First Comprehensive Evaluation of the M.I.C. Evaluator Device Compared to Etest and CLSI Broth Microdilution for MIC Testing of Aerobic Gram-Positive and Gram-Negative Bacterial Species

Abstract

The M.I.C. Evaluator strip (Thermo Fisher Scientific, Basingstoke, United Kingdom) uses a methodology similar to that of Etest. In this first assessment of the M.I.C. Evaluator device, 409 strains of aerobic Gram-positive bacteria (staphylococci, streptococci, and enterococci) and 325 strains of Enterobacteriaceae, Pseudomonas species, and Acinetobacter species were tested by M.I.C. Evaluator strip, Etest, and broth microdilution as a reference standard. The Gram-positive bacteria included staphylococci (methicillin-resistant Staphylococcus aureus, methicillin-susceptible S. aureus, and coagulase-negative staphylococci), Streptococcus pneumoniae, beta-hemolytic streptococci and viridians group strains, vancomycin-resistant enterococci, and other enterococci. The Gram-negative bacteria included 250 strains of 60 Enterobacteriaceae species plus 50 Pseudomonas and 25 Acinetobacter species. A total of 14 antimicrobial agents (depending on the species) were included. The same methodology and reading format were used for M.I.C. Evaluator strips and Etest. Broth microdilution methodology was performed according to CLSI document M07-A8. For the clinical strains, >95% of results were plus or minus one doubling dilution for all species. There were fewer than 5% minor errors, fewer than 3% major errors, and fewer than 1% very major errors. M.I.C. Evaluator strips and Etest often reported higher MICs than the reference broth microdilution method. The M.I.C. Evaluator strips provided results comparable to those of the predicate Etest device and are of value for the accurate testing of MICs for these important pathogens.

INTRODUCTION

There are three methods that permit accurate MIC testing. These are broth and agar dilution tests and a gradient endpoint method that incorporates the antimicrobial agent as a defined gradient onto the back of a solid plastic strip. MICs are read where the zone of inhibition intersects with the drug concentration on the strip. Until now, the only gradient endpoint device available was Etest, developed by AB Biodisk, Solna, Sweden, and currently marketed by bioMérieux. This method has previously been shown to be an accurate and simple alternative to reference broth or disk methods (9, 12, 15, 18).

The M.I.C. Evaluator strip (M.I.C.E) is a new gradient endpoint susceptibility test device. Both the M.I.C. Evaluator strip and Etest are performed similarly and have excellent utility for directed MIC testing against specific agents and microorganisms isolated from clinical samples. The advantage of these devices is the ease of use and accurate MIC results for patient management. MIC data also are valuable for clinical pharmacists to determine optimal pharmacodynamic dosing. In these situations, it is important to be able to detect resistance early in the course of treatment (5, 7, 10). To ensure that any new device is effective in the clinical laboratory, appropriate validation studies, including both control strains and an extensive collection of clinical isolates, is essential to ensure evaluable results for patient care.

Gram-positive and Gram-negative aerobic and facultative anaerobic bacteria occupy the largest group of microorganisms associated with infections in patients in critical care settings. The consequences of inappropriate therapy or of inadequate dosing in these patients may have important clinical consequences. It is, however, important that the in vitro MIC device provides an accurate estimation of the efficacy of the antimicrobial agent against the infecting organism. We compared the new M.I.C. Evaluator strip to Etest to investigate any discordance between the two devices, as well as to the standard CLSI reference broth microdilution method (2), on a collection of staphylococci, streptococci, and enterococci. We also evaluated the device on a large collection of enterobacteriaceae, as well as Pseudomonas species and Acinetobacter species. This is the first comprehensive evaluation of this new gradient endpoint device that compares it to the CLSI reference broth dilution method. The M.I.C. Evaluator device has been compared to the British Society for Antimicrobial Chemotherapy (BSAC) method, with observations similar to those in the current study (1, 8). Other collections of bacterial species, including anaerobes and fastidious bacteria, are reported in a companion publication (11).

MATERIALS AND METHODS

Bacterial species. (i) Gram-positive bacteria.

