Performance of the Oxoid M.I.C.Evaluator Strips compared with the Etest assay and BSAC agar dilution.

OBJECTIVES: The Oxoid M.I.C.Evaluator (M.I.C.E; Thermo Fisher Scientific) comprises an antibiotic gradient on a plastic support. We compared its performance with Etest--a similar product--using BSAC agar dilution as a reference.
METHODS: Parallel MIC tests were performed by M.I.C.Evaluator, Etest and agar dilution on Iso-Sensitest agar. In total, 9354 organism/strip combinations were tested by each method, using 1017 bacteria representing clinically important fastidious and non-fastidious species.
RESULTS: Essential agreement of strip MIC values (+/-1 doubling dilution) with the agar dilution reference, with off-scale results excluded, was 89.9% for M.I.C.Evaluator versus 89.5% for Etest (P > 0.05). These proportions were similar, at 89.5% and 89.3% (P > 0.05), respectively, if off-scale values were counted as agreeing if they could agree (e.g. a strip MIC >32 mg/L and an agar dilution MIC of 128 mg/L). For both strips, agreement with agar dilution was best for non-fastidious genera, Moraxella, Listeria, Pasteurella and Campylobacter spp. and weaker for streptococci, anaerobes, Neisseria spp. and, especially, Haemophilus influenzae. Many 'disagreements', especially for H. influenzae, concerned organisms unequivocally resistant by all methods (e.g. ampicillin MIC 256 mg/L by agar dilution, 16 or 32 mg/L by both strips); nevertheless both strips underestimated imipenem MICs for Proteus. There was no difference between the two strip types in the proportion of agreements with agar dilution (P > 0.05); nevertheless their results agreed better with each other than with agar dilution (P < 0.01).
CONCLUSIONS: The M.I.C.Evaluator performed almost identically to the Etest, giving good agreement with BSAC agar dilution.

Introduction

Discs and automated systems with truncated MIC ranges are adequate for most routine susceptibility testing, but precise MIC determinations are needed in difficult settings, such as endocarditis and pneumococcal meningitis. MICs are also needed to guide the treatment of infections caused by multiresistant pathogens, where pharmacodynamically based dose adjustment may be sought. Finally, MICs are needed for organisms or antibiotic/organism combinations where disc testing is demonstrably unreliable, e.g. anaerobes, penicillin against most α-haemolytic streptococci except pneumococci, and glycopeptides against Staphylococcus aureus, where diffusion tests fail to discriminate strains with vancomycin MICs of 8 mg/L,[1] let alone those with the small reductions in susceptibility (MICs 2–4 mg/L) now being associated with poor clinical outcomes.[2]

Classical MIC determinations on agar or in broth are routinely performed by specialist centres, but are inconvenient for diagnostic laboratories, where only a few isolates require these investigations and where the range of drugs to be tested varies with each isolate. Rather, diagnostic laboratories find it more convenient to perform their few MIC tests using pre-formed antibiotic gradients, such as Etest® (AB bioMérieux, Marcy l'Étoile, France). These are versatile and give results in good agreement with the CLSI broth microdilution method, against which they are calibrated.[3] Their agreement with other methods, such as that of the BSAC is less well validated, but is asserted to be acceptable by the BSAC.[4]

The Oxoid M.I.C.EvaluatorTM Strip (Thermo Fisher Scientific, Basingstoke, UK) is a new gradient strip for MIC determinations, asserted by the manufacturer to be suitable for use on Iso-Sensitest agar as well as Mueller–Hinton agar. We compared its performance with that of the Etest Strip and the BSAC agar dilution methods.

Materials and methods

Bacteria

The same 1017 bacterial strains and antibiotics were tested using M.I.C.Evaluator Strips, Etests and the BSAC agar dilution method, with a total of 9354 antibiotic/strain combinations. The organisms were recent clinical isolates, selected to represent a wide range of species and susceptibilities, as summarized in Table 1. Controls comprised Escherichia coli NCTC 10418 and ATCC® 25922™, Pseudomonas aeruginosa ATCC® 27853™, Staphylococcus aureus ATCC® 25923™, ATCC® 29213™ and ATCC® 43300™, Streptococcus pneumoniae ATCC® 49619™ and ATCC® 49620™, Haemophilus influenzae ATCC® 49247™, Neisseria gonorrhoeae ATCC® 49226™, Enterococcus faecalis ATCC® 29212™, Bacteroides fragilis ATCC® 25285™ and Bacteroides thetaiotaomicron ATCC® 29741™, all provided as Culti-Loops™ (Thermo Fisher Scientific).

