Ability of laboratories to detect emerging antimicrobial resistance in nosocomial pathogens: a survey of Project ICARE laboratories

Diagnostic Microbiology and Infectious Disease 38 (2000) 59 67 Surveillance www.elsevier.com/locate/diagmicrobio Ability of laboratories to detect emerging antimicrobial resistance in nosocomial pathogens: a survey of Project ICARE laboratories Christine D. Steward a, *, David Wallace b, Susannah K. Hubert b, Rachel Lawton a, Scott K. Fridkin a, Robert P. Gaynes a, John E. McGowan, Jr b, Fred C. Tenover a a Hospital Infections Program, Centers for Disease Control and Prevention, 1600 Clifton Rd, NE (G08) Atlanta, GA 30333, USA b Rollins School of Public Health, Emory University, Atlanta, GA 30322, USA Abstract A proficiency testing project was conducted among 48 microbiology laboratories participating in Project ICARE (Intensive Care Antimicrobial Resistance Epidemiology). All laboratories correctly identified the Staphylococcus aureus challenge strain as oxacillinresistant and an Enterococcus faecium strain as vancomycin-resistant. Thirty-one (97%) of 32 laboratories correctly reported the Streptococcus pneumoniae strain as erythromycin-resistant. All laboratories testing the Pseudomonas aeruginosa strain against ciprofloxacin or ofloxacin correctly reported the organism as resistant. Of 40 laboratories, 30 (75%) correctly reported resistant MICs or zone sizes for the imipenem- and meropenem-resistant Serratia marcescens. For the extended-spectrum -lactamase (ESBL)-producing strain of Klebsiella pneumoniae, 18 (42%) of 43 laboratories testing ceftazidime correctly reported ceftazidime MICs in the resistant range. These results suggest that current testing generally produces accurate results, although some laboratories have difficulty detecting resistance to carbapenems and extended-spectrum cephalosporins. This highlights the need for monitoring how well susceptibility test systems in clinical laboratories detect emerging resistance. 2000 Elsevier Science Inc. All rights reserved. 1. Introduction Validation of the ability of clinical laboratories to detect both established and novel resistance patterns is critical for the success of any project involving surveillance of emerging antimicrobial resistance. Proficiency testing is an important part of quality assurance and validates the accuracy of antimicrobial susceptibility data reported by microbiology laboratories (Doern et al., 1999; Rosenberg et al., 1997; Tenover et al., 1999). Validation of susceptibility testing results is particularly important in multicenter studies where multiple techniques are used for testing. When proficiency testing indicates that laboratories within the system or network are achieving comparable susceptibility test results, comparisons of resistance rates among institutions become more meaningful. In addition, each laboratory then can compare its results with aggregated results from all other participating laboratories (benchmarking) (Jones et al., 1997a; Pfaller et al., 1999; Steward et al., 1998). Here we present the results from a proficiency testing project undertaken in hospitals participating in Project ICARE (Intensive Care Antimicrobial Resistance Epidemiology). Project ICARE conducts laboratory-based surveillance for antimicrobial resistance and antimicrobial use at U.S. hospitals participating in the National Nosocomial Infections Surveillance System (NNIS) (Archibald et al., 1997; Fridkin et al., 1999). Proficiency testing was one of the tools employed to validate monthly, aggregated antimicrobial susceptibility testing results sent to the Project ICARE master dataset by the hospitals. 2. Materials and methods 2.1. ICARE hospitals * Corresponding author. Tel.: 1-404-639-2825; fax: 1-404-639-1381. E-mail address: cks7@cdc.gov (C.D. Steward). Current address. Bacterial and Mycotic Diseases (D63). Centers for Disease Control and Prevention. 1600 Clifton Road, NE. (G08), Atlanta, GA 30333 In March 1998, 48 microbiology laboratories within the United States participated in the Project ICARE proficiency testing program. The laboratories were located in 26 states in the following regions: Northeast (7 states, 18 laborato- 0732-8893/00/$ see front matter 2000 Elsevier Science Inc. All rights reserved. PII: S0732-8893(00)00161-9

