- Email: [email protected]
A cost-effective method for the presumptive identification of Enterobacteriaceae for diagnostic microbiology laboratories
Pathology
ISSN: 0031-3025 (Print) 1465-3931 (Online) Journal homepage: http://www.tandfonline.com/loi/ipat20
A cost-effective method for the presumptive identification of Enterobacteriaceae for diagnostic microbiology laboratories Lily Siew Yong Ng, Thean Yen Tan & Susan Choo San Yeow To cite this article: Lily Siew Yong Ng, Thean Yen Tan & Susan Choo San Yeow (2010) A costeffective method for the presumptive identification of Enterobacteriaceae for diagnostic microbiology laboratories, Pathology, 42:3, 280-283 To link to this article: http://dx.doi.org/10.3109/00313021003631338
Published online: 30 Mar 2010.
Submit your article to this journal
Article views: 35
View related articles
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ipat20 Download by: [FU Berlin]
Date: 07 November 2016, At: 08:41
Pathology (April 2010) 42(3), pp. 280–283
MICROBIOLOGY
A cost-effective method for the presumptive identification of Enterobacteriaceae for diagnostic microbiology laboratories LILY SIEW YONG NG, THEAN YEN TAN
AND
SUSAN CHOO SAN YEOW
Division of Laboratory Medicine, Changi General Hospital, Singapore
Summary Aims: This study evaluated the use of an abbreviated algorithm for the presumptive identification of Enterobacteriaceae isolated from clinical microbiology specimens. Methods: Identification was based on primary isolation of bacterial pathogens on blood, lactose fermentation based on colonial morphology on MacConkey agar, oxidase and indole tests, and a limited number of conventional biochemical tests. The accuracy of the study algorithm was prospectively evaluated against commercial bacterial identification kits, using clinical isolates from blood, urine and superficial wound and tissue sites. Results: Of 534 isolates, 518 (97%) were accurately identified to genus level. Identification of the study isolates was achieved with a 56% reduction in technologist time and 85% reduction in reagent costs, when compared to the use of a conventional biochemical identification panel. The main limitation of the protocol in the tested bacterial population was that indole-negative Escherichia coli were likely to be misidentified as Enterobacter species. Conclusions: This protocol may be suitable for the presumptive identification of commonly isolated Enterobacteriaceae from non-sterile sites by diagnostic laboratories in resourceconstrained settings. Key words: Enterobacteriaceae, Escherichia coli, Klebsiella, Enterobacter, Proteus. Received 10 September, revised 14 October, accepted 18 October 2009
CLSI standard in a routine clinical setting for the identification of E. coli. 4 We developed an algorithm for the identification of the more frequently clinically encountered genera in the Enterobacteriaceae family based on lactose fermentation on MacConkey agar plate (MAC) and indole production, followed by a limited set of biochemical tests. The algorithm was intended for use with media that is commonly used for the primary isolation of bacterial pathogens from clinical samples, and was developed from initial pilot testing together with the adoption of some tests from the M35-A document. Unlike the M35-A document that focused on rapid testing methods, this study protocol utilised both rapid and conventional biochemical tests with the aim to identify Enterobacteriaceae to genus level using minimal consumables and labour. The aim of the study was not to provide definitive bacterial identification, but to compare the utility of bacterial identification using limited resources against commercially available biochemical identification kits that are commonly used in clinical microbiology laboratories.
