Proteus mirabilis Salmonella spp.* Serratia marcescens Other GNB Aeromonas hydrophila Acinetobacter ursingii Pseudomonas aeruginosa Brucella spp.d Haemophilus influenzae Anaerobic GNB Bacteroides fragilis Bacteroides vulgatus Bacteroides thetaiotaomicron Parabacteroides distasonis Prevotella oralis Veillonella parvula Fusobacterium necrophorum Gram-negative anaerobic rodd Total 180 (78.6 %) 34 (14.8 %) 15 (6.6 %) 229 (100 %) *As all Salmonella spp. were identified only at the genus level by conventional methods at $24 h, MALDI-TOF MS identification at the genus level was accepted as a successful identification. D For biosafety reasons, the isolate was sent for typing by PCR to a reference laboratory. d Could not be identified by MALDI-TOF MS or Vitek 2 because of insufficient bacterial growth even after.24 h. the identification of Streptococcus pneumoniae (2/8), other a-haemolytic streptococci (3/19) and anaerobic GPB (2/10). When aerobic GNB bacteria were considered, the additional culture time further raised the identification rate for Enterobacteriaceae from 173/209 (82.8 %) to 201/209 (96.2 %) correctly identified species, and for non-enterobacteriaceae to 4/6 isolates (66.8 %). Interestingly, 9/14 anaerobic GNB could be identified successfully after 5.5 h. In particular, the method performed unexpectedly well for Bacteroides spp., identifying 8/9 of the isolates. Isolates that could not be identified by this approach. Most of the remaining bacteria were represented by too few isolates to evaluate the performance of the method. Gram-positive rods such as Corynebacterium spp. or Propionibacterium spp. and anaerobes other than Bacteroides spp. were difficult to identify after rapid culture (Tables 1 and 2). Similarly, 24/37 (64.9 %) Streptococcus spp. could not be identified correctly. In general, these microorganisms either showed no growth after rapid culture or did not yield a reliable identification result, i.e. scores v1.7 with MALDI-TOF MS. False species identification was only observed in the case of 4 a-haemolytic streptococci that were incorrectly typed as Streptococcus pneumoniae (data not shown). These isolates were identified (to the species or genus level) by standard microbiological identification tests, including MALDI-TOF MS from overnight cultures. DISCUSSION Hitherto published studies have shown that early effective antimicrobial treatment of patients with BSI is crucial in

5 O. Altun and others reducing mortality rates and hospital-related costs due to these infections (Beekmann et al., 2003; Kumar et al., 2006). It is, therefore, essential to develop rapid diagnostic methods to identify micro-organisms that cause BSI. The present study evaluated MALDI-TOF MS following a short-term culture for identification of bacteria from positive blood culture bottles. A large number of clinical monomicrobial blood cultures were analysed prospectively. The proposed method could accurately identify 424/515 (82.3 %) bacteria from monomicrobial cultures after a 5.5 h subculture from positive blood culture bottles. For 300/515 (58.3 %) bacteria, identification was already obtained after 2.5 h. The overall bacterial identification rates obtained at 5.5 h are comparable to the results of direct MALDI-TOF MS identification from blood culture bottles (Vlek et al., 2012; Wüppenhorst et al., 2012; Fothergill et al., 2013). However, direct MALDI-TOF MS methods may be difficult to implement and integrate alongside conventional diagnostic methods because of an added hands-on time aspect. Benefits of the method evaluated here were thus: no need of extra reagents or kits, and no additional steps outside routine procedures, such as sample lysis, washing or centrifuging. In addition to the identification of micro-organisms, antibiotic susceptibility testing (AST) provides essential information on resistance and allows the clinician to tailor appropriate antimicrobial therapy accordingly. A clear advantage of the rapid culture method is that it allows initiation of AST simultaneously with identification from agar