60 C.D. Steward et al. / Diagnostic Microbiology and Infectious Disease 38 (2000) 59 67 ries), South (10 states, 18 laboratories), Midwest (6 states, 8 laboratories), and West (3 states, 4 laboratories). 2.2. Proficiency testing organisms The Hospital Infections Program at the Centers for Disease Control and Prevention (CDC) sent the following six challenge isolates to participating laboratories: a homogeneous oxacillin-resistant Staphylococcus aureus, a vancomycin-resistant Enterococcus faecium (NJ1) (Tenover et al., 1993), an erythromycin-resistant Streptococcus pneumoniae with reduced susceptibility to penicillin, a ciprofloxacinand ofloxacin-resistant Pseudomonas aeruginosa, an imipenem- and meropenem-resistant Serratia marcescens (Yigit et al., 1999), and Klebsiella pneumoniae ATCC 700603, which is an extended-spectrum -lactamase (ESBL) producer. Each organism was selected from the CDC culture collection based on its resistance phenotype and genotype. The genotypes of the S. aureus (meca), E. faecium (vana), and K. pneumoniae (bla SHV-18 ) were confirmed by using specific polymerase chain reaction (PCR) assays (Clark et al., 1993; Mabilat & Goussard, 1993; Rasheed et al., 1997; Vannuffel et al., 1995). The P. aeruginosa strain was shown to maintain ciprofloxacin resistance after ten passages on nonselective trypticase soy agar plates containing 5% sheep blood (blood agar plates) (BD 1, Sparks, MD). The S. marcescens contained a bla SME-1 -like carbapenemase and outer membrane porin changes, producing high-level resistance to imipenem and meropenem (Yigit et al., 1999). The S. marcescens continued to be imipenem-resistant after eight passages on non-selective blood agar plates. 2.3. Isolate preparation 1 Use of trade names is for identification purposes only and does not constitute endorsement by the Public Health Service or the U.S. Department of Health and Human Services In preparation for shipment, the isolates were transferred to blood agar plates from storage in defibrinated sheep blood (Lampire Biologic Laboratories, Pipersville, PA) at 70 C. After a 24 h incubation, the plates were inspected for purity, and colonies were inoculated into 49 sets of nutrient agar stabs (one set for each ICARE laboratory and the CDC laboratory). The S. pneumoniae challenge strain was inoculated onto slants containing trypticase soy agar with 5% sheep blood (BD). The stabs and slants were incubated at 35 C overnight. After incubation, the stabs and slants were examined for growth and one set of organisms was sent to each of the 48 ICARE laboratories via overnight delivery. The CDC set was left at room temperature overnight. The proficiency testing protocol required that each organism be transferred to a blood agar plate and incubated at 35 C overnight upon arrival at the ICARE laboratories. One additional transfer was required before testing was performed. In the same manner, the organisms were transferred to blood agar plates twice in the CDC laboratory before testing. 2.4. Isolate testing Laboratories performed routine antimicrobial susceptibility testing on all organisms and routine identification to species level on the E. faecium and K. pneumoniae (identified in the instructions to the laboratory directors only as Enterococcus species and Gram-negative bacillus). Susceptibility test systems used by the ICARE laboratories included Vitek (biomérieux, Inc., Hazelwood, MO), MicroScan (Dade Behring, Inc., West Sacramento, CA), disk diffusion, Sensititre (Trek Diagnostics, Westlake, OH), and Etest (AB BIODISK North America, Inc., Piscataway, NJ). The oxacillin disk test was used by some hospitals to screen the S. pneumoniae strain for potential penicillin resistance (National Committee for Clinical Laboratory Standards, 1997a; National Committee for Clinical Laboratory Standards, 1998). As the study referee, CDC tested all isolates using broth microdilution plates made in-house, according to National Committee for Clinical Laboratory Standards (NCCLS) guidelines (National Committee for Clinical Laboratory Standards, 1997b). Quality control for broth microdilution was performed using the following organisms: S. aureus ATCC 29213, S. aureus ATCC 43300, S. aureus ATCC 43387, E. coli ATCC 25922, P. aeruginosa ATCC 27853, E. faecalis ATCC 29212, E. faecalis ATCC 51299, and S. pneumoniae ATCC 49619. For identification of the E. faecium to species level, laboratories used Vitek cards, MicroScan panels, API 20 Strep strips (biomérieux, Inc.), Minitek disks (BD), and RapID STR (Innova Diagnostics, San Diego, CA). For identification of the K. pneumoniae to species level, laboratories used Vitek cards, MicroScan panels, Sensititre panels, and API 20E strips (biomérieux, Inc.). The CDC laboratory used reference biochemical tests to identify each isolate to species level (Facklam et al., 1999; Farmer et al., 1980; Farmer, 1999). The quantitative results (minimum inhibitory concentration [MIC] and disk diffusion values) and the categorical interpretations (susceptible, intermediate, and resistant) were transcribed by each laboratory to worksheets supplied by the Project ICARE laboratory. The worksheets, which contained a list of preprinted antimicrobial agent names and spaces for additional agents, also included spaces for test date, test method, identification, and identification method, where appropriate. On the worksheet for the S. pneumoniae challenge strain, space was provided only for MICs of penicillin, cefotaxime, and ceftriaxone. At the bottom of each worksheet, the laboratory was invited to list any comments about the susceptibility profile that would be reported to the patient s chart. The worksheets from each laboratory were sent to the Project ICARE laboratory for analysis.