MATERIALS AND METHODS Growth of oxidase negative, Gram negative bacilli was observed on trypticase soy agar with 5% sheep blood agar (BAP) and MacConkey agar plates (MAC). Tests were performed from colonies obtained from BAP. An exemption for ethical review was provided for this study by the local Institutional Review Board. Identification methods
INTRODUCTION The identification of bacterial isolates from clinical samples remains one of the fundamental tasks of a clinical microbiology laboratory. However, many laboratories also face financial and labour constraints.1 The identification of Gram negative bacilli has traditionally been performed using a panel of inoculated biochemical reactions, but there is also increasing use of commercial identification kits and semi-automated systems, such as Vitek or Phoenix. The usefulness of simple biochemical reactions or tests to identify bacteria was recognised by the Clinical Laboratory Standards Institute (CLSI) when they released the M35-A standard for the identification of common bacterial isolates.2 However, the M35-A document relies primarily on rapid test methods, and identifies only Escherichia coli and Proteus species.3 To date, we are aware of only one published article that validates the use of the
Step-wise testing was performed according to the criteria in the test flowchart, as shown in Fig. 1. Based on step-wise criteria, organisms were identified using six separate arms of the protocol (labelled A, B, C, D, F1– F5 and G). Gram negative bacilli with characteristic swarming motility on BAP were identified as Proteus species, and further speciation was achieved using Protocol G of the identification flowchart. Non-swarming, oxidase negative, Gram negative bacilli were tested for indole production using a rapid indole test (Becton Dickinson, USA) and observation of lactose fermentation on MAC. Adjunctive identification methods for lactose fermenting, indole positive, oxidase negative, Gram negative bacilli Beta-haemolytic isolates were identified as E. coli (Protocol B).5 Rapid testing for PYRase activity (OBIS Pyr; Oxoid, United Kingdom) was performed on non-beta-haemolytic isolates, and PYR negative isolates were identified as E. coli (Protocol C). The following two tests were performed on PYR positive isolates: lysine decarboxylation, and the Voges and Proskauer (VP) test. The results from these overnight tests allowed
Print ISSN 0031-3025/Online ISSN 1465-3931 # 2010 Royal College of Pathologists of Australasia DOI: 10.3109/00313021003631338
PRESUMPTIVE IDENTIFICATION OF ENTEROBACTERIACEAE
281
Fig. 1 Flow-chart used to aid the presumptive identification of Enterobacteriaceae (Protocols A–H). (þ), positive; (7), negative; b-Glu, beta-glucuronidase testing; F, glucose fermenter; H2S, presence of hydrogen sulphide; Indole, rapid indole testing; KG, Kliger iron agar; Lys, lysine decarboxylase; Mot, motility; Orn, ornithine decarboxylase; PYR, PYRase activity; VP, Voges and Proskauer test.
discrimination between Klebsiella species and Citrobacter species by referring to Table 1 (Protocol D).
Table 1 Supplemental tests for Protocols D and F3 Test
Adjunctive identification methods for non-lactose fermenting, indole positive, oxidase negative, Gram negative bacilli These isolates were tested for the presence of beta-glucuronidase (Rosco, Denmark). Isolates positive for beta-glucuronidase were identified as E. coli (Protocol A).6 All other tested isolates required identification by an alternative method (Protocol H).
Adjunctive identification methods for lactose fermenting, indole negative, oxidase negative, Gram negative bacilli A set of three additional tests was performed: glucose fermentation and hydrogen sulphide (H2S) production using Kliger Iron agar (KIA), ornithine decarboxylase, and motility using Motility Test Medium (all three media from Biomedia Laboratories, Singapore). Glucose fermenting, motile isolates that do not produce H2S were identified as Enterobacter species (Protocols F1 and F5). Glucose fermenting, non-motile, ornithine negative isolates were identified as Klebsiella species (Protocol F2). Glucosefermenting, motile isolates that produced H2S were identified as Citrobacter freundii (Protocol F4). Glucose fermenting, ornithine positive, H2S negative, non-motile isolates required an additional VP test to differentiate between Enterobacter species (VP positive) and E. coli (VP negative) (Protocol F3). Alternatively, such isolates could be identified by a rapid 4 h commercial assay such as RapID ONE (Remel, United Kingdom).
Validation of identification methods The algorithm was prospectively evaluated against 596 clinical isolates of Enterobacteriaceae (see Table 2), predominantly isolated from urinary tract specimens (n ¼ 301) but also from wound and tissue cultures (n ¼ 162) and
Protocol
VP
Lys
Identification
Protocol D
+ 7 7 þ
þ 7
Klebsiella species Citrobacter species E. coli Enterobacter species
Protocol F3
þ, positive; 7 , negative; + , variable; Lys, lysine decarboxylation; VP, Voges and Proskauer.
blood cultures (n ¼ 133). The identification of the test isolates was first determined by RapID ONE (Remel) or Vitek ID-GN (Biome´rieux, France), after which study isolates were identified using the study algorithm. The accuracy of bacterial identification was compared to those obtained by the commercial kits.
Determination of labour requirements and consumable costs The time required to set up, perform and interpret each individual test was determined by stopwatch, and was measured over multiple occasions for different technologists. Timings for ID-GN testing included only test set up, as testing and result interpretation were performed by the semiautomated Vitek 2 instrument (Biome´rieux, France). The average time for each individual test was used to calculate the time required to complete testing for each protocol. The cost of each individual test was calculated using the most recent available costing in 2008 in Singapore dollars (SGD), and converted to
282
NG et al.