Table 1

Summary MIC parameters (mg/L) for the test strains, as determined by the BSAC agar dilution method

	Enterobacteriaceae (250)a	Non-fermenters (76)b	Staphylococci (154)c	Enterococci (100)d	Streptococci (161)e	Neisseria (56)f	Haemophilus (56)g	Anaerobes (101)h	Moraxella (10)i	Listeria (21)j	Pasteurella (10)k	Campylobacter (22)l
Amoxicillin/clavulanic acid range	0.25–256						0.25–8	0.015–16	0.03–0.25
MIC50	16						1	0.25	0.06
MIC90	64						2	2	0.25
Amoxicillin range	0.5 to >256				0.008–4		0.25–256
MIC50	256				0.06		2
MIC90	>256				1		128
Ampicillin range	0.5 to >256			0.06 to >256	0.008–4	0.013–128	0.25–256			0.12–0.25
MIC50	256			2	0.125	0.5	1			0.25
MIC90	>256			>256	2	2	128			0.25
Cefotaxime range	0.008–256		0.25 to >256		0.008–4	≤0.002–0.5	0.004 to >8
MIC50	0.25		4		0.06	0.25	0.015
MIC90	128		>256		0.5	0.125	0.06
Ciprofloxacin range	0.015 to >32	0.03 to >32	0.12 to >32		0.5–16	0.004–32	0.004–16					0.06 to >32
MIC50	0.06	0.5	1		2	0.008	0.008					0.12
MIC90	>32	>32	>32		4	4	0.015					32
Erythromycin range			0.06 to >256		0.06 to >256		1–64			0.06–0.12		0.25–4
MIC50			0.5		0.12		8			0.12		0.5
MIC90			>256		32		16			0.12		1
Gentamicin range	0.25–512	0.125 to >1024	0.016 to >1024	2 to >1024	1–64
MIC50	0.5	16	0.25	256	8
MIC90	64	>1024	32	>1024	64
Imipenem range	0.06–4	0.03 to >32		0.25–2m	0.002–0.5		0.06–2	0.008–4
MIC50	0.25	2		1m	0.016		0.5	0.12
MIC90	2	32		2	0.12		2	1
Levofloxacin range	0.015 to >32		0.12 to >32		0.5–4		0.008–4
MIC50	0.125		0.5		1		0.015
MIC90	16		32		2		0.06
Linezolid range			0.5–32	1–64	0.5–2
MIC50			2	2	1
MIC90			2	16	2
Oxacillin range			0.06 to >256		0.03–16
MIC50			8		0.5
MIC90			>256		8
Penicillin G range			0.016–128		0.004–2	0.015–64		0.008 to >512			0.016–0.25
MIC50			8		0.03	0.12		4			0.12
MIC90			128		1	1		128			0.12
Tetracycline range			0.12–256		0.125–128	0.25–16	0.12–16
MIC50			0.5		0.5	0.25	0.5
MIC90			64		64	4	2
Vancomycin range			0.5–4	0.5 to >256	0.25–1
MIC50			1	8	0.5
MIC90			2	>256	0.5
Metronidazole range								0.015 to >256
MIC50								0.5
MIC90								1

aComprising 50 Citrobacter spp., 50 Enterobacter spp., 50 E. coli, 50 Klebsiella spp., 29 Proteus spp. and 21 Salmonella spp.

bComprising 26 Acinetobacter spp., 42 P. aeruginosa and 8 Pseudomonas spp.

cComprising 104 S. aureus and 50 coagulase-negative staphylococci.

dComprising 50 Enterococcus faecium, 40 E. faecalis and 10 other named Enterococcus spp.

eComprising 34 Streptococcus Lancefield A, 34 Streptococcus Lancefield B, 39 S. pneumoniae and 54 other α-haemolytic streptococci.

fComprising 27 N. gonorrhoeae and 29 Neisseria meningitidis.

gComprising 46 H. influenzae and 10 Haemophilus parainfluenzae.

hComprising 32 Clostridium spp., 46 Bacteroides spp., 8 Anaerococcus spp., 7 Peptoniphilus spp., 3 Finegoldia spp. 3 Peptostreptococcus spp. and 2 Parvimonas spp.

iComprising 10 Moraxella catarrhalis.

jComprising 11 Listeria monocytogenes, 9 Listeria innocua and 1 Listeria ivanovii.

kComprising 7 Pasteurella multocida and 3 unidentified Pasteurella spp.

lComprising 12 Campylobacter jejuni, 4 Campylobacter coli, 5 Campylobacter fetus and 1 Campylobacter lari.

mOnly E. faecalis tested.