A total of 409 fresh or recently frozen clinical strains of staphylococci, streptococci, and enterococci were tested. All isolates were from individual patients at the University of Alberta Hospital, Edmonton, Canada. They were identified in our laboratory using primary tests, standard automated methods, or molecular tests as required. Strains were stored at −70°C in litmus milk prior to testing. Strains were also characterized for resistance determinants where appropriate. Methicillin resistance in Staphylococcus aureus was confirmed by cefoxitin disc testing, PBP 2′ testing, and mecA gene testing. There were 157 strains of S. aureus, including methicillin-susceptible strains (MSSA) (84), methicillin-resistant S. aureus (MRSA) (20), and coagulase-negative staphylococci (CNS) (53); 151 strains of streptococci, including Streptococcus pneumoniae (60), beta-hemolytic streptococci (40), and viridans group streptococci (51); and 100 strains of enterococci (52 vancomycin-resistant enterococci [VRE] and 48 Enterococcus species). The species of strains of coagulase-negative staphylococci and vancomycin-susceptible enterococci were not fully determined, since this did not bear on the comparison of antimicrobial test results. Among the 51 strains included as viridans group streptococci, there were 13 S. mitis, 4 S. salivarius, 3 S. sanguis, 2 S. dysgalactiae, 1 each of S. mutans and S. equi subsp. zooepidemicus, and 16 S. anginosus strains; the species of 11 strains were not determined. Within the beta-hemolytic streptococci, there were 25 strains of S. pyogenes (group A), 11 strains of S. agalactiae (group B), and 4 strains of other species (group C). Among the vancomycin-resistant enterococci, 29 strains carried VanA and 14 strains carried VanB. There were 8 strains classified as VanC (Enterococcus casseliflavus or Enterococcus gallinarum), and there was 1 strain of E. gallinarum with both VanA and VanC genes. The quality control strains for Gram-positive tests were S. aureus ATCC 29213, S. pneumoniae ATCC 49619, and Enterococcus faecalis ATCC 29212.

(ii) Gram-negative bacteria.

For the enterobacteriaceae, a total of 250 strains, each from individual patients, were tested. There were 16 common genera and 60 species in this collection of strains (Table 1). For Pseudomonas and Acinetobacter species, the following organisms were tested: 50 Pseudomonas species (40 P. aeruginosa, 5 P. stutzeri, 3 P. fluorescens, and 1 each of P. pseudoalcaligenes and P. mendocina) and 25 strains of the Acinetobacter baumannii complex. The same methodologies as those for the Gram-positive species were used for the identification and for the storage of strains prior to testing.

Table 1.

Enterobacterial species tested^a

Species	No. of strains
Escherichia coli	44
Escherichia coli 0157:H7	3
Escherichia fergusonii	1
Escherichia hermanii	1
Escherichia vulneris	1
Klebsiella pneumoniae	29
Klebsiella oxytoca	14
Klebsiella ozaenae	4
Klebsiella rhinoscleromatis	2
Enterobacter cloacae	24
Enterobacter aerogenes	15
Enterobacter sakazakii	3
Enterobacter amnigenus	4
Enterobacter agglomerans	1
Enterobacter gergoviae	2
Enterobacter asburiae	2
Enterobacter cancerogenus	1
Salmonella enterica
Serovar Typhi	2
Serovar Enteritidis	2
Serovar Typhimurium	2
Serovar Paratyphi	4
Serovar Heidelberg	1
Serovar Kentucky	1
Serovar Hadar	1
Serovar Newport	1
Serovar Cholerasuis	1
Serovar Arizonae	1
Serovar Muenchen	1
Serovar Derby	1
Serovar Infantis	1
Serovar Lomalinda	1
Citrobacter freundii	11
Citrobacter koseri	3
Citrobacter amalonaticus	1
Citrobacter braakii	1
Citrobacter farmeri	1
Citrobacter sedlackii	1
Citrobacter werkmanii	1
Citrobacter youngae	1
Shigella sonnei	5
Shigella dysenteriae	4
Shigella boydii	5
Shigella flexneri	5
Proteus mirabilis	10
Proteus vulgaris	7
Proteus penneri	3
Yersinia frederiksenii	1
Yersinia intermedia	1
Yersinia enterocolitica	2
Ewingella americana	1
Serratia marcescens	3
Serratia odorifera	1
Serratia liquefaciens	1
Morganella morganii	3
Providencia stuartii	2
Providencia rettgeri	1
Hafnia alvei	1
Ochrobactrum anthropi	1
Pantoea agglomerans	1
Edwardsiella tarda	1

