Antibiotics in Laboratory Medicine, 6 Ed.

Chapter 2. Antimicrobial Susceptibility on Solid Media

John D. Turnidge and Jan M. Bell

Susceptibility testing on solid media is a widely used alternative to the traditional broth-based testing developed originally to measure minimum inhibitory concentrations. With rare exceptions, it relies on the use of agar as the solidifying agent and the nature of agar, which permits slow diffusion of chemicals through its three-dimensional matrix. There are three formats for testing using solid media: agar dilution, disk diffusion, and gradient diffusion. Of these, the one that has proven most popular and adaptable to routine laboratory testing is disk diffusion. It has been the subject of excellent chapters in previous editions of this book (1,2). The first edition of this book (3) contains an excellent chapter on some of the basic issues with testing in agar that are not fully explored here.

Of note, this chapter addresses the use of agar in susceptibility testing of conventional bacteria of human and animal origin and does not address susceptibility testing of mycobacteria, nocardiae or other aerobic actinomycetes, mycoplasmas, yeast, or molds. Methods for testing such organisms using agar have been described, and readers are referred to the standards from the Clinical and Laboratory Standards Institute (CLSI) (4–7).

Solidifying agents for susceptibility testing and culture media generally need the following features to be useful: (a) water solubility, (b) ability to remain solid at incubation temperatures (≤42°C), (c) ability to liquefy at higher temperatures to permit pouring and incorporation of additives, (d) chemical inertness, (e) relative transparency, and (f) resistance to bacterial degradative enzymes.

Agar

Agar is a natural product obtained from several red seaweeds from the Rhodophyceae class and related seaweeds such as Pterocladia, which collectively are called agarophytes. It takes its name from the Malay word agar-agar, which describes these seaweeds. Agar also has wide application in the food industry as a thickening and emulsifying agent. Agar liquefies when heated to boiling and does not gel until cooled to 45°C to 50°C. After gelling, it requires reheating to near boiling to liquefy again. For the preparation of bacteriologic grade agar, Gelidium species are almost always used because they have the preferred lower gelling temperature of 34°C to 36°C, allowing the addition of supplements (8,9).

The principal components of agar are the polysaccharides agarose and agaropectin. Purified agarose has become one of the mainstays of solid-phase electrophoretic analysis. As a natural product, agar is subject to lot-to-lot variation. Variation occurs in the presence of sulfate ions, which affect the overall negative charge of the polysaccharides, which in turn can affect diffusion of certain chemicals. Calcium is also essential in small amounts to permit gelling. Of greater importance is brand-to-brand and lot-to-lot variation in the concentrations of divalent cations, which can affect the activity of certain antimicrobials, as discussed in the following section.

Other Solidifying Agents

A range of other gelling agents have been experimented with over the years, but none has yet displaced agar for susceptibility tests. In part, this relates not only to cost but also to the daunting challenge of recalibrating end points (minimum inhibitory concentrations [MICs], zone diameters) for a very broad range of bacteria and drugs. Substances that have been examined include Separan NP10 (Dow Chemical Co, Midland, MI), a polyacrylamide that allows smaller concentrations of agar to be used; Gelrite (Merck & Co, Kelco Division, Rahway, NJ), a gellan gum formed from the fermentation products of a Pseudomonas species; and Neutra-Gel (Union Carbide Corp, Tuxedo, NY), a polyoxyethylene polymer that held the greatest promise when combined with a synthetic amino acid medium. Gelrite (Merck & Co, Kelco Division, Rahway, NJ) is the only one still available commercially.

Since the development and promulgation of the European Committee of Antimicrobial Susceptibility Testing (EUCAST) disk diffusion methods (10), the agar main medium now used for conventional bacterial susceptibility testing is Mueller-Hinton. Sensitest and Iso-Sensitest agars are still used by minor susceptibility testing methods in Australia (11) and the United Kingdom (12). Mueller-Hinton agar has become the de facto standard agar medium in large part because it was selected for disk susceptibility testing when it was first standardized in the United States by Bauer et al. (13). Curiously, the medium was originally developed for the cultivation of Neisseria species. A large number of criticisms have been leveled at Mueller-Hinton over the years. Its problems include the possibility of different MIC values in broth versus the agar version, antagonism of tetracyclines, high levels of folate synthesis inhibitor antagonists, variation in performance between manufacturers due to difference in peptone sources, poor support for streptococcal species, and variable growth rates with gram-positive bacteria generally (14). Most of these problems were overcome through the intensive efforts of investigators and manufacturers toward its standardization, including intensive quality control procedures (15) and subsequently the development of “golden pound” reference lots (16). Currently, a new International Standards Organization (ISO) standard is being prepared which permits the calibration of Mueller-Hinton agar and broth based on performance rather than by comparison to a reference golden pound (17).