plates as reported recently by Idelevich et al. (2014a). In comparison, other rapid molecular identification methods provide no or only limited AST possibilities (Kothari et al., 2014). Direct MALDI-TOF MS methods may not always be suitable for AST, as some protocols expose the micro-organisms to harsh chemicals including formic acid to extract proteins (Croxatto et al., 2012). Another benefit of using rapid culture with MALDI-TOF MS identification is the use of an extensive MALDI-TOF MS database that covers a wide range of bacterial species. During the study period, 51 different bacterial species were detected by conventional methods and 41/51 (80.4 %) of these isolates could be identified by the method described here. In contrast to molecular diagnostic tests such as QuickFISH and Verigene Gram-positive/ Gram-negative blood culture (Hill et al., 2014; Martinez et al., 2014) that are capable of detecting only a limited range of bacterial species, the present method could identify the majority of the bacterial species isolated from the clinical samples. The bacteria that could not be identified after rapid culture in this study included coagulasenegative staphylococci and a-streptococci, Micrococcus spp., Corynebacterium spp., Propionibacterium spp., Brucella spp., Haemophilus influenzae and Eggerthella lenta. This limitation might apply as well to other slow-growing micro-organisms, such as yeasts. This was reported in a study by Verroken et al. (2015), who observed poor identification results with rapid yeast culture. Recently, Bhatti et al. (2014) presented a rapid culture and MALDI-TOF method using pre-warmed blood agar plates and different plating techniques with and without previous centrifugation of the blood culture broth. Significant identification rates of bacterial isolates were obtained with the pellet streak method, 68 % after 2 h and 96 % after 6 h. However, double centrifugation of the samples was required, demanding longer hands-on time and making the procedure overall more time-consuming. The identification rates after direct subculture of blood broth samples (i.e. direct streak or spot) reached 49 % at 2 h and 92 % at 6 h, which is similar to our present findings. Thus, the comparison implies that the use of pre-warmed plates does not provide a clear advantage for early detection of micro-organisms, but requires instead an additional step in the procedure. The study by Bhatti et al. (2014) is moreover limited by the comparatively small number of bottles tested (n 5 134), reducing the sample sizes for specific and potentially difficult bacteria. In the present study, the higher number of samples ensured a better representation of the diversity of the micro-organisms detected in routine clinical practice. Another previous study on rapid culture on blood agar plates followed by MALDI-TOF identification evaluated multiple rapid culture time points (including 2 and 6 h), but also two different levels of identification score (i1.7 or i2.0) in 152 positive blood cultures. While that study showed lower identification rates for Gram-positive cocci at 2 h, identification rates for both Gram-positive cocci and Gram-negative rods at 6 h with a score of i1.7 were very similar to ours in the present study (Idelevich et al., 2014b). In a larger study with 453 monomicrobial episodes, Verroken et al. (2015) showed that MALDI-TOF MS could identify 369/453 (81.5 %) of isolates after 5 h incubation on solid medium. In contrast to the present study, the authors analysed only one incubation time point. The information that a considerable number of samples can already be identified after 2.5 h appears, however, highly relevant for the application of this method in the clinical laboratory. Obtaining identification results at such an early time point might eventually help the clinicians to adjust antibiotic treatment in a more timely way. The clinical importance of early effective treatment in patients with severe sepsis was shown previously (Kumar et al., 2006). In contrast to the study by Verroken et al. (2015), where only 2/10 anaerobic bacteria were detected after 5 h, we were able to identify 11/24 of the isolated anaerobic bacteria after a 5.5 h incubation; among those were 8/9 Bacteroides spp., 1/1 Parabacteroides spp. and 2/3 Clostridium spp. The underlying reason for the discrepant finding for anaerobic bacteria is not clear and might relate to the 1350 Journal of Medical Microbiology 64