^aSixteen genera and 250 strains were tested.

Four standard quality control strains were tested against the Gram-negative isolates as appropriate. These were Escherichia coli ATCC 25922, P. aeruginosa ATCC 27853, S. aureus ATCC 29213, and Enterococcus faecalis ATCC 29212. CLSI quality control MIC ranges were available for all antimicrobial agents tested in this study.

Antimicrobial agents. (i) Gram-positive bacteria.

A total of 13 antimicrobial agents were tested against the Gram-positive bacterial strains. The agents tested were ampicillin, amoxicillin, amoxicillin-clavulanate, cefotaxime, ciprofloxacin, erythromycin, gentamicin, levofloxacin, linezolid, oxacillin, penicillin, tetracycline, and vancomycin. Agents were tested as appropriate for the species.

(ii) Gram-negative bacteria.

In this study, nine agents produced for the M.I.C. Evaluator strips were tested: ampicillin, amoxicillin, amoxicillin-clavulanate, cefotaxime, tetracycline, gentamicin, ciprofloxacin, imipenem, and levofloxacin. The concentration ranges were similar for Etest and the M.I.C. Evaluator device. Etest strips were purchased from the manufacturer (bioMérieux). M.I.C. Evaluator strips were supplied by Thermo Fisher Scientific, Basingstoke, United Kingdom.

CLSI broth microdilution tests.

Standard CLSI broth microdilution methodology (2) was used to test for susceptibility, with media of cation-supplemented Mueller-Hinton broth (CSMHB) for staphylococci, enterococci, and for the Gram-negative species and CSMHB with 5% horse blood for streptococci. Strains were incubated at 35°C for 20 to 24 h, and susceptibility was determined according to standard CLSI breakpoints (3).

Gradient endpoint device testing.

The same strains were tested using the same inocula as those used for the broth microdilution tests. The bacterial strains were inoculated onto appropriate Mueller-Hinton agar (Thermo Fisher, Basingstoke, United Kingdom) according to CLSI M07-A8 (2). M.I.C.E and Etest strips were added to the plates according to the manufacturers' instructions and were incubated for 20 to 24 h at 35°C. Complete inhibition was read for each manufacturers' strip at the point where the elliptical zone intersected with the strip. For trimethoprim-sulfamethoxazole, erythromycin, tetracycline, and linezolid, where 80% inhibition was taken as the endpoint, the manufacturer's diagrams and photographs were used for the analysis of the test results.

Performance criteria.

Standard performance criteria were included in this method for comparisons of the two gradient endpoint devices against the reference broth microdilution method, including ≥95% quality control strains within acceptable ranges, ≥90% essential and categorical agreement, ≤10% minor errors, ≤3% major errors, and ≤ 1.5% very major errors. Essential agreement was defined for this study as a result being plus or minus one doubling dilution of that from broth microdilution. These criteria are identified in FDA guidance documents (16) and are universally used.

RESULTS

The quality control organisms for both Gram-positive and Gram-negative isolates were tested a total of seven times during this evaluation. Quality control strains were included with each test run. Results for all of the strains tested showed that the MICs were within published CLSI ranges for each of the antimicrobial agents tested. Each strain tested in the study was isolated from a different patient, although as would be expected in most current hospital settings, it is possible that some of the resistant strains were in fact clonal and associated with ongoing outbreaks.

For the Gram-negative enterobacterial isolates, a total of 250 strains of 16 genera and 60 species were tested by all three methods (Table 1). For isolates of Pseudomonas and Acinetobacter species, a total of 75 strains were tested.