Pathology (2010), 42(3), April
Table 2 Bacterial isolates tested by the study method
Genus Citrobacter species Enterobacter species
No. tested isolates
No. identified correctly (%)
Tested by protocol (n)
45
45 (100%)
D (4) F4 (41) F1 (120) F3 (16) F5 (4) A (20) B (12) C (95) F1–F5 (36) D (28) F2 (143) F3 (3) G D (1) F3 (1)
140
139 (99%)
E. coli
163
151 (93%)
Klebsiella species
174
173 (99%)
Proteus species Other Enterobacteriaceae
10 2
10 (100%) 1 (50%)
United States dollars (USD) based on an exchange rate of $1.00 SGD to $0.75 USD. The cost of testing was calculated for each protocol, based on the sum of the costs of each individual test required.
RESULTS Identification of species Using step-wise testing (Fig. 1), 62 isolates of the 596 study isolates required identification using alternative methods (Protocol H), and were excluded from the analysis for accuracy of identification. These 62 isolates included Citrobacter koseri (n ¼ 38), Citrobacter species (n ¼ 5), Edwardsiella tarda (n ¼ 2), E. coli (n ¼ 9), non-swarming Proteus vulgaris (n ¼ 2), Morganella morganii (n ¼ 5) and Providencia species (n ¼ 1). Of the remaining 534 isolates, 97% (n ¼ 518) were accurately identified to genus level (Table 2). The identification protocols A, B, C and G of the study algorithm achieved 100% accuracy in identification. For the remaining protocols, the most common organisms that were misidentified were indole negative strains of E. coli (n ¼ 12), which were generally misidentified as Enterobacter species or Klebsiella species by protocols F1–F5. One isolate of Enterobacter aerogenes was misidentified as E. coli (Protocol F3), and another isolate of Klebsiella oxytoca was misidentified as Enterobacter species (Protocol D). Finally, two isolates of Kluyvera ascorbata were misidentified as E. coli (Protocols F3 and D). The presumptive genus of 492 (92.1%) of the 534 isolates identified by the study protocol were available within 24 hours of primary isolation on BAP and MAC media, which is comparable to the turnaround time achieved by the use of conventional identification kits. The remaining 42 (7.9%) were identified within 48 hours of primary isolation. Labour and costs of testing The time taken by technologists to identify organisms in the test panel using the study algorithm was calculated to be 40.0 hours. This included the time required to perform a RapID One for test organisms in Protocol H. The equivalent technologist time taken to identify the same organisms by RapID One or Vitek ID-GN was 70.7 hours and 57.8 hours, respectively.
The cost of test consumables required to identify the study organisms was calculated to be USD $1463 for the study algorithm, USD $1721 using RapID One and USD $3230 using Vitek ID-GN.
DISCUSSION This study evaluated the accuracy of an abbreviated algorithm for the provisional identification of commonly encountered Enterobacteriaceae within a clinical microbiology setting. The advantages of the algorithm are that it requires only commonly available bacteriological media, with potential savings in reagent costs and technologist time. The protocol achieved over 95% accuracy, but frequently misidentified indole negative strains of E. coli as Enterobacter species. There were several limitations to the abbreviated algorithm. The study method would not identify E. coli that were negative for beta-glucuronidase. Enterobacteriaceae are generally identified only to genus level. However, the most useful aspect of bacterial identification is to enable the application of interpretative susceptibility criteria and, in general, genus identification is sufficient to apply such susceptibility interpretations.7 Interpretation of lactose fermentation on MAC may be subjective. For example, Acinetobacter species are known to retain a pink hue on MAC,8 while late lactose fermenters such as Enterobacter species may demonstrate faint colour changes on MAC.9 The unwary microbiologist may mistake pigmented Serratia species as a lactose fermenter, while Aeromonas species (although quickly differentiated by an oxidase test) may also be mistaken as lactose fermenting Enterobacteriaceae. The actual savings in cost and time will depend on the distribution of bacterial pathogens isolated from each individual laboratory and will also be affected by local cost structures in each laboratory (for example, reagent costs and labour costs). However, the data presented in this study should make it possible for each laboratory to estimate the possible financial savings. Bacteria are complex organisms, and identification by a limited number of biochemical tests will always be subject to phenotypic variation within a given bacterial population. For example, Klebsiella species and Enterobacter species are differentiated in this testing algorithm by motility and ornithine decarboxylase. However, ornithine negative (or weakly positive ornithine) isolates of Enterobacter species may be misidentified as Klebsiella species.10 Similarly, 51% of Shigella species and Salmonella species may be both indole and beta-glucuronidase positive.11 Because of the sheer diversity of Enterobacteriaceae, definitive identification using a limited number of tests is not possible, and misidentification of uncommonly encountered bacterial genera by the study algorithm is likely. Because of these limitations, this protocol would not be appropriate for the identification of Enterobacteriaceae from sterile sites, but may be useful for bacterial identification from urine cultures or for microbiology laboratories practising in resource-constrained settings. The authors would also recommend that other laboratories validate the accuracy of this methodology, as bacterial ecology is often distinct to different hospital or regional settings.