Susceptibility tests

Tests with both the M.I.C.Evaluator and Etest Strips were performed on Iso-Sensitest agar (Thermo Fisher Scientific) with the supplements, incubation periods and conditions specified in the BSAC agar dilution method, except that: (i) based on advice for Etests, p-nitrophenyl glycerol was omitted from the media for strip tests with Proteeae, whereas it was included, at 50 mg/L, as an anti-swarming agent in agar dilution MIC determinations; and (ii) based on the package insert for the M.I.C.Evaluator Strips, Iso-Sensitest agar was used throughout for tests with imipenem, although the BSAC advocates Mueller–Hinton agar for this and other carbapenems. Oxacillin strips were tested on Mueller–Hinton agar supplemented with 2% NaCl as per the BSAC agar dilution method.

For all isolates the inocula for strip tests were matched to a 0.5 McFarland standard. Results were read in accordance with the manufacturers' directions, which are essentially identical for both strip products. For bactericidal antibiotics, the MIC was taken as the point of termination of all growth; for bacteriostatic agents as the point of 80% inhibition. In those cases where growth terminated between two points on the strip scale, the MIC was rounded to the higher value. The following strip types were used: 0.002–32 µg strips of ciprofloxacin, cefotaxime, imipenem, levofloxacin and penicillin G; 0.015–256 µg strips of amoxicillin, amoxicillin/clavulanic acid, ampicillin, cefotaxime, erythromycin, gentamicin, linezolid, metronidazole, oxacillin, penicillin G, tetracycline and vancomycin; and 0.06–1024 µg strips of gentamicin. Storage was at 2–8°C for the M.I.C.Evaluator Strips and −20°C for Etests. Boxes were allowed to equilibrate at room temperature for at least 1 h before opening.

Agar dilution MICs were determined by the BSAC method, as current in 2007,[5] on oblong 12 × 8 cm plates, inoculated using a 96-point inoculator. Results were read with an automated optical reader (Perceptive Instruments, Haverhill, UK) and were manually corrected as necessary.

Data handling

Data analysis was performed in Microsoft Office Excel®, by comparing the number of doubling dilutions difference between each of the gradient test strip results and the reference method. Intermediate MIC values (1.5, 3 or 6 mg/L etc.) read from the M.I.C.Evaluator Strip or Etest scales were rounded up to the next highest value on the standard doubling dilution scale (i.e. 2, 4 or 8 mg/L) for calculations.

‘Essential agreement’ was defined as the percentage of cases where the MIC result by the test method was within one doubling dilution of that by the BSAC reference method among cases where both values were on-scale.

‘Potential agreement’ was defined as the percentage of cases where the MIC result by the test method was within one doubling dilution of the reference method or was off-scale by one method and potentially in agreement with the reference method (e.g. >32 mg/L by one method and 256 mg/L by the other). Either essential or potential agreement could appear higher because the denominators and the numerators both varied between the two measurements. Correlation between MICs by different methods was calculated using log2 MICs, with off-scale values excluded. Comparisons of performance were by χ2 test.

Results and discussion

Performance of M.I.C.Evaluator and Etest Strips versus BSAC agar dilution

A total of 1017 isolates and 18 strip types were tested, with 77 antibiotic/organism combinations represented. Essential agreement was achieved for 89.9% of tests with the M.I.C.Evaluator versus 89.5% for the Etest (P > 0.05) (Tables 2 and 3); potential agreement was achieved in 89.5% for the M.I.C.Evaluator versus 89.3% for the Etest (P > 0.05) (Tables 4 and 5). The proportions of strip-type/organism combinations for which various target levels of agreement with the agar dilution reference (≥95%, ≥90% ≥85% etc.) was achieved did not differ significantly between the two strip types (P > 0.05) irrespective of the target criterion selected (Table 6).