The number of susceptible (S), intermediate (I), and resistant (R) strains and their essential agreements in comparisons of M.I.C. Evaluator strips and Etest to the standard reference method for each of the Gram-positive genera are summarized in Tables 2 (staphylococci), 3 (streptococci), and 4 (enterococci). There were very few resistant streptococci, as would be expected, and S, I, and R results for those species are not included. The combined error rates for all Gram-positive bacteria are shown in Table 5.

Table 2.

Essential agreement for M.I.C.E and Etest compared to broth microdilution for 157 strains of staphylococci

Drug	No. of strains in each resistance category			No. of strains according to no. of doubling dilutions away from reference				Essential agreement (%) (± 1 dilution)		No. of strains according to no. of doubling dilutions away from reference
				>−2		−2				2		>2
	S	I	R	M.I.C.E	Etest	M.I.C.E	Etest	M.I.C.E.	Etest	M.I.C.E	Etest	M.I.C.E	Etest
Cefotaxime	132	9	16	0	0	0	0	105 (67)	128 (82)	31	16	21	13
Ciprofloxacin	104	3	50	0	0	1	1	67 (43)	70 (45	62	55	27	31
Erythromycin	81	0	76	7	6	3	0	147 (94)	150 (96)	0	0	0	1
Gentamicin	133	7	17	3	3	7	8	144 (92)	145 (92)	2	1	1	0
Levofloxacin	102	4	51	0	0	2	2	131 (83)	129 (82)	21	22	3	4
Linezolid	155	1	1	0	23	39	53	116 (74)	81 (52)	2	0	0	0
Oxacillin	111	0	46	0	0	1	1	103 (66)	95 (61)	20	29	34	33
Penicillin	26	0	131	12	17	9	9	110 (70)	110 (70)	20	13	6	8
Tetracycline	142	3	12	1	1	0	0	115 (73)	70 (45)	32	68	9	18
Vancomycin	154	2	1	0	0	0	0	95 (61)	132 (84)	59	9	3	1
Total	1,140	29	401	23	50	62	74	1,133 (72)	1,110 (71)	218	197	83	96

Table 3.

Essential agreement for M.I.C. Evaluator strips (M.I.C.E) and Etest compared to broth microdilution for 151 strains of streptococci

Agent	No. of strains according to no. of doubling dilutions away from reference				Essential agreement (%) (±1 dilution)		No. of strains according to no. of doubling dilutions away from reference
	>−2		−2				2		>2
	M.I.C.E	Etest	M.I.C.E	Etest	M.I.C.E.	Etest	M.I.C.E	Etest	M.I.C.E	Etest
Amoxicillin	1	1	1	3	144 (95)	144 (95)	4	2	1	1
Ampicillin	1	1	6	3	141 (93)	144 (95)	1	1	2	2
Cefotaxime	0	0	1	1	120 (79)	125 (83)	23	20	7	5
Ciprofloxacin	0	0	0	0	136 (90)	139 (91)	13	11	2	1
Erythromycin	1	0	2	1	82 (54)	66 (44)	47	33	19	51
Gentamicin	6	9	4	6	77 (51)	105 (70)	49	24	15	9
Imipenem	2	0	7	2	137 (90)	81 (54)	3	47	2	21
Levofloxacin	0	0	0	0	130 (86)	132 (87)	19	17	2	2
Linezolid	1	1	0	6	139 (92)	142 (94)	11	2	0	0
Oxacillin	1	1	7	2	137 (91)	141 (94)	4	3	2	2
Penicillin	0	0	3	2	133 (88)	137 (90)	13	11	2	1
Tetracycline	1	1	1	1	122 (81)	111 (73)	16	27	11	11
Vancomycin	1	0	1	0	145 (96)	148 (98)	3	2	1	1
Total	15	14	33	27	1,643 (84)	1,615 (82)	206	200	66	105

Table 4.