PRESUMPTIVE IDENTIFICATION OF ENTEROBACTERIACEAE
While there are other specialised media that allow rapid presumptive diagnosis of some Enterobacteriaceae, most of these are not suitable for primary bacterial isolation from clinical samples. The use of chromogenic media together with limited additional biochemical testing has been reported to accurately identify most common Enterobacteriaceae isolated from urine and other clinical samples.12 However, chromogenic media are costly and light-sensitive and therefore have specific storage requirements and a shorter shelf life, requiring more complex media inventory management. In contrast, the basic media used in this study are readily available in most microbiology laboratories, and are commonly used as primary isolation media for clinical microbiological culture. Minor variations to the abbreviated algorithm may also be possible. Although we did not evaluate other media, it is possible that any plate medium that allows the differentiation of lactose fermentation may be used in place of MacConkey agar used in this study, e.g., eosin methylene blue agar plates. Similarly, although we used separate media for the detection of ornithine decarboxylase and motility, a single combination tube test utilising ornithine-indole-motility media (Becton Dickinson) may be used instead to further simplify the testing protocols. In conclusion, this study evaluated the use of a stepwise abbreviated protocol for the presumptive identification of Enterobacteriaceae against commonly available commercial identification kits. The protocol successfully identified over 95% of study isolates to genus level with savings in both technologist time and consumable costs, and may be suitable for use in a resource-constrained setting. Because of limitations in the study methodology, individual laboratories should validate this protocol in a local setting prior to implementation of the scheme. Acknowledgements: This work was supported by a grant from the National Medical Research Council, Singapore.
283
Address for correspondence: Dr T. Y. Tan, Division of Laboratory Medicine, Changi General Hospital, 2 Simei Street 3, Singapore, 529889. E-mail: [email protected]
References 1. Peterson LR, Hamilton JD, Baron EJ, et al. Role of clinical microbiology laboratories in the management and control of infectious diseases and the delivery of health care. Clin Infect Dis 2001; 32: 605–11. 2. National Committee for Clinical Laboratory Standards (NCCLS). Abbreviated Identification of Bacteria and Yeast; Approved Guideline. NCCLS document M35-A. 25th ed. Wayne, Pennsylvania: Clinical Laboratory Standards Institute, 2005. 3. Baron EJ. Rapid identification of bacteria and yeast: summary of a National Committee for Clinical Laboratory Standards proposed guideline. Clin Infect Dis 2001; 33: 220–5. 4. York MK, Baron EJ, Clarridge JE, et al. Multilaboratory validation of rapid spot tests for identification of Escherichia coli. J Clin Microbiol 2000; 38: 3394–8. 5. Clinical Laboratory Standards Institute (CLSI). Abbreviated Identification of Bacteria and Yeast; Approved Guideline. NCCLS document M35-A2. Wayne, Pennsylvania: CLSI, 2008. 6. Perez JL, Berrocal CI, Berrocal L. Evaluation of a commercial betaglucuronidase test for the rapid and economical identification of Escherichia coli. J Appl Bacteriol 1986; 61: 541–5. 7. Livermore DM, Winstanley TG, Shannon KP. Interpretative reading: recognizing the unusual and inferring resistance mechanisms from resistance phenotypes. J Antimicrob Chemother 2001; 48 (Suppl 1): 87–102. 8. Schreckenberger PC, Daneshvar MI, Hollis DG. Manual of Clinical Microbiology. 9th ed. Washington, DC: American Society of Microbiology, 2007; Acinetobacter, Achromobacter, Chryseobacterium, Moraxella, and other nonfermentative Gram-negative rods. 9. Flournoy DJ, Wongpradit S, Silberg SL. Facilitating identification of lactose-fermenting Enterobacteriaceae on MacConkey agar. Proc Oklahoma Acad Sci 1990; 70: 5–8. 10. Claeys G, De Baere T, Wauters G, et al. Extended-spectrum betalactamase (ESBL) producing Enterobacter aerogenes phenotypically misidentified as Klebsiella pneumoniae or K. terrigena. BMC Microbiol 2004; 4: 49. 11. Heizmann W, Doller PC, Gutbrod B, et al. Rapid identification of Escherichia coli by Fluorocult media and positive indole reaction. J Clin Microbiol 1988; 26: 2682–4. 12. Ohkusu K. Cost-effective and rapid presumptive identification of gramnegative bacilli in routine urine, pus, and stool cultures: evaluation of the use of CHROMagar orientation medium in conjunction with simple biochemical tests. J Clin Microbiol 2000; 38: 4586–92.