6 Rapid culture and MALDI-TOF MS small number of isolates analysed. Further studies with larger numbers of clinical samples containing anaerobic bacteria are warranted. The shortcoming of MALDI-TOF MS in accurately differentiating Streptococcus pneumoniae from other a- haemolytic streptococci is a well-known phenomenon (Neville et al., 2011; Schulthess et al., 2013). The present method could identify only 2/8 Streptococcus pneumoniae isolates correctly, probably based on limitations in the resolution of the MALDI-TOF MS analysis. In addition, detection of other a-haemolytic streptococci with the present method was poor, probably due to slow growth of these bacteria. Therefore, usage of the present method for identification of Streptococcus pneumoniae and other a- streptococci is limited. The new algorithms developed for MALDI-TOF MS that accurately identify Streptococcus pneumoniae might lead to improved identification of this bacterium by the present method. Another limitation of the present study was the selected MALDI-TOF MS score of i1.7, following the manufacturer s recommendations. However, the identification score results with the present method were compared to conventional methods and accepted as successful identification only after concordance with overnight culture and MALDI-TOF MS scores of i2.0 or w93 % probability results by the Vitek 2 system. It is therefore plausible to suggest that the present results reflect the clinical performance of the method. In conclusion, the present study demonstrates, in a large number of clinical samples, that MALDI-TOF MS following short-term culture on solid medium is a reliable, easy and rapid method for the identification of the majority of important bacteria from blood culture bottles with monomicrobial bacterial growth. ACKNOWLEDGEMENTS The authors would like to thank Monika Kedfors for her excellent technical assistance and Petra Lüthje for critical reading of the manuscript. REFERENCES Beekmann, S. E., Diekema, D. J., Chapin, K. C. & Doern, G. V. (2003). Effects of rapid detection of bloodstream infections on length of hospitalization and hospital charges. J Clin Microbiol 41, Bhatti, M. M., Boonlayangoor, S., Beavis, K. G. & Tesic, V. (2014). Rapid identification of positive blood cultures by matrix-assisted laser desorption ionization-time of flight mass spectrometry using prewarmed agar plates. J Clin Microbiol 52, Chapin, K. & Musgnug, M. (2003). Evaluation of three rapid methods for the direct identification of Staphylococcus aureus from positive blood cultures. J Clin Microbiol 41, Croxatto, A., Prod hom, G. & Greub, G. (2012). Applications of MALDI-TOF mass spectrometry in clinical diagnostic microbiology. FEMS Microbiol Rev 36, Dark, P., Blackwood, B., Gates, S., McAuley, D., Perkins, G. D., McMullan, R., Wilson, C., Graham, D., Timms, K. & Warhurst, G. (2015). Accuracy of LightCycler( H ) SeptiFast for the detection and identification of pathogens in the blood of patients with suspected sepsis: a systematic review and meta-analysis. Intensive Care Med 41, Fothergill, A., Kasinathan, V., Hyman, J., Walsh, J., Drake, T. & Wang, Y. F. (2013). Rapid identification of bacteria and yeasts from positiveblood-culture bottles by using a lysis-filtration method and matrixassisted laser desorption ionization-time of flight mass spectrum analysis with the SARAMIS database. J Clin Microbiol 51, Goto, M. & Al-Hasan, M. N. (2013). Overall burden of bloodstream infection and nosocomial bloodstream infection in North America and Europe. Clin Microbiol Infect 19, Hill, J. T., Tran, K.