Species-driven interpretation guidelines in case of a single-sampling strategy for blood culture

Eur J Clin Microbiol Infect Dis (2011) 30:1537 1541 DOI 10.1007/s10096-011-1257-3 ARTICLE Species-driven interpretation guidelines in case of a single-sampling strategy for blood culture D. Leyssene & S. Gardes & P. Vilquin & J.-P. Flandrois & G. Carret & B. Lamy Received: 17 December 2010 / Accepted: 28 March 2011 / Published online: 18 April 2011 # Springer-Verlag 2011 Abstract The purpose of this paper is to define guidelines to interpret positive blood cultures (BCs) to distinguish bloodstream infection (BSI) from contamination in BCs drawn with a single venipuncture. During a 2-year period, each positive BC set (comprising six bottles from a single venipuncture) was prospectively categorised by clinicians, bacteriologists and hospital epidemiologists as BSI or contamination. For each case, the number of positive bottles per set, results from Gram staining and microorganism identification were analysed in order to define interpretation guidelines. We analysed 940 positive BC sets. The BSI rate in monomicrobial BC sets was positively correlated with the number of positive bottles. The positive predictive value was 88% with one and 100% with 2 positive bottles for Escherichia coli; 100% forstaphylococcus aureus, Pseudomonas and Candida spp., regardless of the number of positive bottles; 3.5%, 61.1%, 78.9% and 100% for coagulase-negative staphylococci (CoNS) with one, two, three and 4 positive bottles, respectively. Using a single-sampling strategy, interpretation guidelines for monomicrobial positive BCs are based on the number of positive bottles per set, results from Gram staining and microorganism identification: 4 positive bottles ( 2 with Gram-negative bacilli) always led to a diagnosis of BSI. The CoNS BSI rate positively correlates with the number of positive bottles. D. Leyssene : J.-P. Flandrois : G. Carret Université de Lyon, 69000 Lyon, France D. Leyssene : J.-P. Flandrois : G. Carret CNRS, UMR5558, Laboratoire de Biométrie et Biologie Evolutive, Faculté de Médecine Lyon Sud Charles Mérieux, Université Lyon 1, 69921 Oullins Cedex, France S. Gardes : P. Vilquin Unité d Hygiène et d Epidémiologie, Centre Hospitalier Lyon Sud, 69495 Pierre-Bénite, France B. Lamy Laboratoire de Bactériologie, Centre Hospitalier Régional Universitaire, 34000 Montpellier, France D. Leyssene (*) : J.-P. Flandrois : G. Carret Laboratoire de Bactériologie, Centre Hospitalier Lyon Sud, 69495 Pierre-Bénite, France e-mail: davidleyssene@yahoo.fr Introduction Bloodstream infection (BSI) diagnosis relies on blood cultures (BCs) drawn within a 24-h period. There are two main strategies for obtaining blood, the most popular of which consists of two to three separate BC draws (hereafter multi-sampling strategy ) [1 4]. The alternative singlesampling strategy, which consists of drawing four to six bottles of blood from a single venipuncture [5, 6], has been developed on the basis of at least two studies that clearly show that the timing of blood specimen collection is unimportant and that there is no difference in the ability to detect BSI when BCs are taken simultaneously or at different times [6 8]. The rationale for this approach is based on the fact that: (i) BSIs are rarely truly intermittent during the short period of drawings (several hours to 24 h) and are, instead, rather continuous, with a very low density of microorganisms [7, 9 11] and (ii) the collection of a large blood volume with a single venipuncture allows sensitive BSI detection with a lower contamination rate and

