Results of an interlaboratory test on antimicrobial susceptibility testing of bacteria from animals by broth microdilution

International Journal of Antimicrobial Agents 27 (2006) 482–490

Results of an interlaboratory test on antimicrobial susceptibility testing of bacteria from animals by broth microdilution J¨urgen Wallmann a,∗ , Alexander B¨ottner b , Luc Goossens c , H. Mohamed Hafez d , Katrin Hartmann e , Heike Kaspar a , Corinna Kehrenberg f , Manfred Kietzmann g , Dieter Klarmann h , G¨unter Klein i , Peter Krabisch j , Tilman K¨uhn k , Gabriele Luhofer l , Angelika Richter m , Bianka Schulz e , Stefan Schwarz f , Claudia Sigge n , Wolfgang Traeder c , Karl-Heinz Waldmann o , Christiane Werckenthin p , for the Working Group ‘Antimicrobial Resistance’ of the German Veterinary Society, Eva Zschiesche b a

p

Federal Office of Consumer Protection and Food Safety (BVL), Diedersdorfer Weg 1, D-12277 Berlin, Germany b Intervet Innovation GmbH, Schwabenheim, Germany c Pfizer GmbH, Karlsruhe, Germany d Institut f¨ ur Gefl¨ugelkrankheiten, Freie Universit¨at Berlin, Germany e Medizinische Kleintierklinik, Ludwig-Maximilians-Universit¨ at M¨unchen, Germany f Institut f¨ ur Tierzucht, Bundesforschungsanstalt f¨ur Landwirtschaft (FAL), Neustadt-Mariensee, Germany g Institut f¨ ur Pharmakologie, Toxikologie und Pharmazie, Tier¨arztliche Hochschule Hannover, Germany h Veterin¨ arinstitut Oldenburg, LAVES, Germany i Institut f¨ ur Lebensmittelqualit¨at und -sicherheit, Tier¨arztliche Hochschule Hannover, Germany j TGD Bayern, Poing, Germany k Th¨ uringer Medizinal-, Lebensmittel- und Veterin¨aruntersuchungsamt, Jena, Germany l Landesuntersuchungsamt Rheinland-Pfalz, Institut f¨ ur Lebensmittel tierischer Herkunft, Koblenz, Germany m Institut f¨ ur Pharmakologie und Toxikologie, Freie Universit¨at Berlin, Germany n Bundesverband f¨ ur Tiergesundheit (BfT), Bonn, Germany o Klinik f¨ ur kleine Klauentiere, Tier¨arztliche Hochschule Hannover, Germany Institut f¨ur Medizinische Mikrobiologie, Infektions- und Seuchenmedizin, Ludwig-Maximilians-Universit¨at M¨unchen, Germany Received 10 October 2005; accepted 19 December 2005

Abstract A standard operating procedure for the determination of minimum inhibitory concentrations (MICs) of antimicrobial agents by the broth microdilution method was developed and evaluated for its fitness for use in an interlaboratory ring trial involving 46 routine diagnostic laboratories. All laboratories tested five strains (one reference strain and four field strains) against a total of 22 different antimicrobial agents. Gram-negative strains were tested against 16 different antimicrobial agents and Gram-positive strains against 14 different antimicrobial agents. Tests were performed once a week for three consecutive weeks. At least 80% of the results determined by 35 of the 46 participating laboratories were within the expected range (mode MIC ± 1 dilution step), with the 18 participating laboratories experienced in MIC determination showing a slightly higher mean percentage of accurate results (89.3% reproducible results) than the 28 non-experienced laboratories (86.7% reproducible results). The most accurate results were obtained for the Escherichia coli field strain, whilst the results for the Streptococcus uberis field strain showed the highest error rate. Among the 22 antimicrobial agents tested, the highest variabilities in the results (mean value for all antimicrobial agents 12.3%) were recorded for ceftiofur (27.8%), penicillin G (20.8%) and cefoperazone (20.6%). © 2006 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved. Keywords: Susceptibility testing; Interlaboratory test; Broth microdilution; Minimum inhibitory concentration; Antimicrobial agents

∗

Corresponding author. Tel.: +49 1888 412 2050; fax: +49 1888 412 2956. E-mail address: [email protected] (J. Wallmann).