Iso-Sensitest agar was a common choice in some European countries until supplemented by EUCAST methods (18). Iso-Sensitest was developed by Oxoid from the original diagnostic sensitivity test agar and then Sensitest agar and is designed to minimize the amount of variable nutrients and maximize the defined components. To a lesser extent, problems have been encountered over the years in the performance of Iso-Sensitest (19). Nevertheless, it remains the basis of the British Society of Antimicrobial Chemotherapy disk diffusion test (12). Paper disk method (PDM) antibiotic sensitivity medium, long recommended (along with Iso-Sensitest) by the Swedish Reference Group on Antibiotics (20), has been withdrawn from the market.

The medium recommended for anaerobe susceptibility testing is Brucella agar or Wilkins-Chalgren (21,22). Less attention has been paid overall to the suitability of different media for anaerobes. Studies by a number of investigators have shown that supplemented Brucella agar supports the growth of a wider range of anaerobes than other media (23–25) and is now recommended as the reference medium by CLSI (22).

There are considerable differences in the formulas of different agars (Table 2.1).

Antagonists of Folate Synthesis Inhibitors

Paraaminobenzoic acid (p-ABA) is a potent inhibitor of sulfonamides. Concentrations found in certain media such as peptone water and nutrient agar will virtually eliminate sulfonamide activity (26). Susceptibility testing agar used in any of the current published methods has minimal concentrations of p-ABA.

Thymidine and thymine in sufficient concentrations antagonize the dihydrofolate reductase (DHFR) inhibitors such as trimethoprim, increasing MICs and reducing zone diameters in agar diffusion tests. This nucleoside and its pyrimidine base possibly act by competing for the target enzyme. Thymidine is by far the more potent of the two, and methods to reduce or eliminate its presence in agar will restore DHFR inhibitor activity. Indeed, one common method is to use lysed horse blood, which is rich in the enzyme thymidine phosphorylase, which converts thymidine to thymine and 2-deoxyribose-1-phosphate. Further, most bacteria, except Enterococcus faecalis, cannot utilize thymine as a substrate (26). Some bacterial strains require thymidine for growth and will grow poorly or not at all on susceptibility testing agar. Because such strains are naturally DHFR inhibitor resistant, no problems of testing occur when thymidine is added back into the medium.

Although Mueller-Hinton agar is purportedly low in thymidine and thymine, occasionally, problems can arise. CLSI has developed a quality control procedure to test for low thymidine content using either of two American Type Culture Collection (ATCC) strains of E. faecalis and trimethoprim-sulfamethoxazole disks (27). This procedure is now embraced in the new ISO standard to Mueller-Hinton lot acceptance criteria (17).

Calcium, Magnesium, Zinc, and Manganese

The principal problem with Mueller-Hinton over the years has been the variability in the concentrations of divalent cations, especially calcium (Ca⁺⁺) and magnesium (Mg⁺⁺), which can have significant effects on the activity of aminoglycosides and some other antimicrobials, particularly against Pseudomonas aeruginosa (28,29). For the broth, this has now been overcome during the manufacturing process, and the concentrations are adjusted to within an acceptable range specified by ISO 16782 (17). However, the addition of agar to Mueller-Hinton will alter the cation concentration in unpredictable ways, such that for some newer antimicrobial agents heavily influenced by cation concentration, such as Ca⁺⁺ with daptomycin (30) and manganese (Mn⁺⁺) with tigecycline (31,32), it has not been possible to develop agar dilution or diffusion standards. The great variability in cation concentrations with different brands of Mueller-Hinton agar has recently been highlighted (31,33).