C.D. Steward et al. / Diagnostic Microbiology and Infectious Disease 38 (2000) 59 67 61 2.5. Data analysis Results for each specific antimicrobial agent were analyzed separately. Testing errors were determined by comparing CDC broth microdilution MICs and categorical interpretations based on NCCLS criteria (National Committee for Clinical Laboratory Standards, 1998) with the results from the participating laboratories. Acceptable results were MICs within 1 dilution from the reference result or test interpretations (e.g., susceptible, intermediate, or resistant) resulting in no error or a minor error when compared to the reference interpretation. Minor errors occurred when the test result or reference result was intermediate and the other result was susceptible or resistant. MIC results outside 1 dilution, major errors (test result resistant and reference result susceptible), and very major errors (test result susceptible and reference result resistant) were considered incorrect results. Disk diffusion errors were based on categorical interpretations. For the antimicrobial agents studied, no categorical interpretation changes occurred in NCCLS guidelines between 1998 and 1999 (National Committee for Clinical Laboratory Standards, 1998; National Committee for Clinical Laboratory Standards, 1999). 2.6. Supplemental information In February 1998, before the initiation of the proficiency testing program, all laboratories filled out a survey describing their microbiology methods. This survey included detailed information on the antimicrobial susceptibility methods used to test bacteria. If the susceptibility test method was automated, laboratories recorded the specific panel or card types used for testing. The panel and card type information was extracted and used as supplemental information in the proficiency testing analysis. In March 1998, along with testing of the isolates, each laboratory was asked to record the last ten MIC or disk diffusion zone sizes measured for the control strain, E. coli ATCC 25922. These quality assurance data were used to supplement the proficiency testing analysis, specifically for imipenem and ceftazidime. In a follow-up study conducted in February 1999, a subset of Vitek users were sent the imipenem- and meropenem-resistant S. marcescens strain that was previously sent for proficiency testing in March 1998. The five Vitek users who participated had reported the strain as imipenemsusceptible, -intermediate, or -resistant in the proficiency testing. In the follow-up study, the isolate was tested in the same manner as described for the previous proficiency testing challenge. Results of the 1999 follow-up study were compared with those from the 1998 proficiency testing report. 3. Results Of the 48 laboratories participating in the Project ICARE proficiency testing program, 21 (44%) reported correct results for all 27 organism-antimicrobial agent combinations analyzed. Of the 27 laboratories producing errors, 20 (74%) produced only one error, five (18%) produced two errors, one (4%) produced three errors, and one laboratory (4%) produced four errors among the 27 organism-antimicrobial agent combinations analyzed. Detailed analysis of the errors is shown below. 3.1. Oxacillin/methicillin-resistant S. aureus (MRSA) All 48 laboratories correctly reported the isolate as oxacillin- and penicillin-resistant. Oxacillin and penicillin reference MICs were 16 g/ml and 2 g/ml, respectively. Of the 45 laboratories that tested the isolate against erythromycin, all 45 correctly reported the organism as erythromycin-resistant (reference MIC 8 g/ml). Most laboratories used Vitek cards (26 laboratories) or MicroScan conventional panels (16 laboratories) for susceptibility testing. Other methods included disk diffusion (three laboratories), MicroScan Auto or Touchscan panels (two laboratories), or MicroScan rapid panels (one laboratory). 3.2. Vancomycin-resistant E. faecium Forty-three (90%) of 48 laboratories correctly identified the enterococcal challenge organism as an E. faecium. One laboratory incorrectly reported the organism as an E. durans. Four laboratories (two Vitek users and two disk diffusion users) reported that they do not identify enterococci to species level. Of the 44 laboratories that identified the organism to species level, 34 used the same system for identification and vancomycin susceptibility testing (21 Vitek users and 13 MicroScan users). The other 10 laboratories used different methods for identification, including Sensititre, Minitek, RapID STR, and API 20 STREP. All laboratories correctly reported the organism as vancomycin-resistant (reference MIC 512-1024 g/ml) (Table 1). Forty-five (98%) of 46 laboratories correctly reported ampicillin-susceptible MICs or zone sizes (reference MIC 4 8 g/ml). One Vitek user incorrectly reported a resistant ampicillin MIC of 16 g/ml (Table 1). One MicroScan user correctly reported an ampicillin MIC of 4 g/ml (susceptible), but incorrectly interpreted the MIC as intermediate. Thirty-nine (98%) of 40 laboratories correctly reported the organism as penicillin-resistant (reference MIC 32 g/ ml). One Vitek user incorrectly reported a susceptible penicillin MIC of 8 g/ml (Table 1).

62 C.D. Steward et al. / Diagnostic Microbiology and Infectious Disease 38 (2000) 59 67 Table 1 E. faecium: Results by Test Method Antimicrobial Agent (n a ) Laboratory Test Method MicroScan b Vitek b Disk Rapid Conventional Auto or diffusion b Touchscan Etest b Vancomycin (48) key result - 16/16 2/2 23/23 4/4 3/3 Ampicillin (46) - 16/16 1/1 23/24 3/3 2/2 Penicillin (40) 1/1 14/14 2/2 20/21 1/1 1/1 b Number of correct MICs or disk diffusion results/total number of laboratories reporting a result for the antimicrobial agent. 3.3. Erythromycin-resistant S. pneumoniae with reduced susceptibility to penicillin Thirty-one (97%) of 32 laboratories testing erythromycin correctly reported the pneumococcal challenge strain as resistant (Table 2). One laboratory incorrectly reported an erythromycin zone size of 26 mm (susceptible). The other 18 disk diffusion users correctly reported a resistant erythromycin zone size. The S. pneumoniae produced a reference oxacillin disk test zone size of 15 mm, indicating that penicillin MICs should be performed to confirm penicillin resistance. Penicillin reference MICs for this organism were 0.03 to 0.12 g/ml (susceptible to intermediate). [According to NCCLS criteria, penicillin MICs of 0.06 are susceptible, and MICs of 0.12 1 g/ml are intermediate (National Committee for Clinical Laboratory Standards 1998).] In this survey, 24 laboratories performed an oxacillin disk test and a penicillin MIC, 18 laboratories performed a penicillin MIC only, four laboratories performed an oxacillin disk test and noted that a referral laboratory would perform an MIC, and two laboratories performed the oxacillin disk test and did not report a subsequent penicillin result. All 30 laboratories performing an oxacillin disk test correctly reported oxacillin disk screen results of 19 mm (range: 6 19 mm). No laboratories reported testing penicillin by disk diffusion. Forty-two laboratories (35 Etest users and seven MicroScan MicroStrep panel users) reported penicillin MICs. Forty laboratories reported penicillin MICs of 0.016 0.125 g/ml. Two laboratories, both Etest users, reported intermediate MICs of 0.64 g/ml, an MIC not found on the penicillin Etest strip. Four laboratories reported penicillin Etest results of 0.064 g/ml and interpreted the MIC incorrectly as intermediate. One laboratory reported an MIC of 0.094 g/ml and interpreted the MIC incorrectly as susceptible. All laboratories reporting cefotaxime and/or ceftriaxone results used MicroScan MicroStrep panels (seven laboratories) or Etest (17 laboratories) and correctly reported those drugs as susceptible. No laboratories reported disk diffusion testing of cephalosporins. 3.4. Ciprofloxacin- and ofloxacin-resistant P. aeruginosa Forty-seven laboratories correctly reported the P. aeruginosa isolate as resistant to ciprofloxacin (46 laboratories; reference MIC 8 g/ml), ofloxacin (22 laboratories; reference MIC 8 g/ml), or both drugs. One laboratory did not report results for any fluoroquinolones. For other antimicrobial agents, all laboratories correctly reported the organism as ceftazidime-susceptible (reference MIC 2 g/ ml), and laboratories reporting either imipenem (43 laboratories), meropenem (two laboratories), or both drugs correctly reported the organism as carbapenem-susceptible. The imipenem and meropenem reference MICs were 1 2 g/ml and 0.25 g/ml, respectively. The most commonly used susceptibility testing methods for the P. aeruginosa were Vitek cards and MicroScan Table 2 S. pneumoniae: Test Results Antimicrobial Agent (n a ) Reference Laboratory Broth Microdilution MIC Results b Pooled Results by Category c Susceptible Intermediate Resistant Erythromycin (32) key result 8 g/ml (R) 1-31 Penicillin (42) 0.03 0.12 g/ml (S or I) 33 9 - Cefotaxime (24) 0.12 0.25 g/ml (S) 24 - - Ceftriaxone (20) 0.25 0.5 g/ml (S) 20 - - b R, resistant; I, intermediate; S, susceptible. c Based on reported MIC or zone size.