0924-8579/$ – see front matter © 2006 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved. doi:10.1016/j.ijantimicag.2005.12.011

J. Wallmann et al. / International Journal of Antimicrobial Agents 27 (2006) 482–490

1. Introduction Emerging resistance to antimicrobial agents in bacteria is a worldwide problem in human and veterinary medicine [1,2]. Valid data are necessary to assess the antimicrobial susceptibility of bacterial pathogens in the routine diagnostic laboratory as well as changes in the susceptibility patterns in surveillance and monitoring programmes [3]. Such data can only be generated using standardised procedures for in vitro antimicrobial susceptibility testing [4–7]. There are several different dilution and diffusion methods that are used in routine diagnostic laboratories to determine the in vitro susceptibility of bacterial pathogens to antimicrobial agents. In addition, institutions such as the Clinical and Laboratory Standards Institute (CLSI) in the USA [8], the British Society for Antimicrobial Chemotherapy (BSAC) [9], the French Comit´e de l’Antibiogramme de la Soci´et´e Franc¸aise de Microbiologie (CA-SFM) [10] and the Deutsches Institut f¨ur Normung e.V. (DIN) in Germany [11] have established guidelines for the performance of these tests. A comparison revealed that these guidelines not only differed in relevant parameters such as size of the inoculum or the media allowed, but they also differed in the breakpoints used for assignment of the sizes of the growth inhibition zone determined in disk diffusion tests and the minimum inhibitory concentrations (MICs) determined in dilution tests to the qualitative categories ‘susceptible’, ‘intermediate’ or ‘resistant’ [12,13]. A questionnaire sent to different veterinary diagnostic laboratories in Germany in 2003 revealed that disk diffusion and/or microdilution broth methods are both commonly used and also that the tests are performed and the results are evaluated on the basis of several of the guidelines mentioned above. Based on this information, we concluded that comparison of the results is not necessarily possible. To generate comparable data in the veterinary field based on the use of a standardised microdilution broth procedure and unified microtitre plates, the working group ‘Antimicrobial Resistance’ of the German Veterinary Society (DVG) took action in the following points. First, a standard operating procedure (SOP) for performance of in vitro susceptibility testing was developed. This SOP closely followed the specifications described in the CLSI document M31-A2 [8], since this document is currently the only one worldwide that deals with in vitro susceptibility testing of bacteria from animal origin and contains veterinary-specific breakpoints. In the next step, uniform layouts for microtitre plates were developed for testing bacterial pathogens from food-producing animals (mainly cattle and pigs), from companion animals (mainly dogs, cats and horses) as well as from mastitis cases in dairy cattle [14,15]. These layouts were based on the antimicrobial agents licensed in Germany for the control of the most relevant bacterial pathogens in the respective animal species or in bovine mastitis and the breakpoints available for evaluation of the results. Furthermore, class representatives were used in these layouts based on known cross-resistances

483

between members of specific classes of antimicrobial agents. Finally, well received courses were organised to introduce laboratory personnel to performance of the broth microdilution method and evaluation of the results obtained with this method. Based on these aforementioned activities, a ring trial was started in which a total of 46 laboratories participated. Since correct species identification prior to in vitro susceptibility testing is essential in choosing the correct test medium and supplements (if necessary), all laboratories had to identify the four field strains to species level. In addition to validating the SOP for its fitness for use [16–18], the results were expected to provide information on the consistency of the results for the five different strains, the 22 different antimicrobial agents tested and the three independent test series performed in each laboratory. Moreover, the results from experienced laboratories and from those not experienced in this methodology were compared to evaluate whether experience is a key factor for the determination of accurate MIC values.