Essential agreement for M.I.C.E and Etest compared to broth microdilution for 100 enterococcal strains

Agent	No. of strains in each resistance categorya			No. of strains according to no. of doubling dilutions away from reference				Essential agreement (%) (±1 dilution)		No. of strains according to no. of doubling dilutions away from reference
				>−2		−2				2		>2
	S	I	R	M.I.C.E	Etest	M.I.C.E	Etest	M.I.C.E	Etest	M.I.C.E	Etest	M.I.C.E	Etest
Ampicillin	63	0	38	0	0	1	1	92 (91)	95 (94)	8	5	0	0
Gentamicin	NA			0	0	0	0	72 (71)	86 (85)	21	14	8	1
Imipenem	NA			0	0	2	1	97 (96)	96 (95)	2	3	0	1
Linezolid	70	31	0	1	0	0	10	100 (99)	91 (90)	0	0	0	0
Vancomycin	61	9	31	0	0	1	1	78 (77)	92 (91)	20	6	2	2
Total	194	40	69	1	0	4	13	439 (88)	460 (92)	51	28	10	4

^aNA, no CLSI breakpoints available.

Table 5.

Error rates for 409 strains of staphylococcal, streptococcal, and enterococcal species

Agent	No. (%) of errors
	Minor		Major		Very major
	M.I.C.E	Etest	M.I.C.E	Etest	M.I.C.E	Etest
Amoxicillin	0	0	0	0	0	0
Ampicillin	4 (2)	5 (3)	0	0	0	0
Cefotaxime	20 (6)	15 (5)	12 (4)	9 (3)	0	0
Ciprofloxacin	6 (4)	7 (4)	2 (1)	1 (1)	0	0
Erythromycin	16 (5)	16 (5)	7 (2)	8 (3)	0	0
Gentamicin	9 (6)	5 (3)	0	0	2 (1)	0
Imipenem	2 (1)	1 (1)	0	0	0	0
Levofloxacin	2 (1)	2 (1)	0	0	0	0
Linezolid	27 (7)	31 (8)	1 (1)	1 (1)	0	0
Oxacillin	1 (1)	1 (1)	16 (10)	16 (10)	0	0
Penicillin	6 (2)	4 (1)	1 (1)	0	7 (2)	6 (2)
Tetracycline	1 (1)	1 (1)	2 (1)	2 (1)	0	0
Vancomycin	15 (4)	10 (2)	1 (1)	0	0	0
Total	109 (3.3)	98 (2.5)	42 (1.5)	37 (1.4)	9 (0.2)	6 (0.2)

For the staphylococci, essential agreement was <90% for all agents except erythromycin and gentamicin. Categorical agreement, however, was >95% for all agents except cefotaxime and oxacillin. M.I.C. Evaluator strips and Etest performed in an equivalent manner. As shown in Table 2, both gradient endpoint devices gave larger numbers of results at two or more dilutions above the reference method MIC for most of the agents tested with staphylococci. The major issue that was discovered was with oxacillin and penicillin. Higher MICs for the two gradient devices resulted in some major errors for oxacillin, and some lower gradient penicillin MICs resulted in a number of very major errors. Therefore, the gradient endpoint devices would report oxacillin as resistant, whereas the reference method called these strains susceptible. The opposite occurred for penicillin (Table 5). On closer examination, all of the oxacillin discrepancies were among methicillin-resistant S. aureus isolates that were captured by M.I.C.E strips and Etest but not by broth microdilution. For penicillin, discordant results all occurred among oxacillin-susceptible S. aureus strains. We did not observe any small colonies (so-called fuzzy zones) surrounding the gradient endpoint strips in these isolates. There were two very major errors with gentamicin, one for an S. pyogenes isolate and one for an S. anginosus isolate.

For the streptococci (Table 3), essential agreement was low for erythromycin, gentamicin, tetracycline, and cefotaxime, but the categorical agreement was >95% for all agents that have CLSI breakpoints except for erythromycin (90 and 87% for M.I.C. Evaluator and Etest, respectively). The essential agreement of Etest (54%) for imipenem was much lower than that of M.I.C.E (90%). As in the case of staphylococci, streptococcal MICs with the gradient endpoint devices tended to be higher than the reference MICs (Table 3). Most of the streptococci have very low MICs for all agents tested, such that a 2-fold difference in MIC between the gradient endpoint and reference broth method might only reflect a change from a MIC of 0.004 to 0.016, for example. This would not affect categorical agreement or reporting on the likely efficacy of that agent.