C.D. Steward et al. / Diagnostic Microbiology and Infectious Disease 38 (2000) 59 67 63 Table 3 P. aeruginosa: Results by Test Method Antimicrobial Agent (n a ) Laboratory Test Method MicroScan b Vitek b Disk Rapid Conventional Auto or Touchscan Diffusion b Ciprofloxacin (46) - 16/16 2/2 22/22 6/6 key result Ofloxacin (22) - 10/10-9/9 3/3 key result Ceftazidime (48) - 16/16 2/2 24/24 6/6 Imipenem (43) - 15/15 2/2 21/21 5/5 Meropenem (2) - 1/1 - - 1/1 b Number of correct MICs or disk diffusion results/total number of laboratories reporting a result for the antimicrobial agent. conventional panels (Table 3). No laboratories used MicroScan rapid panels for P. aeruginosa testing. 3.5. Imipenem- and meropenem-resistant S. marcescens Thirty-nine of 48 laboratories reported results for imipenem, and one laboratory reported results for both imipenem and meropenem for the S. marcescens challenge strain. Of the 40 laboratories reporting imipenem results, 30 (75%) correctly reported an imipenem MIC of 16 g/ml (resistant) (Table 4). One laboratory reported the organism correctly as imipenem resistant (MIC 8 g/ml) but incorrectly as meropenem susceptible (MIC 4 g/ml). To test imipenem, 21 laboratories used Vitek cards, 14 used MicroScan conventional panels, two used MicroScan Auto or Touchscan panels, two used disk diffusion, and one used MicroScan rapid panels. Of the six laboratories incorrectly reporting imipenem-susceptible MICs or zone sizes, one used disk diffusion (zone size reported as 16 mm) and five used Vitek (MICs reported as 4 g/ml). Imipenemintermediate MICs were reported by three Vitek users and one MicroScan rapid panel user. Of the eight Vitek users incorrectly reporting imipenem-susceptible or -intermediate MICs, six different Vitek cards were used for testing, and five of the six card types were the same as card types used in laboratories that correctly reported resistant imipenem MICs for this organism. Eight of the ten laboratories incorrectly reporting imipenem-susceptible or -intermediate MICs or zone sizes had quality control results for E. coli ATCC 25922 that were within the acceptable range for imipenem (MICs, 0.06 0.25 g/ml, and disk diffusion zone sizes, 26 32 mm) (National Committee for Clinical Laboratory Standards 1998). Two of the ten laboratories did not report imipenem quality control data. When the February 1999 test results of the S. marcescens challenge strain were compared to the original proficiency test results, three of the five Vitek users reported the same results both times. In both studies, two of the five Vitek users correctly reported an imipenem MIC of 16 g/ml (resistant), and one incorrectly reported an imipenem MIC of 8 g/ml (intermediate). The other two laboratories reported results of 4 g/ml (susceptible) to 8 g/ml in the 1998 proficiency testing study and 8 g/ml to 16 g/ml in the 1999 follow-up study. For ceftazidime, 37 (90%) of 41 laboratories correctly reported the organism as susceptible (Table 4). All four laboratories reporting incorrect (intermediate or resistant) results were MicroScan users. One laboratory reported a ceftazidime MIC of 16 g/ml (intermediate), and three other laboratories reported ceftazidime MICs of 16 g/ml (resistant). The four MicroScan users incorrectly reporting ceftazidime-intermediate or -resistant MICs used four different MicroScan panels for testing. All four panel types were the same as panel types used in laboratories correctly reporting the organism as ceftazidime-susceptible. All laboratories correctly reported ciprofloxacin, ofloxacin, or both as susceptible (Table 4). 3.6. ESBL-producing K. pneumoniae ATCC 700603 All 48 laboratories correctly identified this organism as K. pneumoniae. Forty-four laboratories used the same system for organism identification and susceptibility testing (25 Vitek; 19 MicroScan). Four laboratories reported disk diffusion results for at least one drug. Two of these laboratories Table 4 S. marcescens: Test Results Antimicrobial agent (n a ) Reference Laboratory Broth Microdilution MIC Results b Pooled Results by Category c Susceptible Intermediate Resistant Imipenem (40) key result 32 g/ml (R) 6 4 30 Meropenem (1) key result 16 32 g/ml (R) 1 - - Ceftazidime (41) 2 g/ml (S) 37 1 3 Ciprofloxacin (46) 0.06 0.25 g/ml (S) 46 - - Ofloxacin (27) 0.25 0.5 g/ml (S) 27 - - b R, resistant; I, intermediate; S, susceptible. c Based on reported MIC or zone size.