1538 Eur J Clin Microbiol Infect Dis (2011) 30:1537 1541 allows an earlier antibiotic treatment. This strategy has been approved in France by the French Society for Microbiology as an alternative to the multi-sampling strategy [12, 13] and has been recommended in our hospital since 2004. The drawback of the single-sampling strategy is that the standard interpretation guidelines to differentiate BSIs from contaminations, which are based on the types of species recovered and the number of positive BC sets, cannot be used [14]. Hence, the aim of this study was to analyse BC results, collected using the single-sampling strategy over a 2-year period, in order to define interpretation guidelines applicable to the practice of the single-sampling strategy. Materials and methods Blood culture procedures A BC set was defined as a six-bottle set of blood (three aerobic, three anaerobic) drawn from a single venipuncture. To minimise the risk of contamination, a skin disinfection protocol using povidone iodine plus 70% ethyl alcohol [15] was systematically used and blood was preferentially obtained by direct venipuncture rather than from an intravascular catheter [16, 17]. All BCs were monitored by the BacT/Alert system (biomérieux, Marcy l Etoile, France) for 5 days. A set was declared positive when at least one bottle tested positive for microorganisms. Each positive bottle was subjected to Gram staining and subcultured on bacteriological media as appropriate. Patients and clinical interpretation The study was conducted in a 1,000-bed teaching hospital accounting for approximately 45,000 admissions and in three long-term-care hospitals with a total of 760 beds. All BC sets taken from adult patients between 2007 and 2008 were analysed. An episode was defined as all of the microorganisms isolated from the same patient within a 7-day period. Each episode was prospectively categorised by clinicians, hospital epidemiologists and bacteriologists as either BSI or contamination, taking into account clinical (e.g. fever, hypothermia, chills, blood pressure, heart rate, respiratory rate, organ failure) and biological (white blood cell count) signs of infection, results of others cultures, the decision of whether to treat and the clinical course. Data analysis To define interpretation guidelines, all BC results were analysed according to the number of positive bottles per set, results from Gram staining, microorganism identification and clinical categorisation as BSI or contamination. Table 1 Blood culture (BC) isolates from monomicrobial episodes Microorganism Categorisation Number of positive bottles Frequency Frequency (%) of BSI and C per microorganism 1 2 3 4 5 6 Frequency (%) Coagulase-negative staphylococci BSI 4 11 15 4 5 60 99 (44.8) 221 (27.5) C 111 7 4 0 0 0 122 (55.2) Escherichia coli BSI 22 22 20 15 15 62 156 (98.1) 159 (19.8) C 3 0 0 0 0 0 3 (1.9) Staphylococcus aureus BSI 15 10 11 7 7 48 98 (100) 98 (12.2) Other Enterobacteriaceae BSI 20 11 5 8 7 39 90 (100) 90 (11.2) Pseudomonas spp. BSI 11 9 26 0 0 0 46 (100) 46 (5.7) Anaerobic microorganisms BSI 17 4 10 1 0 1 33 (82.5) 40 (5.0) C 7 0 0 0 0 0 7 (17.5) Enterococcus spp. BSI 7 4 1 3 2 7 24 (96) 25 (3.1) C 1 0 0 0 0 0 1 (4) Corynebacterium spp. BSI 1 0 1 0 0 0 2 (11.1) 18 (2.2) C 13 1 2 0 0 0 16 (88.9) Other microorganisms (includingmicrococcus spp., Candida spp. and Haemophilus spp.) BSI 18 10 27 2 5 22 84 (77.8) 108 (13.4) C 22 2 0 0 0 0 24 (22.2) Total BSI 115 81 116 40 41 239 632 (78.5) 805 (100) C 157 10 6 0 0 0 173 (21.5) BSI: bloodstream infection; C: contamination