ISSN: 0031-3025 (Print) 1465-3931 (Online) Journal homepage: http://www.tandfonline.com/loi/ipat20
A cost-effective method for the presumptive identification of Enterobacteriaceae for diagnostic microbiology laboratories Lily Siew Yong Ng, Thean Yen Tan & Susan Choo San Yeow To cite this article: Lily Siew Yong Ng, Thean Yen Tan & Susan Choo San Yeow (2010) A costeffective method for the presumptive identification of Enterobacteriaceae for diagnostic microbiology laboratories, Pathology, 42:3, 280-283 To link to this article: http://dx.doi.org/10.3109/00313021003631338
Published online: 30 Mar 2010.
Submit your article to this journal
Article views: 35
View related articles
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ipat20 Download by: [FU Berlin]
Date: 07 November 2016, At: 08:41
Pathology (April 2010) 42(3), pp. 280–283
MICROBIOLOGY
A cost-effective method for the presumptive identification of Enterobacteriaceae for diagnostic microbiology laboratories LILY SIEW YONG NG, THEAN YEN TAN
AND
SUSAN CHOO SAN YEOW
Division of Laboratory Medicine, Changi General Hospital, Singapore
Summary Aims: This study evaluated the use of an abbreviated algorithm for the presumptive identification of Enterobacteriaceae isolated from clinical microbiology specimens. Methods: Identification was based on primary isolation of bacterial pathogens on blood, lactose fermentation based on colonial morphology on MacConkey agar, oxidase and indole tests, and a limited number of conventional biochemical tests. The accuracy of the study algorithm was prospectively evaluated against commercial bacterial identification kits, using clinical isolates from blood, urine and superficial wound and tissue sites. Results: Of 534 isolates, 518 (97%) were accurately identified to genus level. Identification of the study isolates was achieved with a 56% reduction in technologist time and 85% reduction in reagent costs, when compared to the use of a conventional biochemical identification panel. The main limitation of the protocol in the tested bacterial population was that indole-negative Escherichia coli were likely to be misidentified as Enterobacter species. Conclusions: This protocol may be suitable for the presumptive identification of commonly isolated Enterobacteriaceae from non-sterile sites by diagnostic laboratories in resourceconstrained settings. Key words: Enterobacteriaceae, Escherichia coli, Klebsiella, Enterobacter, Proteus. Received 10 September, revised 14 October, accepted 18 October 2009
CLSI standard in a routine clinical setting for the identification of E. coli. 4 We developed an algorithm for the identification of the more frequently clinically encountered genera in the Enterobacteriaceae family based on lactose fermentation on MacConkey agar plate (MAC) and indole production, followed by a limited set of biochemical tests. The algorithm was intended for use with media that is commonly used for the primary isolation of bacterial pathogens from clinical samples, and was developed from initial pilot testing together with the adoption of some tests from the M35-A document. Unlike the M35-A document that focused on rapid testing methods, this study protocol utilised both rapid and conventional biochemical tests with the aim to identify Enterobacteriaceae to genus level using minimal consumables and labour. The aim of the study was not to provide definitive bacterial identification, but to compare the utility of bacterial identification using limited resources against commercially available biochemical identification kits that are commonly used in clinical microbiology laboratories.