The divalent cations of Ca⁺⁺ and Mg⁺⁺ are well-known antagonists of aminoglycosides. The antagonism is complex and cannot simply be accounted for by the concentrations of the cations themselves (34). It appears to be affected to a large extent by other constituents such as sodium chloride and phosphate. The effects are most obvious when testing P. aeruginosa (35–37). The variability of concentrations in Mueller-Hinton agar in the past has been one reason that certain European countries have favored Iso-Sensitest agar, where the divalent cation concentrations are defined in the formulation. Ostensibly, agar itself is processed to remove free cations and anions (14), but recent studies have demonstrated that there can still be a wide range of calcium and magnesium concentrations in different batches of Mueller-Hinton agar (30,31,33). In contrast to Mueller-Hinton broth, the concentration of divalent cations in Mueller-Hinton is not stipulated in the ISO 16782 standard. However, other susceptibility test media have not been subject to control of their divalent cations, and the reproducibility of aminoglycoside results with different lots has not been examined.

The concentration of Ca⁺⁺ is critical to the interpretation of daptomycin susceptibility (30). Daptomycin activity varies greatly with Ca⁺⁺ concentration, and specified concentrations are required in the media (usually those seen physiologically). For this reason, daptomycin has not yet been completely standardized for tests in agar, and supplementation is required for broth testing (30). Cation concentrations are also known to affect the activity of the polymyxins in susceptibility tests at least for P. aeruginosa and Acinetobacter baumannii (33,35). Calcium and magnesium concentrations can also affect the action of tetracycline against Pseudomonas species, although this is of little importance because tetracyclines are not considered clinically active against these species.

The concentrations of another cation, zinc (Zn⁺⁺), in Mueller-Hinton (38–40) agar and Iso-Sensitest agar (41) has an impact on the activity of imipenem and possibly other carbapenems, at least for common nonfermentative gram-negative. Again, the concentrations of Zn⁺⁺ can be quite variable in different brands of Mueller-Hinton agar (32,33).

A summary of two recent publications that have examined cation concentrations in Mueller-Hinton agar is presented in Table 2.2.

Sodium Chloride

Less well known is the effect of sodium chloride concentration on the activity of aminoglycosides (42,43). As summarized by Waterworth (34), variations in NaCl can have quite significant effects: an increase in concentration from 22 to 174 mM can increase the MIC of gentamicin by as much as 32-fold. NaCl is not part of the Mueller-Hinton formulation, but the manufacture to a reference standard at least generates consistent results. NaCl is part of the Iso-Sensitest formulation, but viable amounts could also come from the hydrolyzed casein and peptones.

pH

Major variation in the pH of the medium can result in major changes in the activity of aminoglycosides, macrolides, and tetracyclines. Aminoglycoside activity is substantially increased in alkaline conditions and substantially inhibited in acidic conditions. Similar effects are observed with macrolides. Susceptibility testing agars are manufactured to performance standards of pH, and in some testing methods, such as those of CLSI, it is recommended to confirm the pH in the cooled, prepoured state or with a surface pH meter after pouring, once the agar has been prepared from the dried powder in the routine laboratory.

The pH effect does become significant, however, when agar plates are incubated in increased concentrations of CO₂; such is recommended for streptococci and Haemophilus species in most methods. Rosenblatt and Schoenknecht (44) have shown an increase in pH from 7.4 to 8.4 over a period of 24 hours when Mueller-Hinton blood agar plates are incubated in 5% to 7% CO₂ in air. Carbon dioxide is absorbed onto the surface during incubation, some of which will be converted to carbonic acid initially and then carbonate ions, first decreasing and later increasing the pH at the surface (44). Acidity is known to reduce the activity of macrolides in particular (including the azalides and ketolides) (45–55) and of aminoglycosides to some extent (56,57) while increasing the activity of tetracyclines (2). In agar-based tests, this effect will result in higher MICs and smaller zones for macrolides and aminoglycosides, as the pH at the surface will be more acidic at the critical time.

Additives

Some bacterial species require the addition of specific nutrients to ensure adequate growth. The most common of these is blood, usually sheep or horse blood, at a concentration of 5%. When testing sulfonamides, horse blood is preferred, as it is low in sulfonamide antagonists.