Eur J Clin Microbiol Infect Dis (2011) 30:1537 1541 1539 Results a 300 During the study period, 7,250 BC sets drawn from 4,424 adult patients were processed by the laboratory. Among these, 805 monomicrobial episodes were identified from 720 patients. Fifty-one different genera or species of microorganisms were recovered, including fastidious and slow-growing bacteria (e.g. Haemophilus influenzae, Capnocytophaga spp., Fusobacterium spp.). Aerobic/facultative anaerobic bacteria represented 92.9% of the isolated microorganisms (748 of 805), with coagulase-negative staphylococci (CoNS) (27.5%, 221 of 805), Escherichia coli (19.8%, 159 of 805) and Staphylococcus aureus (12.2%, 98 of 805) being the most prevalent bacteria (Table 1). Anaerobic bacteria represented 5% of the isolated microorganisms (40 of 805), with Bacteroides spp. being the most prevalent (2.9% - 23 of 805). All of the yeast found belonged to the genus Candida (2.1%, 17 of 805). When evaluated in relation to the number of positive bottles, the monomicrobial BC distribution was U-shaped (Fig. 1a) and included 632 (78.5%) BSIs and 173 (21.5%) contaminations. The BSI rate positively correlated with the number of positive bottles per set (Fig. 1a). The positive predictive value (PPV) of BSI was only 42.3% with one positive bottle, but rose to 89.0%, 95.1% and 100% when two, three and 4 bottles were positive, respectively. When the Gram staining results were included, the PPV was 23.4%, 80.4%, 90.5% and 100% with one, two, three and 4positive bottles, respectively, for Gram-positive cocci and 93.1% and 100% with one and 2 positive bottles, respectively, for Gram-negative bacilli. When including the microorganism identification results (Table 1), trends that distinguished contaminants from pathogens emerged in two areas: the number of positive bottles and the level of PPV of BSI. In the first area, frequent skin contaminants such as Corynebacterium spp. or Micrococcus spp. grew in a maximum of 3 out of 6 bottles, while pathogens such as E. coli or S. aureus grew more frequently in 6 out of 6 bottles. In the second area, the PPV of BSI for Corynebacterium spp. was low (7%, 0% and 33% with one, two and three positive bottles, respectively) (Table 1) and all Micrococcus spp. recovered were categorised as contaminants, whereas the PPV was 88% and 100% with one and 2 positive bottles, respectively, for E. coli and 100% with both one and 2 positive bottles for S. aureus. Similarly, all Pseudomonas spp. and Candida spp. isolated were classified as true pathogens regardless of the number of positive bottles. Finally, the PPV of CoNS, which are microorganisms of questionable significance, was 3.5%, 61.1%, 78.9% and 100% with one, two, three and 4 positive bottles, respectively (Fig. 1b). Therefore, the probability of CoNS being of clinical significance positively correlated with the number of positive bottles. Number of episodes b BSI and contamination rates (%) 250 200 150 100 50 0 100 80 60 40 20 0 During the study period, 135 polymicrobial episodes occurred: 70 were categorised as BSI, 13 as BSI and contamination and 14 as contamination (data not shown). Fifty episodes (5.3% of all positive episodes) could not be clearly categorised. Discussion 1 of 6 2 of 6 3 of 6 4 of 6 5 of 6 6 of 6 Number of positive bottles Contamination Bloodstream infection 1 of 6 2 of 6 3 of 6 4 of 6 5 of 6 6 of 6 Number of positive bottles Contamination Bloodstream infection Fig. 1 Distribution of bloodstream infection (BSI) and contamination in relation to the number of positive bottles for (a) all monomicrobial episodes and (b) coagulase-negative staphylococci (CoNS) episodes While not universally employed, the single-sampling strategy is being adopted by an increasing number of hospitals in France and also in Denmark (unpublished data), even though methodological and ethical issues make