MATERIALS AND METHODS Growth of oxidase negative, Gram negative bacilli was observed on trypticase soy agar with 5% sheep blood agar (BAP) and MacConkey agar plates (MAC). Tests were performed from colonies obtained from BAP. An exemption for ethical review was provided for this study by the local Institutional Review Board. Identification methods
INTRODUCTION The identification of bacterial isolates from clinical samples remains one of the fundamental tasks of a clinical microbiology laboratory. However, many laboratories also face financial and labour constraints.1 The identification of Gram negative bacilli has traditionally been performed using a panel of inoculated biochemical reactions, but there is also increasing use of commercial identification kits and semi-automated systems, such as Vitek or Phoenix. The usefulness of simple biochemical reactions or tests to identify bacteria was recognised by the Clinical Laboratory Standards Institute (CLSI) when they released the M35-A standard for the identification of common bacterial isolates.2 However, the M35-A document relies primarily on rapid test methods, and identifies only Escherichia coli and Proteus species.3 To date, we are aware of only one published article that validates the use of the
Step-wise testing was performed according to the criteria in the test flowchart, as shown in Fig. 1. Based on step-wise criteria, organisms were identified using six separate arms of the protocol (labelled A, B, C, D, F1– F5 and G). Gram negative bacilli with characteristic swarming motility on BAP were identified as Proteus species, and further speciation was achieved using Protocol G of the identification flowchart. Non-swarming, oxidase negative, Gram negative bacilli were tested for indole production using a rapid indole test (Becton Dickinson, USA) and observation of lactose fermentation on MAC. Adjunctive identification methods for lactose fermenting, indole positive, oxidase negative, Gram negative bacilli Beta-haemolytic isolates were identified as E. coli (Protocol B).5 Rapid testing for PYRase activity (OBIS Pyr; Oxoid, United Kingdom) was performed on non-beta-haemolytic isolates, and PYR negative isolates were identified as E. coli (Protocol C). The following two tests were performed on PYR positive isolates: lysine decarboxylation, and the Voges and Proskauer (VP) test. The results from these overnight tests allowed
Print ISSN 0031-3025/Online ISSN 1465-3931 # 2010 Royal College of Pathologists of Australasia DOI: 10.3109/00313021003631338
PRESUMPTIVE IDENTIFICATION OF ENTEROBACTERIACEAE
281
Fig. 1 Flow-chart used to aid the presumptive identification of Enterobacteriaceae (Protocols A–H). (þ), positive; (7), negative; b-Glu, beta-glucuronidase testing; F, glucose fermenter; H2S, presence of hydrogen sulphide; Indole, rapid indole testing; KG, Kliger iron agar; Lys, lysine decarboxylase; Mot, motility; Orn, ornithine decarboxylase; PYR, PYRase activity; VP, Voges and Proskauer test.
discrimination between Klebsiella species and Citrobacter species by referring to Table 1 (Protocol D).
Table 1 Supplemental tests for Protocols D and F3 Test
Adjunctive identification methods for non-lactose fermenting, indole positive, oxidase negative, Gram negative bacilli These isolates were tested for the presence of beta-glucuronidase (Rosco, Denmark). Isolates positive for beta-glucuronidase were identified as E. coli (Protocol A).6 All other tested isolates required identification by an alternative method (Protocol H).
Adjunctive identification methods for lactose fermenting, indole negative, oxidase negative, Gram negative bacilli A set of three additional tests was performed: glucose fermentation and hydrogen sulphide (H2S) production using Kliger Iron agar (KIA), ornithine decarboxylase, and motility using Motility Test Medium (all three media from Biomedia Laboratories, Singapore). Glucose fermenting, motile isolates that do not produce H2S were identified as Enterobacter species (Protocols F1 and F5). Glucose fermenting, non-motile, ornithine negative isolates were identified as Klebsiella species (Protocol F2). Glucosefermenting, motile isolates that produced H2S were identified as Citrobacter freundii (Protocol F4). Glucose fermenting, ornithine positive, H2S negative, non-motile isolates required an additional VP test to differentiate between Enterobacter species (VP positive) and E. coli (VP negative) (Protocol F3). Alternatively, such isolates could be identified by a rapid 4 h commercial assay such as RapID ONE (Remel, United Kingdom).
Validation of identification methods The algorithm was prospectively evaluated against 596 clinical isolates of Enterobacteriaceae (see Table 2), predominantly isolated from urinary tract specimens (n ¼ 301) but also from wound and tissue cultures (n ¼ 162) and
Protocol
VP
Lys
Identification
Protocol D
+ 7 7 þ
þ 7
Klebsiella species Citrobacter species E. coli Enterobacter species
Protocol F3
þ, positive; 7 , negative; + , variable; Lys, lysine decarboxylation; VP, Voges and Proskauer.
blood cultures (n ¼ 133). The identification of the test isolates was first determined by RapID ONE (Remel) or Vitek ID-GN (Biome´rieux, France), after which study isolates were identified using the study algorithm. The accuracy of bacterial identification was compared to those obtained by the commercial kits.
Determination of labour requirements and consumable costs The time required to set up, perform and interpret each individual test was determined by stopwatch, and was measured over multiple occasions for different technologists. Timings for ID-GN testing included only test set up, as testing and result interpretation were performed by the semiautomated Vitek 2 instrument (Biome´rieux, France). The average time for each individual test was used to calculate the time required to complete testing for each protocol. The cost of each individual test was calculated using the most recent available costing in 2008 in Singapore dollars (SGD), and converted to
282
NG et al.
Pathology (2010), 42(3), April
Table 2 Bacterial isolates tested by the study method
Genus Citrobacter species Enterobacter species
No. tested isolates
No. identified correctly (%)
Tested by protocol (n)
45
45 (100%)
D (4) F4 (41) F1 (120) F3 (16) F5 (4) A (20) B (12) C (95) F1–F5 (36) D (28) F2 (143) F3 (3) G D (1) F3 (1)
140
139 (99%)
E. coli
163
151 (93%)
Klebsiella species
174
173 (99%)
Proteus species Other Enterobacteriaceae
10 2
10 (100%) 1 (50%)
United States dollars (USD) based on an exchange rate of $1.00 SGD to $0.75 USD. The cost of testing was calculated for each protocol, based on the sum of the costs of each individual test required.