2. Materials and methods 2.1. Participating laboratories, test strains and microtitre plates Of the 46 participating laboratories, 43 were from Germany and a single laboratory was from each of Austria, the Czech Republic and The Netherlands. Among all participants, 18 (39.1%) laboratories already used the broth microdilution method as a routine method for in vitro susceptibility testing prior to the study. The bacterial strains (on agar slants), microtitre plates, protocols for performance of the tests as well as standardised forms for recording the results were provided by the Federal Office of Consumer Protection and Food Safety (BVL), Berlin, Germany, and were sent off on the same day to all participating laboratories. Each of the five test strains was given a specific code number. The four field strains were chosen randomly from the strain collection of the German National Resistance Monitoring Programme (GERM-Vet) of the BVL. These field strains represented the species Escherichia coli, Pasteurella multocida, Streptococcus agalactiae and Streptococcus uberis. The E. coli, S. agalactiae and S. uberis field strains were cultured from cases of bovine mastitis, whilst the P. multocida strain was isolated from a case of porcine pneumonia. The reference strain Staphylococcus aureus ATCC 29213, which served as an internal control in this test system, was purchased from the German Collection of Microorganisms and Tissue Cultures (DSMZ, Braunschweig, Germany). The three microtitre plates (Sensititre; Trek Diagnostic Systems, East Grinstead, UK) used in this ring trial were the same as those used in the GERM-Vet programme in 2002/2003, including one common plate for Gram-positive and Gram-negative bacteria (NLV31) and a specific plate

484

J. Wallmann et al. / International Journal of Antimicrobial Agents 27 (2006) 482–490

Table 1 Concentration ranges of antimicrobial agents and modal minimum inhibitory concentration (MIC) values for bacterial strains tested in the interlaboratory test Antimicrobial agent

Concentration range (mg/L)

Modal MIC values (mg/L) Field strains Escherichia coli

Amoxicillin/clavulanic acid (2:1) Ampicillin Apramycin Avilamycin Cefoperazone Cefquinome Ceftiofur Cefalothin Cefazolin Colistin Doxycycline Erythromycin Florfenicol Lincomycin Nalidixic acid Oxacillin + 2% NaCl Penicillin G Quinupristin/dalfopristin Spectinomycin Tiamulin Trimethoprim Vancomycin

Reference strain Pasteurella multocida

Streptococcus agalactiae

Streptococcus uberis

Staphylococcus aureus ATCC 29213a

0.015–32

8

0.25

0.06

0.25

0.25

0.03–64 0.03–64 0.06–128 0.03–16 0.008–16 0.008–16 0.015–32 0.015–32 0.03–16 0.015–32 0.015–32 0.03–64 0.03–64 0.015–32 0.03–64 0.015–32 0.25–128 0.25–512 0.03–64 0.06–128 0.06–128

>64 4 – 2 0.06 0.5 32 4 0.25 16 – 8 – 4 – >32 – 32, 64 >64 >128 –

0.25 4 – 0.03 0.06 0.008 0.25 0.5 0.5 0.25 – 0.5 – 1 – 0.125 – 8 4 1 –

0.06 – 0.5 0.25 0.06 0.25 0.25 0.25 – – 0.03 – 0.25 – 0.25 0.06 0.5 – – – 0.5

0.25 – 1 2 0.125 1 1 0.5 – – 0.06 – 64 – 1 0.125 1 – – – 0.5

0.5 – 16 2 0.5 1 0.25 0.5 – – 0.5 – 1 – 0.25 0.5 0.5 – – – 1

(–) not tested. a Values identical with results from Sensititre® for Clinical and Laboratory Standards Institute (formerly National Committee for Clinical Laboratory Standards) recommended quality control organisms.

for Gram-positive bacteria (NLV21) and Gram-negative bacteria (NLV34). Whilst the reference strain was tested on all three microtitre plates, the laboratories first had to differentiate the four field strains and then use microtitre plates NLV31 + NLV21 for the Gram-positive strains and NLV31 + NLV34 for the Gram-negative strains. The following antimicrobial agents were tested on the different microtitre plates: NLV31, ampicillin, amoxicillin/clavulanic acid (2:1), penicillin G, ceftiofur, cefquinome, cefalothin, cefazolin and cefoperazone; NLV21, avilamycin, cefalothin, lincomycin, erythromycin, oxacillin + 2% NaCl, penicillin G, vancomycin and quinupristin/dalfopristin; NLV34, florfenicol, spectinomycin, tiamulin, apramycin, nalidixic acid, trimethoprim, doxycycline and colistin. Based on the layouts of these microtitre plates, it was unavoidable that Grampositive bacteria were tested on both plates for their MICs to penicillin G and cefalothin. Table 1 shows the test ranges of the 22 antimicrobial agents or combinations of antimicrobial agents as well as the expected modal MIC values of the five test strains. In several cases, this test procedure led to the testing of combinations of bacteria and antimicrobial agents, e.g. E. coli–penicillin G, which are not of clinical relevance owing to intrinsic resistance properties of the test strain.