Specific reagent additives are required for certain fastidious species or for nonfastidious species against certain antimicrobial agents, as described in Table 2.3.

Defined Growth Supplement

One important additive, noted here and in CLSI documents as defined growth supplement, is a complex mixture of vitamins, cofactors, and other nutrient substances. The two most recognizable and important brands are IsoVitaleX (Becton, Dickinson and Company, Sparks, MD) and Vitox (Thermo Scientific, West Palm Beach, FL). These products contain 1.1 g L-cystine, 0.03 g guanine, 3 mg thiamine HCl, 13 mg p-ABA, 0.001 to 0.012 g vitamin B₁₂, 0.1 g cocarboxylase, 0.25 g nicotinamide adenine dinucleotide (NAD), 1 g adenine, 10 g L-glutamine, 100 g glucose, 0.02 g ferric nitrate, and 25.9 mg cysteine HCl per liter of water. It is most commonly used at a 1% concentration to ensure the growth of the fastidious organisms, especially Neisseria gonorrhoeae.

Paranitrophenylglycerol (PNPG) has been used among other techniques to prevent the swarming of Proteus mirabilis and Proteus vulgaris across agar surfaces. This is mostly a problem for agar dilution testing when multiple strains including strains of these two species are being tested on a single series of plates. As a result of the swarming, spots close to the Proteus species can be difficult or impossible to read. However, PNPG has been shown to affect the MIC results of a number of bacterial species and antimicrobials, especially Pseudomonas aeruginosa (58,59).

A second antiswarming agent, Dispersol LN, has also been evaluated and shown to affect the activity of some antimicrobials (60). The addition of PNPG or another antiswarming agent is no longer recommended routinely in any method and should only be used if there are data to show that the agent does not interfere with the MICs of that organism–antimicrobial combination. Another natural chemical, 10′(Z),13′(E)-heptadecadienylhydroquinone, has been shown to increase susceptibility of P. mirabilis to polymyxin B (61).

Higher concentrations of agar (e.g., 2%) can be used to inhibit swarming. None of the three options—use of PNPG, use of Dispersol LN, or increasing the concentration of agar—is now recommended.

Strains with Special Growth Requirements

Special problems arise with strains or species with unusual growth requirements. The so-called nutritionally variant “streptococci” Abiotrophia defectiva (previously Streptococcus defectivus) and Granulicatella adiacens (previously Streptococcus adjacens, Abiotrophia adiacens) require pyridoxal for growth. Media can be supplemented with pyridoxal (0.001%) and lysed horse blood to ensure growth of the strains (63–65) and allow interpretation of MICs at least (no breakpoints have been determined).

Occasionally, mutant strains of Staphylococcus aureus will depend on thiamine or menadione (vitamin K₃) for growth. These often exhibit small colonies on primary isolation. The addition of thiamine (2 mg/L) and menadione (0.5 mg/L) to susceptibility testing media will allow susceptibility (MIC) tests to be performed (66).

Strains of Escherichia coli and other gram-negatives dependent on thymine for growth are sometimes encountered. They can be selected for during treatment with folate synthesis inhibitors (67). Strains of Enterobacteriaceae dependent for growth on thymidine, cysteine, or glutamine may be tested by adding the appropriate nutrient to the basal medium (7), although experience with this is limited due to the infrequency with which these strains are isolated.

Agar dilution susceptibility testing is the solid equivalent of broth dilution susceptibility testing, either in macro- or microbroth format. One advantage it offers over broth-based methods is that it allows the simultaneous testing of a large number of strains on a single agar plate, for example, 32 on a 90-mm plate using a Steers-Foltz or similar replicator (Fig. 2.1). It is therefore well suited to the rapid evaluation of new compounds or for large-scale centralized surveillance programs. The comparative disadvantages of this method are that it includes an additional variable (agar) in the medium and that, once prepared, the plates have a limited shelf life owing to degradation of the antimicrobial.