RESULTS Identification of species Using step-wise testing (Fig. 1), 62 isolates of the 596 study isolates required identification using alternative methods (Protocol H), and were excluded from the analysis for accuracy of identification. These 62 isolates included Citrobacter koseri (n ¼ 38), Citrobacter species (n ¼ 5), Edwardsiella tarda (n ¼ 2), E. coli (n ¼ 9), non-swarming Proteus vulgaris (n ¼ 2), Morganella morganii (n ¼ 5) and Providencia species (n ¼ 1). Of the remaining 534 isolates, 97% (n ¼ 518) were accurately identified to genus level (Table 2). The identification protocols A, B, C and G of the study algorithm achieved 100% accuracy in identification. For the remaining protocols, the most common organisms that were misidentified were indole negative strains of E. coli (n ¼ 12), which were generally misidentified as Enterobacter species or Klebsiella species by protocols F1–F5. One isolate of Enterobacter aerogenes was misidentified as E. coli (Protocol F3), and another isolate of Klebsiella oxytoca was misidentified as Enterobacter species (Protocol D). Finally, two isolates of Kluyvera ascorbata were misidentified as E. coli (Protocols F3 and D). The presumptive genus of 492 (92.1%) of the 534 isolates identified by the study protocol were available within 24 hours of primary isolation on BAP and MAC media, which is comparable to the turnaround time achieved by the use of conventional identification kits. The remaining 42 (7.9%) were identified within 48 hours of primary isolation. Labour and costs of testing The time taken by technologists to identify organisms in the test panel using the study algorithm was calculated to be 40.0 hours. This included the time required to perform a RapID One for test organisms in Protocol H. The equivalent technologist time taken to identify the same organisms by RapID One or Vitek ID-GN was 70.7 hours and 57.8 hours, respectively.
The cost of test consumables required to identify the study organisms was calculated to be USD $1463 for the study algorithm, USD $1721 using RapID One and USD $3230 using Vitek ID-GN.
DISCUSSION This study evaluated the accuracy of an abbreviated algorithm for the provisional identification of commonly encountered Enterobacteriaceae within a clinical microbiology setting. The advantages of the algorithm are that it requires only commonly available bacteriological media, with potential savings in reagent costs and technologist time. The protocol achieved over 95% accuracy, but frequently misidentified indole negative strains of E. coli as Enterobacter species. There were several limitations to the abbreviated algorithm. The study method would not identify E. coli that were negative for beta-glucuronidase. Enterobacteriaceae are generally identified only to genus level. However, the most useful aspect of bacterial identification is to enable the application of interpretative susceptibility criteria and, in general, genus identification is sufficient to apply such susceptibility interpretations.7 Interpretation of lactose fermentation on MAC may be subjective. For example, Acinetobacter species are known to retain a pink hue on MAC,8 while late lactose fermenters such as Enterobacter species may demonstrate faint colour changes on MAC.9 The unwary microbiologist may mistake pigmented Serratia species as a lactose fermenter, while Aeromonas species (although quickly differentiated by an oxidase test) may also be mistaken as lactose fermenting Enterobacteriaceae. The actual savings in cost and time will depend on the distribution of bacterial pathogens isolated from each individual laboratory and will also be affected by local cost structures in each laboratory (for example, reagent costs and labour costs). However, the data presented in this study should make it possible for each laboratory to estimate the possible financial savings. Bacteria are complex organisms, and identification by a limited number of biochemical tests will always be subject to phenotypic variation within a given bacterial population. For example, Klebsiella species and Enterobacter species are differentiated in this testing algorithm by motility and ornithine decarboxylase. However, ornithine negative (or weakly positive ornithine) isolates of Enterobacter species may be misidentified as Klebsiella species.10 Similarly, 51% of Shigella species and Salmonella species may be both indole and beta-glucuronidase positive.11 Because of the sheer diversity of Enterobacteriaceae, definitive identification using a limited number of tests is not possible, and misidentification of uncommonly encountered bacterial genera by the study algorithm is likely. Because of these limitations, this protocol would not be appropriate for the identification of Enterobacteriaceae from sterile sites, but may be useful for bacterial identification from urine cultures or for microbiology laboratories practising in resource-constrained settings. The authors would also recommend that other laboratories validate the accuracy of this methodology, as bacterial ecology is often distinct to different hospital or regional settings.