No methodological specifications were made for species identification of the four field strains. Hence, every participating laboratory performed the species identification by an established in-house standard procedure. 2.2. Performance of the tests According to the test protocol, each participating laboratory had to conduct three test series in total, which should be performed on the same day of the week within three consecutive weeks. Two microtitre plates per field strain and two for the reference strain had to be inoculated during each test series. In addition, each laboratory had to check the inocula for purity and had to provide the results of counting the colony-forming units (CFU) of each inoculum. The acceptable range of inocula was 2–8 × 105 CFU/mL. For each test strain, the results of the three test series, including all MIC values and the results of CFU count, purity check and growth controls, had to be filled into a form along with additional comments about observations or problems encountered during the tests. The time frame for the entire ring trial was set at 65 days (starting in early October 2004 and ending in early December 2004). This time period started with the sending off of the test

J. Wallmann et al. / International Journal of Antimicrobial Agents 27 (2006) 482–490

material, included the testing period, sending back the forms with results filled in and evaluation of the results. It ended with communication of the results to the participating laboratories. Each laboratory received its results in a personalised form, but could also obtain the results of the other laboratories in an anonymised presentation. 2.3. Evaluation of the test results All data determined by the 46 laboratories were sent to a central institution for evaluation and documentation of the results. The forms on which the data were registered were coded to guarantee the anonymity of the participating laboratories. Successful participation in the ring trial required that at least 80% of all MIC data determined in a specific laboratory were within the expected range [19]. The expected range was defined as the modal MIC value ± 1 dilution step. The modal MIC value (= the most frequently determined MIC value) for each antimicrobial agent/bacterium combination was calculated upon completion of the ring trial on the basis of the distribution of the MIC values determined in this interlaboratory test. The use of the modal value should be preferred over the arithmetic mean value in the case of MIC values [7]. The measured MIC data represent discrete values. For eval-

485

uation of the data, all results were included in a database and were subsequently checked for consistency. If a test strain did not grow within one test series, the missing values were not included for the evaluation. In the case that one or two of the growth controls present on each microtitre plate were negative, the results obtained from this microtitre plate were withdrawn. The same was true for failure of the purity check as well as if the inoculum size was outside the accepted range.

3. Results 3.1. Species differentiation All 46 laboratories were able to at least perform Gramstaining of the test strains and thus predetermined the correct microtitre plates for the Gram-positive and Gram-negative test strains. Although no specific methods for species differentiation were recommended, the participating laboratories were asked to indicate on the data sheet which method they had used. Based on these data, 31 laboratories used standardised, internationally accepted biochemical identification systems whereas the remaining 15 referred to ‘home made’ identification schemes. Such laboratory-specific sys-

Table 2 Mode variations (log2 steps) for each laboratory with experience in minimum inhibitory concentration (MIC) diagnostics participating in the interlaboratory test

a MIC

modal value ± 1 MIC step indicated by the shaded area. of six and more dilution steps from the MIC modal value.

b Variations

486

J. Wallmann et al. / International Journal of Antimicrobial Agents 27 (2006) 482–490

Table 3 Mode variations (log2 steps) for each laboratory without experience in minimum inhibitory concentration (MIC) diagnostics participating in the interlaboratory test

a MIC

modal value ± 1 MIC step indicated by the shaded area. of six and more dilution steps from the MIC modal value.