64 C.D. Steward et al. / Diagnostic Microbiology and Infectious Disease 38 (2000) 59 67 Table 5 K. pneumoniae: Test Results Antimicrobial Agent (n a ) Reference Laboratory Broth Microdilution MIC Results b Pooled Results by Category c Susceptible Intermediate Resistant Extended-spectrum cephalosporin Ceftazidime (43) 32 64 g/ml (R-R) 8 17 18 Cefotaxime (32) 4 16 g/ml (S-I) 27 3 2 Ceftriaxone (37) 8 16 g/ml (S-I) 18 17 2 Monobactam Aztreonam (16) 64 128 g/ml (R-R) - - 16 Cephamycin Cefotetan (16) 0.5 1 g/ml (S) 16 - - Cefoxitin (20) 32 32 g/ml (R) 5 7 8 Cephalosporin Cefazolin or Cephalothin (48) 32 g/ml (R) - 1 47 b R, resistant; I, intermediate; S, susceptible. c Based on reported MIC or zone size. identified organisms using an API 20E strip, another used a Sensititre panel, and the other used a Vitek card. The laboratories reported a wide spectrum of MICs and zone sizes for testing of cephalosporins and cefoxitin against this organism (Table 5). Forty-three (90%) of the 48 laboratories tested ceftazidime (23 Vitek users, 15 MicroScan conventional panel users, two MicroScan Auto or Touchscan users, two disk diffusion users, and one MicroScan rapid panel user). Eight laboratories, all Vitek users, reported incorrect MICs of 8 g/ml. However, two of the eight laboratories identified the organism as an ESBLproducer and changed the ceftazidime results to resistant, as recommended by NCCLS standard M100-S9 (National Committee for Clinical Laboratory Standards, 1999). Seven of the eight laboratories used Vitek card types that also were used by other laboratories that reported intermediate ceftazidime MICs. The ceftazidime quality control data for E. coli ATCC 25922 for seven of the eight laboratories were within the acceptable range for ceftazidime (MICs, 0.06 0.5 g/ml) (National Committee for Clinical Laboratory Standards 1998). One laboratory did not report ceftazidime quality control data. Results for several other cephems were reported incorrectly. Three laboratories, using three different MicroScan conventional panels, incorrectly reported MICs of 32 g/ml for cefotaxime or ceftriaxone or both, while 12 MicroScan conventional panel users correctly reported susceptible or intermediate MICs. Two of the three panel types used by laboratories incorrectly reporting cefotaxime or ceftriaxone MICs were also used by laboratories correctly reporting susceptible or intermediate MICs. For cefoxitin, five laboratories, all Vitek users, reported an incorrect susceptible MIC of 8 g/ml, while eight Vitek users correctly reported intermediate or resistant cefoxitin MICs. Two of the three card types used by laboratories incorrectly reporting susceptible cefoxitin MICs were used by laboratories correctly reporting intermediate or resistant cefoxitin MICs. For cefazolin, one laboratory, a Vitek user, reported an incorrect intermediate MIC of 16 g/ml. All laboratories reported results for at least one extended-spectrum cephalosporin (ceftazidime, cefotaxime, or ceftriaxone). For the three extended-spectrum cephalosporins and aztreonam, 12 laboratories reported results for all four drugs, 25 reported results for three drugs, nine reported results for two drugs, and two reported results for a single drug (i.e., cefotaxime). By reported categorical interpretation, 35%, 47%, and 35% of Vitek users and 100%, 33%, and 29% of MicroScan conventional panel users reported the K. pneumoniae resistant to ceftazidime, cefotaxime, and ceftriaxone, respectively. Nine laboratories reported the organism as an ESBLproducer but did not change extended-spectrum cephalosporin and aztreonam MIC or zone size interpretations to resistant as suggested by NCCLS for ESBL-producing organisms (National Committee for Clinical Laboratory Standards, 1998). Eleven laboratories reported at least one susceptible or intermediate MIC and changed the extendedspectrum cephalosporin and aztreonam interpretation to resistant. Seven of the 11 laboratories reported the organism as an ESBL-producer. 4. Discussion The ability of clinical laboratories to detect antimicrobial resistance is critical in this era of emerging infectious diseases. This study suggests that current testing in clinical microbiology laboratories generally produces accurate results, although some resistant phenotypes are difficult to detect. Twenty-one laboratories correctly reported all 27 organism-antimicrobial agent combinations analyzed, and twenty laboratories missed only one result. Of the key results, the Project ICARE laboratories had little trouble

C.D. Steward et al. / Diagnostic Microbiology and Infectious Disease 38 (2000) 59 67 65 detecting homogeneous oxacillin resistance in S. aureus (which several other laboratories had failed to detect as oxacillin resistant in previous proficiency testing challenges), high-level vancomycin resistance in E. faecium, and high-level fluoroquinolone resistance in P. aeruginosa. Errors occurred predominantly in carbapenem and extendedspectrum cephalosporin testing. In this study, ten laboratories had difficulty detecting high-level imipenem resistance in the S. marcescens isolate. The results for imipenem from the eight laboratories reporting quality control data were within the acceptable range. While it is known that imipenem testing is difficult due to a variety of factors, most studies have focused on false imipenem resistance due to imipenem degradation in the test panels (Carmeli et al., 1998; Grist, 1992; O Rourke et al., 1991; White et al., 1991) and problems with automated instrument susceptibility test interpretations (Biedenbach & Jones, 1995; Jones et al., 1997b). In our study, the Gramnegative bacilli that were susceptible to imipenem, P. aeruginosa and K. pneumoniae, were tested as susceptible by all laboratories reporting imipenem for