1540 Eur J Clin Microbiol Infect Dis (2011) 30:1537 1541 it extremely difficult, if not impossible, to directly compare it to the multi-sampling strategy. The adoption of strategies that have not been, and cannot be, validated in large-scale, controlled clinical trials may be considered as questionable. Yet, several well-conducted studies have highlighted a number of potential advantages in using a single venipuncture for blood culture: there is no significant difference in the sensitivity of detection between blood cultures obtained simultaneously and those obtained at intervals within a 24-h drawing period [7]; the volume of blood cultured remains the most important variable in detecting bacteraemia in adult patients [1, 11, 18]; drawing all of the bottles at the same time allows one to check that the total blood volume cultured is adequate and avoid the solitary blood culture phenomenon (i.e. only two bottles, which would clearly be insufficient for the diagnosis of a BSI [18]), estimated at approximately 30% with the multi-sampling strategy [19]. The use of a routine single-sampling strategy for blood collection was found to be satisfactory and reproducibly led to the same epidemiology as the multi-sampling strategy [5]. Consistent with Arendrup et al., our results indicate that a single-sampling strategy allows the recovery of a wide spectrum of microorganisms: 51 different genera or species of microorganisms in monomicrobial episodes, including fastidious and slow-growing bacteria. We also found an epidemiology consistent with previous reports that used different strategies to obtain blood [5, 20 26], indicating that the multi- and single-sampling strategies reproducibly led to the same epidemiology. Moreover, only 21.5% (173 of 805) of monomicrobial positive episodes were contaminations, representing a lower rate relative to other studies, which have reported up to 50% false-positive results [27 30]. This is likely related to the single-sampling strategy [5, 6], taking into account that the global rate of contamination found in our study (between 2.8% and 3.4% per drawing) was similar to the overall rate reported elsewhere (3%), regardless of whether the 50 unclassified polymicrobial episodes were all categorised as BSIs or contaminations [2, 4, 11, 31]. The BSI PPV rate was clearly positively correlated with the number of positive bottles per set: 4 positive bottles always led to a diagnosis of BSI regardless of the microorganism identity ( 2 positive bottles with Gram-negative bacilli). When the microorganism identity was known, the interpretation was straightforward in most cases. Frequent skin contaminants grew in a maximum of 3 out of 6 bottles and had a low PPVof BSI, unlike other usual pathogens. The number of positive bottles also reliably predicted the clinical significance of CoNS [1] (Fig. 1b): only one positive bottle per set was associated with a very low risk of BSI (3.5%), while there was no contamination when 4 bottles were positive. Though infrequent, cases with two and three positive bottles remained the most difficult to interpret (18 and 19 episodes with two and three positive bottles out of 805 monomicrobial episodes, respectively). These cases should be carefully evaluated in collaboration with a clinician, examining the patient s clinical history and the presence of vascular-access devices. Therefore, using a six-bottle single-sampling strategy and strict skin disinfection, our results support those of Arendrup et al. [5], suggesting a trend between the clinically significant CoNS and the number of positive bottles per set. It should be noted that discordant results have been reported elsewhere [32, 33]. However, it was unclear whether these authors used a singleor a multi-sampling strategy to obtain blood [33] or whether the inclusion of samples taken from the permanent cuffed central venous catheters may have biased the results [32]. Additionally, compared to other studies [5, 32, 33], the PPV of CoNS result may have been optimised in our study because the results are based on a six-bottle protocol, simply because the discriminating power of an interpretation based on the number of positive bottles is greater when the total number of bottles is greater. Finally, polymicrobial BC sets are usually difficult to interpret [34]. In our study, 50 episodes remained unclassified by clinicians, hospital epidemiologists and bacteriologists. These