Most important, although agar dilution susceptibility testing, such as that described by CLSI (62) and EUCAST (18), has traditionally been accepted as equivalent to broth microdilution, it is not the international (ISO) reference standard, and users of agar dilution need to be aware that there may be differences from the ISO standard for some antimicrobial agent–microorganism combinations, particularly in the light of uncontrolled cation concentrations in Mueller-Hinton agar versus Mueller-Hinton broth as specified for susceptibility testing. In the future, it will be necessary to establish equivalence to the ISO standard for new antimicrobial agents as they are developed. Methods for establishing equivalence have been described (68).

The two major published agar dilution susceptibility testing methods for common human pathogens (18,62) are essentially equivalent. In the usual approach to determining MICs on agar, the antimicrobial is incorporated into molten agar over a series of twofold dilutions, gently rotated to ensure even distribution of the antimicrobial, and then poured into Petri plates. The major elements of the test are described in the following sections.

Antimicrobial Powders

Antimicrobial powders should be obtained as “pure substance” from the manufacturer or from a reputable chemical supplier (e.g., Sigma-Aldrich Corporation). It is not appropriate to use powders found in vials for parenteral drug administration, as these may contain preservatives, surfactants, fillers, or other substances that could interfere with the antimicrobial activity, and their contents may vary legally by as much as 10% above or below the label amount. Each powder should come with an expiration date and an indication of its potency and (sometimes) water content. It is vital that all these data be taken into account before preparation of stock solutions. Data should also be available from the manufacturer or another reliable source on choice of solvent and solubility before preparing stock solutions. One easily accessible source is Table 5A in CLSI’s M100 document, which is updated yearly (69).

Powders should be stored as recommended by the manufacturer. If no instructions for storage are available, then store powders at −20°C in a desiccator, preferably under vacuum (69). This will ensure the product retains its potency for the maximum time.

Choice of Dilution Range

Before preparing stock solutions, it is essential that an appropriate dilution range be chosen. Suggested ranges for different species or bacterial types have been published (21). A full range is generally 10 to 12 doubling dilutions. Doubling dilutions are appropriate because they provide the narrowest integer series on a logarithmic scale, and MICs for a single bug–drug combination are log-normally distributed in the wild type, that is, in the absence of a resistance mechanism. Although any series could be used, the most widely accepted doubling dilution series is that of base 2. This implies that preferably the highest concentration (the concentration at the start) should be a power of 2 (e.g., 2⁷ = 128). Following this pattern will allow comparison with the majority of published data and MICs for published quality control strains.

Preparation of Stock Solutions

Stock solutions should be prepared by weighing out the exact amount of powder using a balance designed for milligram amounts. Ideally, amounts less than 100 mg should not be weighed out. Alternatively, an approximate amount can be weighed out, and using the following formula, the exact volume can be added. The amount weighed is calculated using the formula

where W = weighted amount (µg), V = volume of stock solution required (mL), C = concentration of solution required (µg/mL), and P = potency in µg/mg. Usually, a potency value is provided. If not, it will need to be calculated from the values provided in the certificate of analysis. These values are purity (as measured, e.g., by high-performance liquid chromatography [HPLC]), water content (as measured, e.g., by Karl Fischer analysis), and active fraction, which will be lower for salts than for free acids or bases.

Potency = Purity × Active fraction × (1 − Water content [%])

The choice of final concentration will depend on whether the antimicrobial is being used immediately or whether it is intended for aliquoting and storage. Stock solutions for storage are best prepared as a 10-fold concentration of the highest dilution being used. Concentration choices will also be determined by the solubility of the antimicrobial. Stock solutions should be stored frozen at −20°C or lower and only thawed once before use; any unused thawed stock should be discarded.

Some antimicrobial agents require special solvents to achieve the high concentrations required in stock solutions. A comprehensive list of appropriate solvents and diluents is provided in reference 70 and is kept updated on a yearly basis as new antimicrobial agents in clinical development are added.

Preparation of Dilution Range

In order to avoid compounding minor errors in pipetting that would occur if low volumes were used or with repeated dilution, dispensing and dilution schemes such as those recommended by CLSI (69), the British Society for Antimicrobial Chemotherapy (BSAC) (21), and EUCAST (18) should be followed. These are adaptations of the original dilution scheme proposed by Ericsson and Sherris (71). An antimicrobial-free control plate should be added to any dilution series for quality control (i.e., to ensure that the selected strains are indeed viable).