Table 2

‘Essential agreement’ (%) between M.I.C.Evaluator (M.I.C.E) or Etest and agar dilution (±1 doubling dilution*), counting only on-scale values: non-fastidious species

	Enterobacteriaceae		Non-fermenters		Staphylococci		Enterococci
	M.I.C.E	Etest	M.I.C.E	Etest	M.I.C.E	Etest	M.I.C.E	Etest
Amoxicillin/clavulanic acid	86.9	91.3
Amoxicillin	87.3	88.7
Ampicillin	87.6	87.6					77.3	81.8
Cefotaxime 32	92.7	87.6			98*	83*
Cefotaxime 256	91.1	85.8			80.7*	99.1*
Ciprofloxacin	95.8	96.2	100	100	98.9	98.9
Erythromycin					93.7	98.9
Gentamicin 256	91.6	93.6	93.5	93.5	96	95.3
Gentamicin 1024	94.9	91.8	95.3	93.8	95.6	96.4	88.1	82.1
Imipenem	81.7	82.5	96	88			95.1a	82.9a
Levofloxacin	94.3	92.1			98	99
Linezolid					99.3	99.3	94.2	96.1
Oxacillin					83.2*	74.7*
Penicillin G 32					87.2*	80.2*
Penicillin G 256					83.9*	83*
Tetracycline					98.7	98.7
Vancomycin					95.6	96.2	93.3	93.3
Overall	90.4	89.7	96.2	93.8	93.0	92.5	89.6	87.2
Excluding asterisked values					97.0	97.8

aOnly E. faecalis tested.

Table 3

‘Essential agreement’ (%) between M.I.C.Evaluator (M.I.C.E) or Etest and agar dilution (±1 doubling dilution), counting only on-scale values: fastidious species

	Streptococci		Neisseria		Haemophilus		Anaerobes		Moraxella		Listeria		Pasteurella		Campylobacter
	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	M.I.C.E	Etest	M.I.C.E	Etest	M.I.C.E	Etest
Amoxicillin/clavulanic acid					73.8	65	88	91.7	100	100
Amoxicillin	92.7	84.1			59.4	42.4
Ampicillin	94.2	94.2	83.9	85.7	73.8	60.5					95.2	100
Cefotaxime 32	92	88.3	78.2	78.2	83.7	55.8
Cefotaxime 256	88.3	82.1	57.1	71.4	68.8	85.7
Ciprofloxacin	96.2	95.5	98	98	59.5	88									100	100
Erythromycin	60.9	66.2			97.6	95.2					100	95.4			90.9	90.9
Gentamicin 256	88.9	93.2
Gentamicin 1024	81.6	94.4
Imipenem	95	65			54.8	58	46.5	48.8
Levofloxacin	97.5	96.9			100	97.7
Linezolid	98.1	98.7
Oxacillin	89.5	71.1
Penicillin G 32	61.9	68	72.7	89			88	89.1					100	100
Penicillin G 256	63.8	74.3	80.3	80.3			92	90.9					90	90
Tetracycline	76.6	76.6	98.2	96.4	97.7	95.4
Vancomycin	73	77.9
Metronidazole							84.1	81.3
Overall	84.4	82.9	81.2	85.6	76.9	74.4	79.7	80.4	100	100	97.6	97.7	95.0	95.0	95.5	95.5

Table 4

‘Potential agreement’ (%) between M.I.C.Evaluator (M.I.C.E) or Etest Strips and agar dilution (±1 doubling dilution*), counting all potential agreements as agreement: non-fastidious species

	Enterobacteriaceae		Non-fermenters		Staphylococci		Enterococci
	M.I.C.E	Etest	M.I.C.E	Etest	M.I.C.E	Etest	M.I.C.E	Etest
Amoxicillin/clavulanic acid	78.3	83.7
Amoxicillin	91.1	91.1
Ampicillin	89.1	89.1					70.5	79.4
Cefotaxime 32	89.9	84.1			94.3*	82.3*
Cefotaxime 256	89.1	84.8			77.9*	94.9*
Ciprofloxacin	95.7	94.1	98.7	98.7	95.6	95.6
Erythromycin					98.7	99.3
Gentamicin 256	91.4	93	94.8	93.5	96.2	94.3
Gentamicin 1024	94.9	91.8	94.8	94.8	95.5	96.2	87.2	83.3
Imipenem	81.7	82.5	85.8	73			95.1b	82.9b
Levofloxacin	88.3	84.5			75a	75.6a
Linezolid					99.3	99.3	94.2	96.1
Oxacillin					82.5*	76.8*
Penicillin G 32					86.7*	81.7*
Penicillin G 256					79.2*	77.9*
Tetracycline					98.1	98.7
Vancomycin					95.6	96.2	93.1	93.1
Overall	89.0	87.9	93.5	90.0	90.4	89.9	88.0	86.7
Excluding asterisked values					94.3	94.4

aPoorer potential than essential agreement (see Table 2) for levofloxacin versus staphylococci reflects 25 isolates with MICs of 16 mg/L by agar dilution, but >32 mg/L by both strip methods.

bOnly E. faecalis tested.