b Variations

tems were mainly used for identification of P. multocida. In three cases each (laboratories with and without experience in broth microdilution), the laboratories failed to determine the correct bacterial species of P. multocida and S. uberis. This failure was based on the use of a test kit for human pathogens that could not identify veterinary pathogens properly. 3.2. Results from laboratories with or without experience in the use of broth microdilution Assuming correct performance of the tests, a total of 240 MIC values (16 antimicrobial agents × 5 test strains × 3 test series) was expected from each participating laboratory during the course of this ring trial. However, only 14 (77.8%) of the 18 laboratories experienced in broth microdilution

(Table 2) and only 15 (53.6%) of the 28 non-experienced laboratories (Table 3) provided 240 MIC values. The number of MIC values provided varied between 120 and 240 for the experienced laboratories and between 154 and 240 for the non-experienced laboratories. A detailed analysis of all MIC values revealed that the percentages of MIC values within the accepted ranges for the different antimicrobials (modal MIC ± 1 dilution step) varied from 68.8% to 99.2% (mean value 89.3%) among the laboratories with experience, and from 63.3% to 99.0% (mean value 86.7%) among those without experience. Evaluation of the MIC data in Tables 2 and 3 showed that 10 535 MIC data values were provided, which corresponds to 95.4% of the maximum possible number of MIC values (46 laboratories × 240 MIC values = 11 040). A total

J. Wallmann et al. / International Journal of Antimicrobial Agents 27 (2006) 482–490

487

Table 4 Mode variations (log2 steps) for each bacterial strain

MIC, minimum inhibitory concentration. a MIC modal value ± 1 MIC step indicated by the shaded area. b Variations of six and more dilution steps from the MIC modal value.

of 1298 MIC values (12.3%) were outside the accepted range. This included 443 MIC values (10.7%) of the 4123 values from laboratories with experience and 855 values (13.3%) of the 6412 values from laboratories without experience in broth microdilution. Another 505 MIC data (197 from experienced and 308 from non-experienced laboratories) were missing. A total of 35 (76.1%) of the 46 laboratories had ≥80% of MIC values within the accepted range and were thus considered to have successfully participated in this ring trial. Among the 11 (23.9%) laboratories with <80% correct data, 4 were laboratories with experience and 7 did not have experience in broth microdilution. 3.3. Differences in MIC data with regard to the test strains Determination of a maximum number of 2208 MIC values (46 laboratories × 16 antimicrobial agents × 3 test series) would have been possible for each of the five test strains included in this study. As shown in Table 4, the number of MIC data determined varied between 1999 (90.5%) for the S. agalactiae field strain and 2165 (98.1%) for the E. coli field strain. The largest number of MIC values outside the accepted range was detected for the S. uberis field strain (n = 429), followed by the S. agalactiae field strain (n = 347); the largest number of missing values was seen with the S. agalactiae field strain (n = 209) followed by S. uberis (n = 106) and P. multocida (n = 103) field strains. The percentages of acceptable MIC values determined for these field strains varied between 79.6% for S. uberis and 96.7% for E. coli. Of the 2164 MIC values determined for the reference strain S. aureus ATCC 29213, 156 (7.2%) were not within the accepted range, and in 44 cases no MIC data were provided. Since reference strains are commonly tested side by side with the test strains for quality control purposes, negative or missing results obtained from reference strains usually require repetition of the entire test series.

3.4. Differences in MIC data with regard to the antimicrobial agents tested Depending on the number of test strains and the microtitre plates used, the maximum possible number of MIC values determined for each of the antimicrobial agents tested in this study varied between 276 (apramycin, colistin, doxycycline, florfenicol, nalidixic acid, spectinomycin, tiamulin and trimethoprim present only on NLV34), 414 (avilamycin, erythromycin, lincomycin, oxacillin, quinupristin/dalfopristin and vancomycin present only on NLV21), 690 (ampicillin, amoxicillin/clavulanic acid, cefazolin, cefoperazone, cefquinome and ceftiofur present only on NLV31) and 1104 (penicillin G and cefalothin present on NLV21 and NLV31). As shown in Table 5, these hypothetical values were not reached in a single case. A total of 264 to 269 MIC values (95.7–97.5%) were obtained for the antimicrobial agents tested on plate NLV34, 388 to 395 values (93.7–95.4%) for those on NLV21, 649 to 660 (94.1–95.7%) for those on NLV31, and 1045 (94.7%) or 1054 (95.5%) for penicillin G and cefalothin, respectively. This in turn corresponded to 2.5–6.3% of missing values for the different antimicrobial agents. Based on the MIC values outside the acceptable range, the antimicrobial agents tested in this study could be subdivided into four categories: error ratios of 0% to <5% were determined for doxycycline, florfenicol and nalidixic acid, 5% to <10% for vancomycin, quinupristin/dalfopristin, trimethoprim, tiamulin, spectinomycin, amoxicillin/clavulanic acid, colistin, cefazolin, ampicillin and cefquinome, 10% to <20% for lincomycin, cefalothin, apramycin, erythromycin, oxacillin and avilamycin and 20% to <30% for cefoperazone, penicillin G and ceftiofur. A detailed analysis revealed that most of the 182 MIC values of ceftiofur (27.8% of values) that were outside the accepted range were lower than expected (133 values < accepted range > 49 values) and resulted from testing of S. agalactiae, S. uberis and P. multocida. Of the 217 divergent MIC values for penicillin G (20.8% of val-