Agar Selection

The description of each method identifies which agar media to use for particular bacterial groups or species. As all subsequent interpretations depend on data generated from these media, the nominated media for each method must be used. The ranges of media specified for each method are listed in Table 2.1.

Preparation of Agar Plates

Agars are generally prepared from a dehydrated base following the manufacturer’s instructions with regard to amounts of base, water, and autoclaving. After autoclaving, bottles should be cooled to 45°C to 50°C by placing them in a water bath at this temperature. When the agar has reached this temperature, antimicrobial dilutions and supplements—and blood if required—are added aseptically. An additional plate should be prepared without the incorporation of antimicrobials as a plate to ensure that there is adequate growth of test and control strains on the medium.

Adequate mixing of the antibiotic in the molten agar is essential, and this is best achieved by mixing each antibiotic solution in a small decanted volume of molten agar and then mixing the result with the bottle of molten agar. If the antibiotic solution is added directly to the whole bottle of molten agar, the solution should be warmed to avoid small portions of agar solidifying around the solution as it is added. The antimicrobial solution is mixed into the molten agar by gentle swirling and inverting the bottle rather than by shaking so that frothing is avoided. Plates should be poured onto a level surface as soon as practical after mixing.

The pH of each batch of agar should be measured after preparation. An aliquot of molten agar can be poured into a beaker or cup over a suitably designed pH electrode and allowed to gel and reach room temperature. Alternatively, a surface electrode can be placed on an aliquot of agar poured into a Petri dish and allowed to reach room temperature. It is also possible to macerate a sufficient amount of set agar and submerge the tip of an electrode.

Agar plates are then poured from the cooled molten agar as soon as possible after mixing in order to minimize any impact on the antimicrobial concentration. There is no specified depth, but 20 mL of molten agar in a 90-mm plate is considered adequate. It is advisable to dry the plates, for instance, in a fan-assisted drying cabinet for 10 minutes (21) or in a 35°C to 37°C incubator inverted with the lids off (72), as moisture buildup is common if covers are placed over cooling agar in Petri dishes.

Preparation of Inoculum

Most standards offer the choice of two methods for the preparation of inocula: the “growth” method and the “direct” method. The direct method is often preferred for inocula prepared from fastidious organisms, as growth of these organisms in broth can be a little unpredictable. The solution for final suspension varies somewhat between standards and sometimes between groups or species being tested but is generally either 0.85% to 0.9% NaCl, the same type of broth used in the agar, or phosphate-buffered saline (PBS). For reference work, inocula should be prepared from subcultures in order to ensure purity.

Calibration of Turbidity

Inocula are always calibrated to a turbidity standard, usually a McFarland turbidity standard, which employs particulate suspensions of barium sulfate. McFarland 0.5 is the most common choice and can be prepared as described in various texts (62,72). Briefly, it is a 0.5-mL aliquot of 1.175% w/v BaCl₂·2H₂O added to 99.5 mL of 1% w/v H₂SO₄ (62). Alternative turbidity standards made from other materials such as latex particles but optically equivalent to 0.5 McFarland are available commercially (e.g., Remel Inc, Lenexa, KS). Normally, a visual comparison is made with the turbidity standard. It is important that this be done in good lighting and against a card with a white background and a contrasting black line. More recently, there has been a move to the use of a nephelometer or photometer to achieve this calibration. For instance, at a wavelength of 550 nm, a 5-mL glass tube with 2 mL of inoculum suspension with an optical density of 0.1 to 0.12 approximates a 0.5 McFarland barium sulfate standard (72). Equally, at 625 nm and a light path of 1 cm, the standard has an optical density of 0.08 to 0.10 (62).

Barium sulfate standards are affected by light and heat, and hence they should be stored between uses in a dark place at room temperature. Their turbidity should be checked in a nephelometer or photometer monthly. They also require vigorous agitation at each use.

Growth Method

Usually, two to five morphologically similar colonies are picked from a primary or subculture plate. The wire loop is used to touch each colony, and it is then immersed in about 5 mL of the recommended broth (e.g., tryptic soy broth in the CLSI method). The broth is incubated at 35°C until it equals or exceeds the correct turbidity, generally 2 to 6 hours for rapidly growing pathogens. The broth is then diluted with a suitable sterile fluid such as broth, saline, or PBS according to the instructions until 0.5 McFarland is reached.