Table 5

‘Potential agreement’ (%) between M.I.C.Evaluator (M.I.C.E) or Etest Strips and agar dilution (±1 doubling dilution), counting all potential agreements as agreement: fastidious species

	Streptococci		Neisseria		Haemophilus		Anaerobes		Moraxella		Listeria		Pasteurella		Campylobacter
	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	M.I.C.E	Etest	M.I.C.E	Etest	M.I.C.E	Etest
Amoxicillin/clavulanic acid					72	62.7	88.1	91.8	100	100
Amoxicillin	93.2	85.2			56.8	43
Ampicillin	94.4	94.4	83.9	85.7	72.1	55.8					95.4	100
Cefotaxime 32	92	88.3	82.1	82.1	83.7	55.8
Cefotaxime 256	92	88.3	82.1	87.5	76.7	88.3
Ciprofloxacin	94.4	93.2	94.6	94.6	58.1	86									90.9	100
Erythromycin	63.1	68			97.6	93					100	95.4			90.9	90.9
Gentamicin 256	88.9	93.2
Gentamicin 1024	81.6	94.4
Imipenem	95	65			46.5	48.8	69	69
Levofloxacin	97.5	96.9			97.7	97.7
Linezolid	98.1	98.7
Oxacillin	89.5	71.1
Penicillin G 32	61.9	68	73.2	89.2			84.5	89					100	100
Penicillin G 256	63.8	74.8	80.3	80.3			91.8	90					90.9	90.9
Tetracycline	76.6	76.6	98.2	96.4	97.7	95.4
Vancomycin	73	77.9
Metronidazole							84.5	81.8
Overall	84.7	83.4	84.9	88.0	75.9	72.7	83.6	84.3	100	100	97.7	97.7	95.5	95.5	90.9	95.5

Table 6

Agreement of MICs by strip methods with BSAC agar dilution, based on 77 strip/organism combinations

	No. (out of 77 combinations) with essential agreement			No. (out of 77 combinations) with potential agreement
	M.I.C.Evaluator	Etest	P	M.I.C.Evaluator	Etest	P
≥95%	28	25	>0.05	21	18	>0.05
≥90%	42	39	>0.05	39	38	>0.05
≥85%	52	50	>0.05	49	48	>0.05
≥80%	61	61	>0.05	59	59	>0.05
≥75%	64	64	>0.05	65	66	>0.05
≥70%	68	68	>0.05	70	68	>0.05

It should be added that these ‘global’ agreement rates included antibiotic/organism combinations, e.g. cefotaxime and oxacillin against MRSA and benzylpenicillin against penicillinase-producing S. aureus, where MICs are notoriously fickle and would not ordinarily be determined. Agreement in these cases was poorer than for most other agents tested against staphylococci, and global agreement rates rose if they were excluded from the analysis (Tables 2 and 4).

Even for antibiotic/organism combinations where essential agreement was <90%, the results of the strip tests mostly correlated well with BSAC agar dilution (Tables 7 and 8). Exceptions, with poorer correlation (r ≤ 0.7), were: (i) imipenem against Enterobacteriaceae, where a distortion arose owing to the behaviour of Proteeae, as described below; (ii) cases where high-drug-content strips were tested against species groups where low-content strips were more appropriate, as with cefotaxime against Neisseria spp.; and (iii)—most especially—(Tables 7 and 8), cases where MICs by the reference method were clustered over four or fewer drug dilutions, i.e. over a scarcely wider collective range than the ±1 doubling dilution range conventionally accepted as experimental variation when the MIC for a single organism is repeatedly determined by the same method.

Table 7

Correlation coefficients between MICs determined by strip methods and those found by BSAC agar dilution: non-fastidious species

	Enterobacteriaceae		Non-fermenters		Staphylococci		Enterococci
	M.I.C.E	Etest	M.I.C.E	Etest	M.I.C.E	Etest	M.I.C.E	Etest
Amoxicillin/clavulanic acid	0.949	0.891
Amoxicillin	0.927	0.94
Ampicillin	0.906	0.898					0.962	0.921
Cefotaxime 32	0.951	0.939			0.906	0.913
Cefotaxime 256	0.962	0.957			1	1
Ciprofloxacin	0.924	0.916	0.972	0.971	0.871	0.888
Erythromycin					0.906	0.911
Gentamicin 256	0.939	0.916	0.946	0.944	1	1
Gentamicin 1024	0.957	0.949	0.964	0.961	0.977	0.979	0.933	0.911
Imipenem	0.562a	0.427a	0.951	0.922			0.553	0.446
Levofloxacin	0.951	0.951			0.95	0.95
Linezolid					0.9	0.9	0.947	0.941
Oxacillin					0.907	0.906
Penicillin G 32					0.9	0.9
Penicillin G 256					0.958	0.958
Tetracycline					0.962	0.975
Vancomycin					0.655b	0.7	0.955	0.969
Overall	0.903	0.878	0.958	0.949	0.915	0.921	0.870	0.837