For the 100 strains of enterococci tested, ampicillin had a high essential agreement, but for linezolid and vancomycin the agreements were variable and lower between the two devices compared to reference broth microdilution. For linezolid, all results were clustered within a narrow range of MICs (0.5 to 4 mg/liter), so that while the essential agreements were good, the categorical agreements were lower (74 and 70% for M.I.C. Evaluator and Etest, respectively) because the result for some strains was 4 mg/liter, whereas the MIC for the gradient endpoints devices was 2 mg/liter. For vancomycin, as was the case of other Gram-positive species, the gradient endpoint devices had slightly higher MICs than the reference method. Thus, almost all of the categorical results had minor errors for both linezolid and vancomycin. There was one major error for M.I.C. Evaluator strips and Etest with linezolid and one major error for M.I.C.E with vancomycin (Table 5).

For the Gram-negative species studied (Table 6), essential agreement also was high, except for the two fluoroquinolones ciprofloxacin and levofloxacin, but these results did not affect categorical agreements. Categorical agreement was ≥95% for all strains tested except for ampicillin and amoxicillin-clavulanic acid (both were 90 to 91%). Very major errors were observed only with ampicillin and amoxicillin-clavulanate (Table 7). All of these occurred with P. vulgaris and one C. freundii isolate. All other microorganism-antimicrobial agent combinations were within acceptable rates of error.

Table 6.

Combined essential agreement for 325 strains of enterobacterial, Pseudomonas, and Acinetobacter species for M.I.C.E and Etest compared to broth microdilution

Agent	No. of strains in each resistance category			No. of strains according to no. of doubling dilutions away from reference				Essential agreement (%) (±1 dilution)		No. of strains according to no. of doubling dilutions away from reference
				>−2		−2				+2		>2
	S	I	R	M.I.C.E	Etest	M.I.C.E	Etest	M.I.C.E	Etest	M.I.C.E	Etest	M.I.C.E	Etest
Amoxicillina				7	7	10	12	227 (91)	225 (90)	4	4	2	2
Amoxicillin-clavulanate	154	26	70	1	2	5	10	219 (88)	224 (90)	20	11	4	3
Ampicillin	92	15	143	3	6	10	11	221 (89)	226 (90)	13	3	4	5
Ciprofloxacin	269	8	48	2	1	12	4	257 (76)	281 (85)	49	35	5	4
Cefotaxime	224	1	25	4	6	10	9	225 (90)	221 (88)	9	13	2	1
Gentamicin	275	6	31	3	4	18	17	292 (89)	297 (92)	1	1	3	2
Imipenem	59	1	15	0	0	0	1	70 (93)	61 (81)	5	13	0	0
Levofloxacin	227	6	17	0	0	1	1	209 (84)	192 (77)	36	51	5	6
Total	1,638	65	372	20	26	66	65	1,720 (87)	1,727 (87)	137	131	25	23

^aNo resistance breakpoints were found for the strains tested.

Table 7.

Combined error rates for 325 strains of enterobacterial, Pseudomonas, and Acinetobacter species for M.I.C.E and Etest compared to broth microdilution

Antimicrobial agent	No. (%) of errors
	Minor		Major		Very major
	M.I.C.E	Etest	M.I.C.E	Etest	M.I.C.E	Etest
Ampicillin	17 (7)	16 (6)	1 (0.4)	1 (0.4)	7 (3)	9 (4)
Amoxicillin-clavulanate	21 (8)	21 (8)	0	0	2 (1)	4 (2)
Cefotaxime	3 (1)	2 (1)	1 (0.4)	1 (0.4)	0	0
Ciprofloxacin	2 (1)	6 (2)	0	0	0	0
Gentamicin	9 (3)	3 (1)	1 (0.3)	1 (0.3)	0	0
Imipenem	1 (1)	4 (5)	0	0	0	0
Levofloxacin	2 (1)	1 (0.4)	0	0	0	0
Total	55 (3)	53 (3)	3 (0.1)	3 (0.1)	9 (0.5)	13 (0.9)

DISCUSSION

This study was performed with the first available antimicrobial agents produced for the M.I.C. Evaluator strips. For those agents, the comparison was made first against Etest as the predicate device to identify any major method discrepancies and then against the CLSI broth microdilution method as the reference method.