488

J. Wallmann et al. / International Journal of Antimicrobial Agents 27 (2006) 482–490

Table 5 Mode variations (log2 steps) for each antimicrobial agent

MIC, minimum inhibitory concentration. a MIC modal value ± 1 MIC step indicated by the shaded area. b Variations of six and more dilution steps from the MIC modal value. c Antimicrobial agent used twice on the microtitre plates.

ues), most were higher than expected (56 values < accepted range > 161 values) and based mainly on the results obtained from the S. aureus and S. uberis test strains. The 136 divergent MIC values for cefoperazone (20.6% of values) were mainly from P. multocida and were more evenly distributed (51 values < accepted range > 85 values).

89.1% and then 89.9%, whilst the corresponding percentage for the non-experienced laboratories decreased from 87.6% to 86.8% then to 85.6%.

3.5. Differences in MIC data for three test series

A comparison of the MIC values determined by the participating laboratories in this interlaboratory test and the CLSI approved clinical veterinary breakpoints (as far as available) [20] showed that the results which differed distinctly from the mode MIC also resulted in a wrong classification, for example a susceptible strain as resistant (68 cases). This was particularly true when the mode MIC was close to or at the breakpoint for susceptibility, e.g. penicillin G, quinupristin/dalfopristin and vancomycin with Streptococcus spp. Misclassification of a resistant strain as susceptible was observed in seven cases (e.g. cefalothin and doxycycline with E. coli).

A total of 80 MIC values were expected to be determined by each laboratory during each of the three test series. This should result in 1440 values from the 18 experienced laboratories and 2240 values from the 28 non-experienced laboratories. Table 6 shows that 1371–1376 values (95.2–95.6%) and 2125–2158 values (94.9–96.3%) were obtained from the two groups of laboratories, respectively. The percentage of correct MIC values determined in the laboratories with experience increased slightly during the three test series, from 88.8% to

3.6. Relationship of the MIC data to the published CLSI clinical veterinary breakpoints

J. Wallmann et al. / International Journal of Antimicrobial Agents 27 (2006) 482–490

489

Table 6 Mode variations (log2 steps) for each test series of the interlaboratory test

MIC, minimum inhibitory concentration. a MIC modal value ± 1 MIC step indicated by the shaded area. b Variations of six and more dilution steps from the MIC modal value. c Laboratories with experience in MIC determination. d Laboratories without experience in MIC determination.