M.I.C.E, M.I.C.Evaluator.

aMost of the less-susceptible organisms were Proteeae, for which there was poor correlation between BSAC and strip methods.

bPoor correlation (≤ 0.7) explained by > 90% agar dilution MICs being spread over ≤ 4 dilutions.

Table 8

Correlation coefficients between MICs determined by strip methods and those found by BSAC agar dilution: fastidious species

	Streptococci		Neisseria		Haemophilus		Anaerobes		Moraxella		Listeria		Pasteurella		Campylobacter
	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	M.I.C.E	Etest	M.I.C.E	Etest	M.I.C.E	Etest
Amoxicillin/clavulanic acid					0.7	0.6	0.9	0.9	0.9	0.939
Amoxicillin	0.926	0.915			0.762	0.653
Ampicillin	0.937	0.942	0.9	0.9	0.895	0.853					0.132a	0.311a
Cefotaxime 32	0.9	0.9	0.655	0.717	0.9	0.9
Cefotaxime 256	0.9	0.9	0.5	0.6	0.8	0.9
Ciprofloxacin	0.78	0.757	0.920	0.939	0.6a	0.6a									0.931	0.95
Erythromycin	0.763	0.838			0.9	0.9					0.445a	0.488a			0.681a	0.677a
Gentamicin 256	0.707	0.719
Gentamicin 1024	0.703	0.763
Imipenem	0.989	0.991			0.7a	0.6a	0.793	0.724
Levofloxacin	0.522a	0.47a			0.9	0.9
Linezolid	0.656a	0.654a
Oxacillin	0.916	0.934
Penicillin G 32	0.9	0.9	0.917	0.916			0.946	0.939					0.936	0.948
Penicillin G 256	0.9	0.9	0.930	0.938			0.946	0.95					0.798	0.867
Tetracycline	0.919	0.958	0.9	0.9	0.949	0.928
Vancomycin	0.285a	0.386a
Metronidazole							0.56	0.5
Overall	0.8	0.808	0.817	0.844	0.811	0.763	0.829	0.803	0.9	0.939	0.288	0.399	0.867	0.907	0.806	0.813

M.I.C.E, M.I.C.Evaluator.

aPoor correlation (≤0.7) explained by >90% agar dilution MICs being spread over ≤4 dilutions.

In general, agreement between the two strip tests and the BSAC reference method was best for non-fastidious organisms, including staphylococci, enterococci, Enterobacteriaceae and non-fermenters, and for members of the genera Moraxella, Listeria, Pasteurella and Campylobacter (Tables 2–5). For these organisms, and taking all antibiotics combined, there was >89% essential agreement with the BSAC reference for the M.I.C.Evaluator Strips and >87% for Etest Strips. Essential agreement was less good, at between 83% and 88% for all antibiotics combined, for anaerobes, Neisseria spp. and Streptococcus spp. and poorest, at 72.7%–75.9%, for H. influenzae. Agreement rates for H. influenzae were lowered by imipenem (which would rarely be tested against the species) and, more importantly, by amoxicillin and ampicillin. Many of the underlying disagreements with these latter drugs were for β-lactamase-positive isolates found highly resistant by all methods, but with substantially higher MICs by one method than another (e.g. agar dilution amoxicillin MIC 256 mg/L, but only 32 mg/L by the strip methods). MICs are rarely determined for such unequivocal isolates in clinical practice. Among amoxicillin-borderline H. influenzae isolates—where strip tests are more likely to be used—four of seven with amoxicillin MICs of 1 mg/L by agar dilution (i.e. just susceptible) were found susceptible using both strip types, with MICs of 0.5–1 mg/L, whereas three proved resistant, with MICs of 1.5–2 mg/L; among five with agar dilution MICs of 2–4 mg/L (i.e. just resistant), one was found susceptible to amoxicillin with both strip types, with MICs of 0.5–1 mg/L, one was resistant, with MICs of 8 mg/L (Etest) and 16 mg/L (M.I.C.Evaluator) and three gave mixed results, with MICs of 1 mg/L with one strip and 1.5 mg/L with the other. Tristram[6] similarly tested M.I.C. Evaluator ampicillin strips against H. influenzae, though taking CLSI broth microdilution as a reference, and found 90% agreement (±1 doubling dilution) for ampicillin-susceptible isolates and those with β-lactamase-negative ampicillin resistance, versus only 65%–75% for those with β-lactamase.