those organisms. One possible reason for under-reporting of imipenem resistance in the challenge strain of S. marcescens was that a particular test method could not accurately detect carbapenem resistance. Ten laboratories using three different methods incorrectly reported the organism as susceptible or intermediate to imipenem. However, each method was also used by at least one other laboratory that reported the organism as imipenem-resistant. Even though eight of the ten laboratories reporting incorrect results were Vitek users, no one particular card type was associated with underreporting of imipenem. Therefore, one method or card type was not likely the cause of the under-reporting. To further investigate under-reporting, the isolate of S. marcescens was sent again to five Vitek users in February 1999. Reasons for the changes in MICs between the 1998 proficiency testing and 1999 follow-up study are unclear but may be related to heightened awareness of the problems of imipenem testing. Project ICARE hospitals were informed about potential imipenem testing problems through proficiency testing feedback and validation study results (Steward et al., 1998). Another reason for some of the under-reporting of imipenem resistance in the S. marcescens strain could be due to the isolate itself. Carbapenem resistance in this isolate is due to a combination of carbapenemase production and porin changes. Thus, subpopulations that have either lost the bla SME-1 -like carbapenemase or have reduced expression of the enzyme could produce carbapenem-intermediate results. Furthermore, the porin changes could revert, causing the organism to test as carbapenem-susceptible if the carbapenemase was also lost. Ceftazidime errors included reports of both false resistance and false susceptibility. Four MicroScan users incorrectly reported ceftazidime intermediate or resistant MICs for the S. marcescens, while eight Vitek users incorrectly reported the K. pneumoniae as ceftazidime-susceptible. Reasons for the inaccurate reporting of ceftazidime results are unclear. As with imipenem, the quality control results for each hospital were reviewed and all reported results were within range. The cards and panels that were used in one laboratory providing incorrect results were used in other laboratories that provided correct results. However, four of the six Vitek users that incorrectly reported the S. marcescens as imipenem-susceptible also incorrectly reported the K. pneumoniae as ceftazidime- susceptible or cefoxitinsusceptible, which suggests that these cards may have been under-inoculated. The K. pneumoniae isolate sent for proficiency testing was an ESBL-producer. In 1998, at the time of this proficiency testing, NCCLS recommended screening for ESBL production in isolates of K. pneumoniae, K. oxytoca, and E. coli using unique breakpoints of 2 g/ml for cefpodoxime, ceftazidime, or aztreonam (National Committee for Clinical Laboratory Standards, 1998). In addition, these guidelines suggested that cephalosporins and aztreonam should be reported as resistant by stating that strains of Klebsiella spp. and E. coli may be clinically resistant to cephalosporin and aztreonam therapy by virtue of extendedspectrum production, despite apparent in vitro susceptibility to some of these agents (National Committee for Clinical Laboratory Standards, 1998). The 1999 NCCLS guidelines provide more detailed information about screening and confirming ESBL-producing organisms, including the recommendation to use more than one agent for screening (National Committee for Clinical Laboratory Standards, 1999). In addition, the new guidelines recommend that all ESBLproducing strains be reported as resistant for all penicillins, cephalosporins, and aztreonam, despite in vitro test results. From the proficiency testing results, it is clear that confusion regarding ESBL identification and reporting was not uncommon at the time of this project. A wide variety of combinations of extended-spectrum cephalosporins and aztreonam was tested. Only 11 of 48 laboratories changed the extended-spectrum cephalosporin and aztreonam MIC or zone size interpretations to resistant as suggested by NCCLS for ESBL-producing organisms (National Committee for Clinical Laboratory Standards, 1998). Of these 11, seven reported the organism as an ESBL-producer. Nine other laboratories also reported the strain as an ESBL-producer even though they did not modify all extended-spectrum cephalosporin and aztreonam results to be resistant. The remaining 28 laboratories did not report the organism as an ESBL-producer and did not modify drug interpretations to reflect ESBL status. More specific NCCLS guidelines on ESBL detection have been published since this proficiency testing program was performed (National Committee for Clinical Laboratory Standards, 1999). Problems associated with testing VRE were first reported in 1993 (Tenover et al., 1993), and difficulties still are apparent. One laboratory identified the E. faecium as an E. durans and reported a vancomycin disk diffusion zone size