Direct Colony Suspension Method

In this method, two to five colonies are touched or picked up and suspended directly in the fluid of choice. Clearly, the turbidity should be adjusted subsequently as for the growth method. This method is acceptable in almost all situations. Hence, it remains an option. The choice of growth versus direct method depends on laboratory workflow. Almost all standards prefer the direct suspension method for fastidious organisms such as Haemophilus species and N. gonorrhoeae.

Plate Inoculation

In almost all circumstances, it is preferable to use a replicator apparatus to inoculate the agar plates, such as a Steers replicator (see Fig. 2.1). These are expensive items but last indefinitely. They generally come with 32 to 36 pins to deliver this number of strains to a 90-mm agar plate. The size of the pins is critical to decisions about how or whether the inoculum to be used for various species is to be further diluted. Prongs of 3 mm deliver approximately 2 µL (range 1 to 3 µL) to the plate surface; prongs of 2.5 mm generally deliver around 1 µL (18), while prongs of 1 mm deliver around 10-fold less (0.1 to 0.2 µL) (73). Hence, some standards recommend dilution for some organism groups or species when using the 3-mm pins.

There is some variation between standards regarding whether further dilution of the standardized inoculum prepared as outlined earlier should be performed and how it should be performed. The commonly stated intention is that each spot on the plate should contain around 10⁴ colony-forming units (CFUs). However, in preparing standardized inocula, there are considerable differences between species in the number of CFU per milliliter for a given turbidity (74).

Incubation

Temperature and Duration

Most organisms should be incubated at 35°C to 37°C. An incubator set at 36°C can generally operate within this range. The CLSI method specifies 35°C ± 2°C for all bacteria. The duration of incubation depends on the species. For rapidly growing species such as the Enterobacteriaceae, the glucose-nonfermenting gram-negative bacilli, Enterococcus species, and Staphylococcus species, overnight incubation for a minimum of 16 hours is needed. Longer incubation is recommended for fastidious species (18 to 20 hours plus), for Staphylococcus species when testing against antistaphylococcal penicillins (24 hours), for Campylobacter species (24 to 48 hours depending on the standard and incubation temperature), for Helicobacter pylori (CLSI 3 days), and for anaerobes (42 to 48 hours).

Atmosphere

Enrichment of air with CO₂ is recommended in most standards for a variety of fastidious species, including Streptococcus pneumoniae, Haemophilus influenzae, sometimes other Streptococcus species, and N. gonorrhoeae. The concentration is usually 5%, although some standards tolerate ranges of 4% to 6%. The effect on the activity of certain drugs under these circumstances was discussed earlier.

Campylobacter species and H. pylori demand incubation in a microaerophilic atmosphere, identical to that recommended for primary isolation. Commercial systems are available for generating the required microaerophilic environment.

Anaerobic bacteria should obviously be incubated in the standard anaerobic environment used for primary isolation of anaerobes. Either an anaerobic chamber or jar is acceptable.

Plate Stacking

The effect of stacking plates on the incubation temperature is not widely recognized. After placement in the incubator, plates in the middle of a stack will take longer to reach the desired incubation temperature than plates at the top and bottom (75,76). This is important because the incubation temperature can have a profound effect on the generation time of bacteria, which in turn will affect the end point determination. In general, no more than five plates should be used in a stack. Even then, it can take up to 4 hours for the center plate to reach the incubator temperature (76).

Reading

Plates should be read with optimum lighting, preferably on a dark, nonreflecting surface. For instance, in Figure 2.2, growth has obviously occurred at some spots but not at others. The MIC is taken as the first concentration at which no growth occurs. Most standards recommend that the appearance of one or two colonies or a faint haze can be ignored. However, if there are one or two colonies at a number of concentrations rather than a single one above the putative MIC, then the MIC is that concentration at which no colonies are seen. If this is the case, the purity of the strain should be checked, as it may be an indication of contamination. The only significant exception to these reading rules occurs in the reading of the end points for folate synthesis inhibitors, where growth can diminish gradually over a range of concentrations. CLSI recommends that the end point be read as that concentration resulting in 80% inhibition of growth (62). Other bacteriostatic drugs, such as chloramphenicol, clindamycin, tetracycline, and linezolid, can exhibit the same phenomenon to a lesser degree.