Tests with penicillin against streptococci also deserve comment, being another case where gradient MIC test strips are particularly useful—e.g. for pneumococci with borderline resistance and for endocarditis isolates—but where agreement appeared rather low, at 62%–68%. These figures, however, exaggerate the ‘disagreements’, which largely concerned β-haemolytic streptococci with penicillin MICs of ≤0.008 mg/L by agar dilution and 0.023 (rounded to 0.03) mg/L by the strip methods—a difference of no practical consequence. Penicillin MICs by Etest and M.I.C.Evaluator Strips for pneumococci with agar dilution MICs from 0.12 to 2 mg/L are shown in Table 9, indicating perfect essential agreement for the two high-content strips, though rather poorer for the low-content M.I.C.Evaluator Strip.

Table 9

Strip MICs for pneumococci with agar dilution penicillin MICs of 0.125–2 mg/L

	Agar	Etest		M.I.C.Evaluator
	penicillin G	penicillin G 32	penicillin G 256	penicillin G 32	penicillin G 256
H051360048	0.125	0.125	0.125	0.125	0.125
H014200026	0.25	0.25	0.38	0.38	0.38
H051380053	1	1.5	1.5	1.5	1
H051440589	1	2	1.5	1.5	1.5
H051460137	1	0.75	0.75	2	1
H051380039	1	0.75	1	1	0.75
H050300113	1	1.5	1.5	1	1
PN2940	1	0.75	0.75	1	1
H040440029	1	3a	2	4a	2
H051440587	1	1	1.5	2	1
H050720467	1	2	1.5	2	1.5
H059200073	1	1	1.5	3a	1.5
H051240370	1	2	1.5	3a	1.5
H050420065	1	1	1.5	1.5	1.5
H045200156	1	1.5	1.5	3a	0.5
PN2177	2	3	3	4a	3a
H040440026	2	1	1.5	1.5	1
H050480159	2	2	2	3a	2
PN1847	2	1.5	2	16a	3a
H051240093	2	2	1.5	3a	1.5
H050440059	2	2	2	3a	2
H045240059	2	2	1.5	2	1.5
Essential agreement with reference (%)	–	95.5	100	77.3	100

Bold font indicates MIC disagreement of >1 doubling dilution.

aCategorization difference versus breakpoints of: susceptible, ≤0.06 mg/L; and resistant, >2 mg/L.

Agreement rates for Enterobacteriaceae were poorer with imipenem than with other strip types. This was largely owing to the strips giving ∼8-fold lower imipenem MICs for Proteus spp. Agreement was far better for other species, as shown by the geometric mean MIC values (Table 10); these are a valid summary parameter since the MICs were normally distributed and unimodal within each species. It is possible that higher MICs by agar dilution for Proteeae reflected the inclusion of p-nitrophenyl glycerol to prevent swarming, but we can find no published assertion that this compound interferes with the activity of imipenem; moreover, broth dilution MICs, e.g. by the CLSI method, are similarly high for Proteus spp. with no anti-swarming agent present.[7] An alternative explanation, perhaps more likely, is simply that in seeking to ignore swarming into the zone of inhibition, the recorder tends to underestimate the real MICs.

Table 10

Geometric mean MICs (mg/L) of imipenem for Enterobacteriaceae genera

	Agar dilution	Etest	M.I.C.Evaluator
Citrobacter spp.	0.44	0.32	0.23
Enterobacter spp.	0.42	0.61	0.36
E. coli	0.15	0.23	0.15
Klebsiella spp.	0.40	0.34	0.23
Proteus spp.	1.73	0.27	0.29
Salmonella spp.	0.37	0.31	0.22

Other specific concerns could be identified, e.g. there were 19 enterococci with gentamicin MICs of 256–512 mg/L, counting as high-level resistant by BSAC criteria, but nine of these gave MICs ≤128 mg/L by Etest, counting as susceptible, and five did so by M.I.C.Evaluator. Only 2/34 enterococci with gentamicin MICs of ≥1024 mg/L gave MICs ≤128 mg/L by the strip methods, each of them with both products, whilst 2/51 enterococci with a gentamicin MIC ≤128 mg/L by agar dilution appeared resistant, with MICs ≥256 mg/L in strip tests, one with the Etest only and one with both products.