4. Discussion In principle, every antimicrobial agent can select for resistant bacteria and hence the choice of the most effective antimicrobial agent(s) in the control of bacterial infections is crucial. Determination of valid data for the in vitro susceptibility of bacteria is an essential requirement to support the clinician’s decision for or against a particular antimicrobial agent. However, this requires a methodology that allows determination of quantitative data in terms of MIC values. The broth microdilution method is considered the method of choice for most fast-growing aerobic bacterial pathogens for this purpose [21]. During evaluation of the results, it became obvious that there were no relevant differences between the data provided by experienced and non-experienced laboratories (Tables 2 and 3). This observation strongly suggests that the broth microdilution method can be learnt quickly and that reproducible results can be obtained within a short period of time. Hence, introduction of this method into a laboratory that formerly performed disk diffusion is in principle possible without problems regarding correct performance of the tests and evaluation of the data obtained. Several laboratories in Germany are currently in the process of changing the methodology of in vitro susceptibility testing and successfully use the SOP developed by the working group ‘Antimicrobial Resistance’. These laboratories also observe very good acceptance of their results, which are now communicated to their clients as an MIC value supplemented with the associated qualitative category ‘susceptible’, ‘intermediate’ or ‘resistant’. When looking at the additional comments indicated on the data forms, a considerable number of laboratories encountered problems in correct preparation of the inoculum as well as in reading the MIC values of the P. multocida, S. agalactiae and S. uberis test strains. These factors might play a role in the higher error ratios of 14.0%, 17.4% and 20.4%, respec-

tively, noted with these strains compared with those of 3.3% and 7.2% for E. coli and the S. aureus reference strain. A higher variability in the results obtained from Enterococcus spp. compared with E. coli and S. aureus strains was also detected in an interlaboratory test performed in five Scandinavian countries [22]. In the present interlaboratory test, the participating laboratories stated that preparation of the inoculum and the decision of growth versus no growth were unproblematic for the E. coli field strain as well as the S. aureus reference strain. These factors are believed to be of major importance for the high reproducibility of the results obtained for these two test strains. A high level of consistency in the results of MIC determination for S. aureus ATCC 29213 against five different antimicrobial agents utilised in veterinary medicine was also documented in an interlaboratory test performed at six university laboratories in the USA [7]. The results of this study showed that 21 (75.0%) of the 28 laboratories not experienced in broth microdilution and 14 (77.8%) of the 18 laboratories that already use this method routinely produced 81.8–99.2% of results within the accepted MIC range of the modal MIC ± 1 dilution step. A retrospective analysis of the 11 cases in which the required level of 80% accurate values was not achieved revealed individual modifications of the test protocol and the use of media other than those recommended as causes for the failure to produce acceptable results. Based on the overall positive results of this interlaboratory test and the lack of significant intralaboratory variations, which illustrates the uniformity of the test protocol, the working group ‘Antimicrobial Resistance’ will make this SOP available to laboratories interested in performing broth microdilution. This might be of particular relevance to laboratories seeking accreditation [23]. Such a harmonised SOP that has shown its fitness for use in a large ring trial and that is based on the internationally accepted CLSI document M31-A2 [8] will help to ensure the quality of in vitro susceptibility testing in routine diagnostics in veterinary medicine.

490

J. Wallmann et al. / International Journal of Antimicrobial Agents 27 (2006) 482–490

References [1] Schwarz S, Kehrenberg C, Walsh TR. Use of antimicrobial agents in veterinary medicine and food animal production. Int J Antimicrob Agents 2000;17:431–7. [2] Chaslus-Dancla E, Lafont J-P, Martel J-L. Spread of resistance from food animals to man: the French experience. Acta Vet Scand Suppl 2000;93:53–61. [3] Salisbury JG, Nicholis TJ, Lammerding AM, Turnidge J, Nunn MJ. A risk analysis framework for the long-term management of antibiotic resistance in food-producing animals. Int J Antimicrob Agents 2002;20:153–64. [4] Schwarz S, B¨ottner A, Hafez HM, et al. Antimicrobial susceptibility testing of bacteria isolated from animals: methods for in-vitro susceptibility testing and their suitability with regard to the generation of the most useful data for therapeutic applications. Berl M¨unch Tier¨arztl Wochenschr 2003;116:353–61 [in German]. [5] Wallmann J, Schr¨oter K, Wieler LH, Kroker R. National antibiotic resistance monitoring in veterinary pathogens from sick foodproducing animals: the German programme and results from the 2001 pilot study. Int J Antimicrob Agents 2003;22:420– 8. [6] Hindler JF, Swenson JM. Susceptibility test methods: fastidious bacteria. In: Murray PR, Baron EJ, Jorgensen JH, Pfaller MA, Yolken RH, editors. Manual of clinical microbiology. 8th ed. Washington, DC: American Society for Microbiology; 2003. p. 1128–40. [7] Odland BA, Erwin ME, Jones ER. Quality control guidelines for disk diffusion and broth microdilution antimicrobial susceptibility tests with seven drugs for veterinary applications. J Clin Microbiol 2000;38:453–5. [8] Clinical and Laboratory Standards Institute. Performance standards for antimicrobial disk and dilution susceptibility tests for bacteria isolated from animals, approved standard. 2nd ed. NCCLS document M31-A2. Wayne, PA: CLSI; 2002. [9] Andrews J.M. Determination of minimum inhibitory concentrations. J Antimicrob Chemother 2001;48(Suppl. 1):5–16. [Erratum: J Antimicrob Chemother 2002;49:1049]. [10] Courvalin P, Soussy CJ. Report of the Comit´e de l’Antibiogramme de la Soci´et´e Francaise de Microbiologie. Conditions techniques generales pour le methodes de dilution et de diffusion en milieu gelose. Clin Microbiol Infect 1996;2(Suppl. 1):1–49. [11] Medizinische mikrobiologie und immunologie: diagnostische Verfahren. Deutsches Institut f¨ur Normung e.V. (Hrsg.). 3. Auflage, DIN-Taschenb¨ucher; 2000. Band 222, p. 151–294.