episodes involved bacteria for which the data were too small to determine guidelines (e.g. α-hemolytic streptococci) and could also not be clearly categorised using our criteria. Given the small number of cases, further studies are required in order to validate interpretation criteria for polymicrobial BCs. In summary, we defined the following interpretation guidelines for monomicrobial positive cultures which should be useful for laboratories using a single-sampling strategy to diagnose BSIs: 4 positive bottles ( 2 with Gram-negative bacilli) always leads to a diagnosis of BSI, interpretation should be easy in most cases with common contaminants and pathogens, and the rate of CoNS bacteremia should positively correlate with the number of positive bottles. We found this strategy to be sensitive, specific and convenient, requiring only one venipuncture and offering the possibility of earlier antibiotic treatment. Acknowledgement This document has been checked for English language accuracy by American Journal Experts. References 1. Washington JA 2nd (1975) Blood cultures: principles and techniques. Mayo Clin Proc 50(2):91 98, Review 2. Weinstein MP (2003) Blood culture contamination: persisting problems and partial progress. J Clin Microbiol 41(6):2275 2278 3. Beekmann SE, Diekema DJ, Doern GV (2005) Determining the clinical significance of coagulase-negative staphylococci isolated

Eur J Clin Microbiol Infect Dis (2011) 30:1537 1541 1541 from blood cultures. Infect Control Hosp Epidemiol 26(6):559 566 4. Bekeris LG, Tworek JA, Walsh MK, Valenstein PN (2005) Trends in blood culture contamination: a College of American Pathologists Q-Tracks study of 356 institutions. Arch Pathol Lab Med 129(10):1222 1225 5. Arendrup M, Jensen IP, Justesen T (1996) Diagnosing bacteremia at a Danish hospital using one early large blood volume for culture. Scand J Infect Dis 28(6):609 614 6. Lamy B, Roy P, Carret G, Flandrois JP, Delignette-Muller ML (2002) What is the relevance of obtaining multiple blood samples for culture? A comprehensive model to optimize the strategy for diagnosing bacteraemia. Clin Infect Dis 35(7):842 850 7. Li J, Plorde JJ, Carlson LG (1994) Effects of volume and periodicity on blood cultures. J Clin Microbiol 32(11):2829 2831 8. Riedel S, Bourbeau P, Swartz B, Brecher S, Carroll KC, Stamper PD, Dunne WM, McCardle T, Walk N, Fiebelkorn K, Sewell D, Richter SS, Beekmann S, Doern GV (2008) Timing of specimen collection for blood cultures from febrile patients with bacteremia. J Clin Microbiol 46(4):1381 1385 9. Jonsson B, Nyberg A, Henning C (1993) Theoretical aspects of detection of bacteraemia as a function of the volume of blood cultured. APMIS 101(8):595 601 10. Wain J, Diep TS, Ho VA, Walsh AM, Nguyen TT, Parry CM, White NJ (1998) Quantitation of bacteria in blood of typhoid fever patients and relationship between counts and clinical features, transmissibility, and antibiotic resistance. J Clin Microbiol 36(6):1683 1687 11. Riedel S, Carroll KC (2010) Blood cultures: key elements for best practices and future directions. J Infect Chemother 16(5):301 316, Epub 2010 May 21 12. Groupe Rémic de la Société Française de Microbiologie (2007) Référentiel en microbiologie médicale (bactériologie et mycologie). Vivactis Plus, Paris 13. Groupe Rémic de la Société Française de Microbiologie (2010) Référentiel en microbiologie médicale. Société Française de Microbiologie, Paris 14. MacGregor RR, Beaty HN (1972) Evaluation of positive blood cultures. Guidelines for early differentiation of contaminated from valid positive cultures. Arch Intern Med 130(1):84 87 15. Calfee DP, Farr BM (2002) Comparison of four antiseptic preparations for skin in the prevention of contamination of percutaneously drawn blood cultures: a randomized trial. J Clin Microbiol 40(5):1660 1665 16. Everts RJ, Vinson EN, Adholla PO, Reller