66 C.D. Steward et al. / Diagnostic Microbiology and Infectious Disease 38 (2000) 59 67 of 6 mm (resistant). Although vancomycin-resistant E. durans have been reported (Cercenado et al., 1996), this identification is uncommon and has been confused with E. faecium (Singer et al., 1996). Thus, potential E. durans should be confirmed biochemically (Facklam et al., 1999). Most clinical enterococcal isolates that are vancomycinresistant are E. faecium or E. faecalis (Clark et al., 1993). In our study, one laboratory misinterpreted an E. faecium ampicillin MIC result of 4 g/ml, which is susceptible, as intermediate; however, there is no intermediate breakpoint for ampicillin for enterococci (National Committee for Clinical Laboratory Standards, 1998). For the S. pneumoniae challenge strain, two laboratories reported an oxacillin disk test result of 19 mm but failed to report a penicillin MIC result. According to NCCLS guidelines, organisms producing zone sizes of 19 mm should be further tested using a penicillin MIC method to confirm resistance (National Committee for Clinical Laboratory Standards, 1998). Another laboratory misinterpreted an Etest penicillin MIC result of 0.094 g/ml (which should be rounded up to 0.1 g/ml) as susceptible, and four laboratories misinterpreted an MIC of 0.064 g/ml (susceptible) as intermediate. The misinterpretations may have stemmed from rounding errors. The value of 0.094 g/ml may have been rounded down to meet the susceptible breakpoint of 0.06 g/ml, while the value of 0.064 g/ml most likely was rounded up to meet the intermediate breakpoints of 0.12 1 g/ml. MIC values should only be rounded up for Etest values that fall between two-fold dilutions (e.g., a result of 3 g/ml is rounded to 4 g/ml) (AB BIODISK North America Inc., 1997). In addition, NCCLS uses only 2 decimal places for MIC values and thus 0.064 g/ml should be considered susceptible. In general, Project ICARE laboratories performed well. The susceptibility testing errors that occurred with imipenem and extended-spectrum cephalosporin testing most likely are a result of the particular organisms tested, the susceptibility test systems themselves, and the way that an individual test was performed. This study highlights the need for monitoring how well susceptibility test systems detect emerging resistance and the need for more education about detecting and reporting emerging resistance in clinical laboratories. Acknowledgments We thank Hesna Yigit, Kamile Rasheed, and Jana Swenson for helpful discussions. We thank the microbiology personnel at Project ICARE hospitals for testing the organisms and providing results. Phase 3 of Project ICARE was supported in part by grants to the Rollins School of Public Health of Emory University by Astra-Zeneca Pharmaceuticals, Wilmington, DE (full sponsor); Pfizer, Incorporated, New York, NY (full sponsor); Roche Laboratories, Nutley, NJ (full sponsor); American Society for Health-System Pharmacists Research and Education Foundation, Bethesda, MD; Bayer Corporation, Pharmaceuticals Division, West Haven, CT; Kimberly Clark Corporation, Roswell, GA; National Foundation for Infectious Diseases, Bethesda, MD; and Rhône-Poulenc Rorer, Collegeville, PA (now Aventis Pharma). References AB BIODISK North America Inc. (1997). Etest MIC determination of antibiotics package insert. Piscataway, NJ: AB BIODISK North America Inc. Archibald, L., Phillips, L., Monnet, D., McGowan J. E., Jr., Tenover, F., & Gaynes, R. (1997). Antimicrobial resistance in isolates from inpatients and outpatients in the United States: the increasing importance of the intensive care unit. Clin Infect Dis, 24, 211 215. Biedenbach, D. J., & Jones, R. N. (1995). Interpretive errors using an automated system for the susceptibility testing of imipenem and aztreonam. Diagn Microbiol Infect Dis, 21, 57 60. Carmeli, Y., Eichelberger, K., Soja, D., Dakos, J., Venkataraman, L., DeGirolami, P., & Samore, M. (1998). Failure of quality control measures to prevent reporting of false resistance to imipenem, resulting in a pseudo-outbreak of imipenem-resistant Pseudomonas aeruginosa. J Clin Microbiol, 36, 595 597. Cercenado, E., Unal, S., Eliopoulos, C. T., Rubin, L. G., Isenberg, H. D., Moellering R. C., Jr., & Eliopoulos, G. M. (1996). Characterization of vancomycin resistance in Enterococcus durans. J Antimicrob Chemother, 36, 821 825. Clark, N. C., Cooksey, R. C., Hill, B. C., Swenson, J. M., & Tenover, F. C. (1993). Characterization of glycopeptide-resistant enterococci from U.S. hospitals. Antimicrob Agents Chemother, 37, 2311 2317. Doern, G. V., Brueggemann, A. B., Pfaller, M. A., & Jones, R. N. (1999). Assessment of laboratory performance with Streptococcus pneumoniae antimicrobial susceptibility testing in the United States: a report from the College of American Pathologists Microbiology Proficiency Survey Program. Arch Pathol Lab Med, 123, 285 289. Facklam, R. R., Sahm, D. F., & Teixeira, L. M. (1999). Enterococcus. In P. R. Murray, E. J. Baron, M. A. Pfaller, F. C. Tenover, & R. H. Yolken (Eds.), Manual of Clinical Microbiology (pp. 297 305). Washington, D.C.: American Society for Microbiology. Farmer, J. J. (1999). Enterobacteriaceae: Introduction and Identification. In P. R. Murray, E. J. Baron, M. A. Pfaller, F. C. Tenover, & R. H. & Yolken (Eds.), Manual of Clinical Microbiology (pp. 442 458).Washington, D.C.: American Society for Microbiology. Farmer, J. J., Asbury, M. A., Hickman, F. W., & Brenner, D. J., the Enterobacteriaceae Study Group. (1980). Enterobacter sakazakii: a new species of Enterobacteriaceae isolated from clinical specimens. Int J Syst Bacteriol, 30, 569 584. Fridkin, S. K., Steward, C. D., Edwards, J. R., Pryor, E. R., McGowan J. E., Jr., Archibald, L. K., Gaynes, R. P., & Tenover, F. C., Project Intensive Care Antimicrobial Resistance Epidemiology Hospitals. (1999). Surveillance of antimicrobial use and antimicrobial resistance in U.S. hospitals: Project ICARE Phase 2. Clin Infect Dis, 29, 245 252. Grist, R. (1992). External factors affecting imipenem performance in dried microdilution MIC plates [letter]. J Clin Microbiol, 30, 535 536. Jones, R. N., Marshall, S. A., Pfaller, M. A., Wilke, W. W., Hollis, R. J., Erwin, M. E., Edmond, M. B., Wenzel, R. P., & the SCOPE Hospital Study Group. (1997a). Nosocomial enterococcal blood stream infections in the SCOPE program: antimicrobial resistance, species occurrence, molecular testing results, and laboratory testing accuracy. Diagn Microbiol Infect Dis, 29, 95 102.