[12] Kahlmeter G, Brown DFJ, Goldstein FW, et al. European harmonization of MIC breakpoints for antimicrobial susceptibility testing of bacteria. J Antimicrob Chemother 2003;52:145–8. [13] Jorgensen JH. Need for susceptibility testing guidelines for fastidious or less-frequently isolated bacteria. J Clin Microbiol 2004;42:493–6. [14] Luhofer G, B¨ottner A, Hafez HM, et al. Proposals of the working group ‘Antimicrobial resistance’ for the configuration of microtitre plates to be used in routine antimicrobial susceptibility testing of bacterial pathogens from infections of large food-producing animals and mastitis cases. Berl M¨unch Tier¨arztl Wochenschr 2004;117:245–51 [in German]. [15] Luhofer G, B¨ottner A, Hafez HM, et al. Layout proposal for in-vitro susceptibility testing of bacterial pathogens isolated from companion animals. In: Proceedings of the 26th Congress of the German Veterinary Medical Society (DVG), 2005. p. 158. [16] Horwitz W. Protocol for the design, conduct and interpretation of method-performance studies. Pure Appl Chem 1995;67:331–4. [17] Pfeifer T, editor. Design review. In: Qualit¨atsmanagement, Strategien, Methoden, Techniken. 2. Auflage, M¨unchen, Wien: Carl Hanser Verlag; 1996. [18] Thrusfield M, editor. Veterinary epidemiology. 2nd ed. Oxford: Blackwell Science Ltd.; 1995. [19] Stock I, Machka K, Rodloff A, Wiedemann B. Qualit¨atssicherung und Qualit¨atskontrollen in der AntibiotikaEmpfindlichkeitsbestimmung von Bakterien mit der Mikrodilution. Chemother J 2001;10:78–98. [20] Clinical and Laboratory Standards Institute. Performance standards for antimicrobial disk and dilution susceptibility tests for bacteria isolated from animals, informational supplement, NCCLS document M31-S1. Wayne, PA: CLSI; 2004. [21] Chaitram JM, Jevitt LA, Lary S, Tenover FC; WHO Antimicrobial Resistance Group. The World Health Organization’s External Quality Assurance System Proficiency Testing Program has improved the accuracy of antimicrobial susceptibility testing and reporting among participating laboratories using NCCLS method. J Clin Microbiol 2003;41:2372–7. [22] Petersen A, Aarestrup FM, Hofshagen M, Sipila H, Franklin A, Gunnarsson E. Harmonization of antimicrobial susceptibility testing among veterinary diagnostic laboratories in the five Nordic countries. Microb Drug Resist 2003;9:381–8. [23] Middlebrook K, Morris A. The effect of proficiency testing participation on laboratory performance. Draft Minutes of the 15th Annual General Meeting of the Canadian Accociation for Enviromental Analytical Laboratories; 2004 June 18; Vancouver, Canada.