The data showed that the M.I.C. Evaluator strips are comparable to Etest and to the broth dilution reference method for these agents against a comprehensive range of aerobic Gram-positive and Gram-negative bacterial species. For most antimicrobial agents with most bacterial species, it was remarkable how similarly the two gradient endpoint methods performed for a large number of strains of these varied species. Additional agents are now being produced and will be tested in the same manner to verify the capabilities of the M.I.C. Evaluator strips.

The categorical agreements for both M.I.C. Evaluator strips and for Etest were comparable. The most notable issues occurred with beta-lactam antimicrobials, erythromycin, and with the fluoroquinolones. Since all of these agents were not tested against every species, it is difficult to know if this is a genus phenomenon or is broadly observed across species. For erythromycin, it is well known that there is less agreement between broth and agar methods, since testing is affected by pH changes from the CO₂ required for incubation in agar. We have also discovered with the enterobacteriaceae, for example, that there are more minor errors with beta-lactam agents. In most cases the gradient endpoint methods tend to give higher MICs than the reference broth dilution method, and this was consistent for the Gram-positive species. For a number of antimicrobial-organism combinations, particularly carbapenems, the Etest has been reported to yield higher MICs than broth microdilution (4, 6, 13, 17). It is possible that the small volume used in the broth method reduces the likelihood of finding slower-growing resistant subpopulations, but this is not well understood.

Unlike our findings with the enterobacteriaceae, categorical agreements with ampicillin for streptococci and enterococci were very high. This is due to very low MICs for these species compared to those for enterobacteriaceae, so the categorical agreements would be expected to be much better. We have noted that with some microorganism-antimicrobial agent combinations, both M.I.C. Evaluator strips and Etest tend to report slightly higher MICs than broth microdilution and may well identify resistance determinants more readily (data not shown). This phenomenon has been observed in comparisons of the Etest to commercial broth dilution systems when testing for resistance markers such as vancomycin in staphylococci (14). In this study, we observed higher oxacillin MICs among the staphylococci, resulting in 10% major errors for both M.I.C. Evaluator strips and Etest compared to the reference broth method. We determined that these errors were for methicillin-resistant staphylococci. This is significant, since the reference broth dilution method would undercall oxacillin resistance. Most laboratories use more than one test to detect MRSA, but our observations support the validity of the gradient endpoint methods to find these isolates.

In more than 90% of minor and major errors, when we examined the results, the M.I.C. Evaluator strip and Etest MICS were higher than the broth microdilution results. For enterococci, there was only one M.I.C.E result where the broth microdilution result was susceptible. These test systems measure different interactions between an antimicrobial agent and microorganism. It is to be expected that there will be some minor differences between the systems. It is quite satisfying that both gradient endpoint systems provide MIC measurements that are very close to those of the broth microdilution reference method.

We expect that these results will establish a mechanism to test additional M.I.C. Evaluator strip agents against Gram-positive and Gram-negative bacteria as they become available. A companion manuscript (11) will report on the validity of the M.I.C. Evaluator strips against anaerobes and other fastidious bacterial species. All of these studies have shown that the M.I.C. Evaluator strips provide accurate estimations of the MIC of these bacteria that are equivalent to results from Etest. Differences with reference broth microdilution observations are due to the different formats for these tests and to the mechanics of the interactions between bacterial cells on agar versus those in broth. For the most part, the results are very consistent and provide a rationale for the use of the M.I.C. Evaluator strips for routine MIC antimicrobial susceptibility testing.