C.D. Steward et al. / Diagnostic Microbiology and Infectious Disease 38 (2000) 59 67 67 Jones, R. N., Marshall, S. A., & Zerva, L. (1997b). Critical evaluation of the Vitek GNS F6 card results compared to standardized, reference susceptibility test methods. Diagn Microbiol Infect Dis, 28, 35 40. Mabilat, C., & Goussard, S. (1993). PCR detection and identification of genes for extended-spectrum -lactamases. In D. H. Persing, T. F. Smith, F. C. Tenover, & T. J. White (Eds.), Diagnostic Molecular Microbiology: Principles and Applications (pp. 553 559). Washington, D.C.: American Society for Microbiology. National Committee for Clinical Laboratory Standards. (1997b). Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically. Fourth edition. NCCLS approved standard M7 A4. Wayne, PA: National Committee for Clinical Laboratory Standards. National Committee for Clinical Laboratory Standards. (1997a). Performance standards for antimicrobial disk susceptibility tests. Sixth edition. NCCLS approved standard M2 A6. Wayne, PA: National Committee for Clinical Laboratory Standards. National Committee for Clinical Laboratory Standards. (1998). Performance standards for antimicrobial susceptibility testing. NCCLS approved standard M100 S8. Wayne, PA: National Committee for Clinical Laboratory Standards. National Committee for Clinical Laboratory Standards. (1999). Performance standards for antimicrobial susceptibility testing. NCCLS approved standard M100 S9. Wayne, PA: National Committee for Clinical Laboratory Standards. O Rourke, E. J., Lambert, K. G., Parsonnet, K. C., Macone, A. B., & Goldmann, D. A. (1991). False resistance to imipenem with a microdilution susceptibility testing system. J Clin Microbiol, 29, 827 829. Pfaller, M. A., Korten, V., Jones, R. N., Doern, G. V., & Turkish Antimicrobial Resistance Study Group. (1999). Multicenter evaluation of the antimicrobial activity for seven broad-spectrum -lactams in Turkey using the Etest method. Diagn Microbiol Infect Dis, 35, 65 73. Rasheed, J. K., Jay, C., Metchock, B., Berkowitz, F., Weigel, L., Crellin, J., Steward, C., Hill, B., Medeiros, A. A., & Tenover, F. C. (1997). Evolution of extended-spectrum -lactam resistance (SHV-8) in a strain of Escherichia coli during multiple episodes of bacteremia. Antimicrob Agents Chemother, 41, 647 653. Rosenberg, J., Tenover, F. C., Wong, J., Jarvis, W., & Vugia, D. J. (1997). Are clinical laboratories in California accurately reporting vancomycin-resistant enterococci? J Clin Microbiol, 35, 2526 2530. Singer, D. A., Jochimsen, E. M., Gielerak, P., & Jarvis, W. R. (1996). Pseudo-outbreak of Enterococcus durans infections and colonization associated with introduction of an automated identification system software update. J Clin Microbiol, 34, 2685 2687. Steward, C. D., McGowan J. E., Jr., Tenover, F. C., Fridkin, S. K., Gaynes, R. P., & Project ICARE Hospitals. (1998). Multicenter study of susceptibility of imipenem demonstrates that testing problems persist, abstr. 769, IDSA, 36th Annual Meeting, Clin Infect Dis, 27, 1064. Tenover, F. C., Mohammed, M. J., Gorton, T. S., & Dembek, Z. F. (1999). Detection and reporting of organisms producing extended-spectrum -lactamases: survey of laboratories in Connecticut. J Clin Microbiol, 37, 4065 4070. Tenover, F. C., Tokars, J., Swenson, J., Paul, S., Spitalny, K., & Jarvis, W. (1993). Ability of clinical laboratories to detect antimicrobial agentresistant enterococci. J Clin Microbiol, 31, 1695 1699. Vannuffel, P., Gigi, J., Ezzedine, H., Vandercam, B., Delmee, M., Wauters, G., & Gala, J. L. (1995). Specific detection of methicillin-resistant Staphylococcus species by multiplex PCR. J Clin Microbiol, 33, 2864 2867. White, R. L., Kays, M. B., Friedrich, L. V., Brown, E. W., & Koonce, J. R. (1991). Pseudoresistance of Pseudomonas aeruginosa resulting from degradation of imipenem in an automated susceptibility testing system with predried panels. J Clin Microbiol, 29, 398 400. Yigit, H., Steward, C. D., Biddle, J. W., & Tenover, F. C. (1999). Characterization of -lactamases and porin changes in carbapenem-resistant isolates of Klebsiella pneumoniae and Serratia marcescens, abstr. A-61 In Abstracts of the, 99th American Society for Microbiology General Meeting, Washington, D.C.: American Society for Microbiology, pp13.