PRESUMPTIVE IDENTIFICATION OF ENTEROBACTERIACEAE
While there are other specialised media that allow rapid presumptive diagnosis of some Enterobacteriaceae, most of these are not suitable for primary bacterial isolation from clinical samples. The use of chromogenic media together with limited additional biochemical testing has been reported to accurately identify most common Enterobacteriaceae isolated from urine and other clinical samples.12 However, chromogenic media are costly and light-sensitive and therefore have specific storage requirements and a shorter shelf life, requiring more complex media inventory management. In contrast, the basic media used in this study are readily available in most microbiology laboratories, and are commonly used as primary isolation media for clinical microbiological culture. Minor variations to the abbreviated algorithm may also be possible. Although we did not evaluate other media, it is possible that any plate medium that allows the differentiation of lactose fermentation may be used in place of MacConkey agar used in this study, e.g., eosin methylene blue agar plates. Similarly, although we used separate media for the detection of ornithine decarboxylase and motility, a single combination tube test utilising ornithine-indole-motility media (Becton Dickinson) may be used instead to further simplify the testing protocols. In conclusion, this study evaluated the use of a stepwise abbreviated protocol for the presumptive identification of Enterobacteriaceae against commonly available commercial identification kits. The protocol successfully identified over 95% of study isolates to genus level with savings in both technologist time and consumable costs, and may be suitable for use in a resource-constrained setting. Because of limitations in the study methodology, individual laboratories should validate this protocol in a local setting prior to implementation of the scheme. Acknowledgements: This work was supported by a grant from the National Medical Research Council, Singapore.
283
Address for correspondence: Dr T. Y. Tan, Division of Laboratory Medicine, Changi General Hospital, 2 Simei Street 3, Singapore, 529889. E-mail: [email protected]
References 1. Peterson LR, Hamilton JD, Baron EJ, et al. Role of clinical microbiology laboratories in the management and control of infectious diseases and the delivery of health care. Clin Infect Dis 2001; 32: 605–11. 2. National Committee for Clinical Laboratory Standards (NCCLS). Abbreviated Identification of Bacteria and Yeast; Approved Guideline. NCCLS document M35-A. 25th ed. Wayne, Pennsylvania: Clinical Laboratory Standards Institute, 2005. 3. Baron EJ. Rapid identification of bacteria and yeast: summary of a National Committee for Clinical Laboratory Standards proposed guideline. Clin Infect Dis 2001; 33: 220–5. 4. York MK, Baron EJ, Clarridge JE, et al. Multilaboratory validation of rapid spot tests for identification of Escherichia coli. J Clin Microbiol 2000; 38: 3394–8. 5. Clinical Laboratory Standards Institute (CLSI). Abbreviated Identification of Bacteria and Yeast; Approved Guideline. NCCLS document M35-A2. Wayne, Pennsylvania: CLSI, 2008. 6. Perez JL, Berrocal CI, Berrocal L. Evaluation of a commercial betaglucuronidase test for the rapid and economical identification of Escherichia coli. J Appl Bacteriol 1986; 61: 541–5. 7. Livermore DM, Winstanley TG, Shannon KP. Interpretative reading: recognizing the unusual and inferring resistance mechanisms from resistance phenotypes. J Antimicrob Chemother 2001; 48 (Suppl 1): 87–102. 8. Schreckenberger PC, Daneshvar MI, Hollis DG. Manual of Clinical Microbiology. 9th ed. Washington, DC: American Society of Microbiology, 2007; Acinetobacter, Achromobacter, Chryseobacterium, Moraxella, and other nonfermentative Gram-negative rods. 9. Flournoy DJ, Wongpradit S, Silberg SL. Facilitating identification of lactose-fermenting Enterobacteriaceae on MacConkey agar. Proc Oklahoma Acad Sci 1990; 70: 5–8. 10. Claeys G, De Baere T, Wauters G, et al. Extended-spectrum betalactamase (ESBL) producing Enterobacter aerogenes phenotypically misidentified as Klebsiella pneumoniae or K. terrigena. BMC Microbiol 2004; 4: 49. 11. Heizmann W, Doller PC, Gutbrod B, et al. Rapid identification of Escherichia coli by Fluorocult media and positive indole reaction. J Clin Microbiol 1988; 26: 2682–4. 12. Ohkusu K. Cost-effective and rapid presumptive identification of gramnegative bacilli in routine urine, pus, and stool cultures: evaluation of the use of CHROMagar orientation medium in conjunction with simple biochemical tests. J Clin Microbiol 2000; 38: 4586–92.