Agreement between M.I.C.Evaluator and Etest Strip results

Concordance between the M.I.C.Evaluator Strip and Etest Strip results was excellent, with >95% potential agreement (±1 dilution) for 54 of 77 strip/organism combinations and 90%–95% for another 12 (Table 11). These proportions are significantly better than between either strip type and agar dilution (P < 0.01, χ2 test; compare Table 6). The only antibiotic/organism combinations with <85% agreement between the two strips were amoxicillin/clavulanic acid against both Enterobacteriaceae (70.6%) and Haemophilus spp. (81.3%) and both ampicillin and amoxicillin against Haemophilus spp. (67.4% and 68.1%, respectively). In the case of ampicillin and amoxicillin, virtually all the underlying disagreements related to strains that were resistant by all three methods, while most (43/75) >1 dilution disagreements for amoxicillin/clavulanic acid against Enterobacteriaceae related to Enterobacter and Citrobacter spp., both of which are inherently resistant to the drug combination.

Table 11

Potential agreement (±1 doubling dilution, off-scale values counted as agreeing) between M.I.C.Evaluator and Etest results

	Enterobacteriaceae	Non-fermenters	Staphylococci	Enterococci	Streptococci	Neisseria	Haemophilus	Anaerobes	Moraxella	Listeria	Pasteurella	Campylobacter
Amoxicillin/clavulanic acid	70.6						81.3	94.5	100
Amoxicillin	98.8				96.9		68.1
Ampicillin	96.1			98	96.9	98.2	67.4			100
Cefotaxime 32	95.3		94.9		98.1	100	97.6
Cefotaxime 256	96.8		86.7		98.1	98.2	86
Ciprofloxacin	97.2	98.7	98.7		99.3	96.4	95.3					95.4
Erythromycin			93.7		93.8		97.6			90.9		86.3
Gentamicin 256	96.1	97.4	98.1		97.5
Gentamicin 1024	98.8	97.4	97.5	99	98.1
Imipenem	96.8	89.7		100	92.6		86	90.9
Levofloxacin	99.6		100		98.1		97.7
Linezolid			98.1	99	96.9
Oxacillin			91.8		94.4
Penicillin G 32			87.4		95.7	100		87.2			90.9
Penicillin G 256			96.8		96.3	100		97.2			100
Tetracycline			94.3		93.2	100	100
Vancomycin			99.3	97	97.5
Metronidazole								98.1
Overall	94.6	95.8	95.2	98.6	96.5	99.0	87.7	93.6	100	95.5	95.5	90.9

Conclusions

Etests are in widespread use for custom MIC determinations in diagnostic laboratories. They have been calibrated to give equivalent MICs to those found by CLSI methodology using Mueller–Hinton agar, and there is a voluminous literature to support their accuracy when used in this manner.[3] The manufacturer (AB Biodisk at the time of this study, now bioMérieux) does not advocate their use on other media, and there are no substantial performance studies on Iso-Sensitest agar, which is the standard medium for the BSAC dilution method. The BSAC nevertheless states that, in general, Etests can be used on Iso-Sensitest agar, with BSAC/EUCAST breakpoints, so long as inocula equivalent to a 0.5 McFarland are used.[4]

The M.I.C.Evaluator Strip has been developed as an alternative strip-based MIC method, which the manufacturer indicates to be suitable for use on either Mueller–Hinton or Iso-Sensitest agar. As with the Etest Strip, it comprises a laminated plastic support carrying a double series of antibiotic-impregnated droplets of diminishing content. Once placed on agar, the strip rapidly releases the antibiotic, delivering a stable gradient. After incubation, the MIC can be read off against a printed scale. The present data show that, used on Iso-Sensitest agar, both M.I.C.Evaluator and Etest Strips gave essentially equivalent results to one another and acceptable agreement with BSAC agar dilution.

Funding

The work was supported by Thermo Fisher Scientific, Basingstoke, UK.

Transparency declarations

D. M. L. has received conference support from Thermo Fisher Scientific, bioMérieux and AB Biodisk, he has received grants and conference support from numerous pharmaceutical companies, he holds shares in AstraZeneca, Merck, Pfizer, Dechra and GlaxoSmithKline and, as Enduring Attorney, managed further holdings in GlaxoSmithKline and Eco Animal Health. D. M. L. is employed by the HPA and sits on the BSAC's Working Party on Susceptibility Testing, and is influenced by their views. J. C. is an employee of Thermo Fisher Scientific. Other authors: none to declare.