LB (2001) Contamination of catheter-drawn blood cultures. J Clin Microbiol 39(9):3393 3394 17. Beutz M, Sherman G, Mayfield J, Fraser VJ, Kollef MH (2003) Clinical utility of blood cultures drawn from central vein catheters and peripheral venipuncture in critically ill medical patients. Chest 123(3):854 861 18. Cockerill FR 3rd, Wilson JW, Vetter EA, Goodman KM, Torgerson CA, Harmsen WS, Schleck CD, Ilstrup DM, Washington JA 2nd, Wilson WR (2004) Optimal testing parameters for blood cultures. Clin Infect Dis 38(12):1724 1730, Epub 2004 May 25 19. Schifman RB, Strand CL, Braun E, Louis-Charles A, Spark RP, Fried ML (1991) Solitary blood cultures as a quality assurance indicator. Qual Assur Util Rev 6(4):132 137 20. Weinstein MP, Towns ML, Quartey SM, Mirrett S, Reimer LG, Parmigiani G, Reller LB (1997) The clinical significance of positive blood cultures in the 1990s: a prospective comprehensive evaluation of the microbiology, epidemiology, and outcome of bacteremia and fungemia in adults. Clin Infect Dis 24(4):584 602 21. Garrouste-Orgeas M, Chevret S, Mainardi JL, Timsit JF, Misset B, Carlet J (2000) A one-year prospective study of nosocomial bacteraemia in ICU and non-icu patients and its impact on patient outcome. J Hosp Infect 44(3):206 213 22. Diekema DJ, Beekmann SE, Chapin KC, Morel KA, Munson E, Doern GV (2003) Epidemiology and outcome of nosocomial and community-onset bloodstream infection. J Clin Microbiol 41 (8):3655 3660 23. Scheckler WE, Bobula JA, Beamsley MB, Hadden ST (2003) Bloodstream infections in a community hospital: a 25-year follow-up. Infect Control Hosp Epidemiol 24(12):936 941 24. Fenner L, Widmer AF, Straub C, Frei R (2008) Is the incidence of anaerobic bacteremia decreasing? Analysis of 114,000 blood cultures over a ten-year period. J Clin Microbiol 46(7):2432 2434, Epub 2008 May 7 25. Lazarovitch T, Freimann S, Shapira G, Blank H (2010) Decrease in anaerobe-related bacteraemias and increase in Bacteroides species isolation rate from 1998 to 2007: a retrospective study. Anaerobe 16(3):201 205, Epub 2009 Sep 23 26. Tsalik EL, Jones D, Nicholson B, Waring L, Liesenfeld O, Park LP, Glickman SW, Caram LB, Langley RJ, van Velkinburgh JC, Cairns CB, Rivers EP, Otero RM, Kingsmore SF, Lalani T, Fowler VG, Woods CW (2010) Multiplex PCR to diagnose bloodstream infections in patients admitted from the emergency department with sepsis. J Clin Microbiol 48(1):26 33, Epub 2009 Oct 21 27. Bates DW, Goldman L, Lee TH (1991) Contaminant blood cultures and resource utilization. The true consequences of falsepositive results. JAMA 265(3):365 369 28. Little JR, Trovillion E, Fraser V (1997) High frequency of pseudobacteremia at a university hospital. Infect Control Hosp Epidemiol 18(3):200 202 29. Archibald LK, Pallangyo K, Kazembe P, Reller LB (2006) Blood culture contamination in Tanzania, Malawi, and the United States: a microbiological tale of three cities. J Clin Microbiol 44 (12):4425 4429, Epub 2006 Oct 4 30. Kallel H, Dammak H, Mahjoubi F, Bahloul M, Turki E, Chelly H, Rekik N, Hammami A, Bouaziz M (2006) Contamination of blood cultures drawn from central vein catheter and peripheral venipuncture. Prospective study of 75 pairs. Pathol Biol 54(1):44 48 31. Hall KK, Lyman JA (2006) Updated review of blood culture contamination. Clin Microbiol Rev 19(4):788 802 32. Peacock SJ, Bowler IC, Crook DW (1995) Positive predictive value of blood cultures growing coagulase-negative staphylococci. Lancet 346(8968):191 192 33. Mirrett S, Weinstein MP, Reimer LG, Wilson ML, Reller LB (2001) Relevance of the number of positive bottles in determining clinical significance of coagulase-negative staphylococci in blood cultures. J Clin Microbiol 39(9):3279 3281 34. Gupta P, Kumhar GD, Kaur G, Ramachandran VG (2005) Clinical significance of polymicrobial bacteremia in newborns. J Paediatr Child Health 41(7):365 368