-D., Barton, K. L., Labreche, M. J. & Sharp, S. E. (2014). Evaluation of the nanosphere Verigene BC-GN assay for direct identification of Gram-negative bacilli and antibiotic resistance markers from positive blood cultures and potential impact for morerapid antibiotic interventions. JClinMicrobiol52, Idelevich, E. A., Schüle,I.,Grünastel, B., Wüllenweber, J., Peters, G. & Becker, K. (2014a). Acceleration of antimicrobial susceptibility testing of positive blood cultures by inoculation of Vitek 2 cards with briefly incubated solid medium cultures. J Clin Microbiol 52, Idelevich, E. A., Schüle,I.,Grünastel, B., Wüllenweber, J., Peters, G. & Becker, K. (2014b). Rapid identification of microorganisms from positive blood cultures by MALDI-TOF mass spectrometry subsequent to very short-term incubation on solid medium. Clin Microbiol Infect 20, Judd, W. R., Stephens, D. M. & Kennedy, C. A. (2014). Clinical and economic impact of a quality improvement initiative to enhance early recognition and treatment of sepsis. Ann Pharmacother 48, Kang, C.-I., Kim, S.-H., Park, W. -B., Lee, K.-D., Kim, H.-B., Kim, E.-C., Oh, M. D. & Choe, K.-W. (2005). Bloodstream infections caused by antibiotic-resistant Gram-negative bacilli: risk factors for mortality and impact of inappropriate initial antimicrobial therapy on outcome. Antimicrob Agents Chemother 49, Kollef, M. H. (2008). Broad-spectrum antimicrobials and the treatment of serious bacterial infections: getting it right up front. Clin Infect Dis 47 (Suppl. 1), S3 S13. Kothari, A., Morgan, M. & Haake, D. A. (2014). Emerging technologies for rapid identification of bloodstream pathogens. Clin Infect Dis 59, Kumar, A., Roberts, D., Wood, K. E., Light, B., Parrillo, J. E., Sharma, S., Suppes, R., Feinstein, D., Zanotti, S. & other authors (2006). Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock. Crit Care Med 34, Martinez, R. M., Bauerle, E. R., Fang, F. C. & Butler-Wu, S. M. (2014). Evaluation of three rapid diagnostic methods for direct identification of microorganisms in positive blood cultures. J Clin Microbiol 52, Neville, S. A., Lecordier, A., Ziochos, H., Chater, M. J., Gosbell, I. B., Maley, M. W. & van Hal, S. J. (2011). Utility of matrix-assisted laser desorption ionization-time of flight mass spectrometry following introduction for routine laboratory bacterial identification. J Clin Microbiol 49, Özenci, V., Tegmark-Wisell, K., Lundberg, C. & Wretlind, B. (2008). Rapid culture and identification: a practical method for early preliminary laboratory diagnosis of sepsis. Clin Microbiol Infect 14, Peters, R. P., van Agtmael, M. A., Simoons-Smit, A. M., Danner, S. A., Vandenbroucke-Grauls, C. M. & Savelkoul, P. H. (2006). Rapid

7 O. Altun and others identification of pathogens in blood cultures with a modified fluorescence in situ hybridization assay. JClinMicrobiol44, Petti, C. A., Woods, C. W. & Reller, L. B. (2005). Streptococcus pneumoniae antigen test using positive blood culture bottles as an alternative method to diagnose pneumococcal bacteremia. J Clin Microbiol 43, Schulthess, B., Brodner, K., Bloemberg, G. V., Zbinden, R., Böttger, E. C. & Hombach, M. (2013). Identification of Gram-positive cocci by use of matrix-assisted laser desorption ionization-time of flight mass spectrometry: comparison of different preparation methods and implementation of a practical algorithm for routine diagnostics. JClinMicrobiol51, Verroken, A., Defourny, L., Lechgar, L., Magnette, A., Delmée, M. & Glupczynski, Y. (2015). Reducing time to identification of positive blood cultures with MALDI-TOF MS analysis after a 5-h subculture. Eur J Clin Microbiol Infect Dis 34, Vlek, A. L. M., Bonten, M. J. M. & Boel, C. H. E. (2012). Direct matrixassisted laser desorption ionization time-of-flight mass spectrometry improves appropriateness of antibiotic treatment of bacteremia. PLoS One 7, e Wüppenhorst, N., Consoir, C., Lörch, D. & Schneider, C. (2012). Direct identification of bacteria from charcoal-containing blood culture bottles using matrix-assisted laser desorption/ionisation time-offlight mass spectrometry. Eur J Clin Microbiol Infect Dis 31, Journal of Medical Microbiology 64