Quality Control

Quality control procedures in the performance of agar dilution are designed to ensure the reproducibility of the test and to confirm the performance of the reagents and the persons conducting the test. Most importantly, it is designed to detect errors of concentration or dilution of the antimicrobials, a not infrequent hazard in agar dilution testing. It is assumed that the laboratory is using reagents and materials from trusted suppliers who undertake quality control checks during manufacture. Using a trusted supplier does not prevent problems that may develop during shipping and handling, and hence, there is a need to have a quality control system at the laboratory level.

Quality Control Strains

The main component of quality control is the testing of reference quality control strains. The number tested varies between standards and with the type of strains being tested. Wherever possible, it is desirable to test a quality control strain of the same family, genus, and species as the strains under examination. The following ATCC reference strains have become almost universal as quality control strains for testing rapidly growing aerobic bacteria: Escherichia coli ATCC 25922, Staphylococcus aureus ATCC 29213, P. aeruginosa ATCC 27853, and Enterococcus faecalis ATCC 29212. Other strains recommended by different standards include Escherichia coli ATCC 35218 (CLSI for β-lactamase inhibitor combinations), Staphylococcus aureus Collection de l’Institut Pasteur (CIP) 6525 (Comité de l’Antibiogramme de la Société Française de Microbiologie [CA-SFM] for methicillin and oxacillin), Haemophilus influenzae ATCC 49217 (CLSI and BSAC), Haemophilus influenzae ATCC 49766 (CLSI), Haemophilus influenzaeNational Collection of Type Cultures (NCTC) 8468 (EUCAST), N. gonorrhoeae ATCC 49226 (CLSI and BSAC), Streptococcus pneumoniae ATCC 49619 (CLSI and BSAC), Helicobacter pylori ATCC 43504 (CLSI), Campylobacter jejuni ATCC 33560 (CLSI), Bacteroides fragilis ATCC 25285 (CLSI), B. fragilis NCTC 9343 (BSAC), Bacteroides thetaiotaomicron ATCC 29741 (CLSI), Eubacterium lentumATCC 43005 (CLSI), and Clostridium difficile ATCC 700057. The CLSI also has special control strains for veterinary testing: Histophilus somni ATCC 700025 and Actinobacillus pleuropneumoniae ATCC 27090.

The selection of quality control strains involves a compromise between the objective of getting strains with MICs close to the center of a dilution series of a range of antimicrobials (62) and the number of strains that would be required to achieve this objective. The strains commonly recommended have also been selected for their stability over long periods of time and repeated subculture. There is nothing to prevent a laboratory from developing and using its own quality control strains.

Storage of Quality Control Strains

Stock cultures of the quality control strains should be maintained in the freezer below −20°C (ideally at −70°C) in a suitable stabilizer such as fetal calf serum, glycerol broth, or skim milk or should be freeze-dried. Some workers recommend that two sets of stock cultures be stored: one set to create working cultures and the other as emergency backup. This will guarantee that the laboratory always has unaltered quality control strains. Working cultures should be stored on agar slants at 2°C to 8°C and subcultured each week for no more than 3 successive weeks. New working cultures are generated each month from frozen or freeze-dried stock and subcultured twice before use.

Frequency of Testing and Corrective Action

Each standard provides MIC quality control ranges for these strains against some or all of the antimicrobials of interest. The recommended quality control strains are included in each test run. When these strains fall within the quality control range, the test run is valid. When one or more results are out of this range, the test run must be considered invalid and be rerun.

Use in Routine Susceptibility Testing

Agar dilution can be adapted to routine susceptibility testing. Essentially, it is a truncated method that incorporates one or two selected concentrations of antimicrobial, usually at breakpoint values (so-called breakpoint susceptibility testing). It has been advocated as an option for routine testing by some authorities in the past (77) but is slowly being supplanted by other methods. The advantages and disadvantages of this method have been described in detail by BSAC (77). When testing multiple strains of gram-negative bacteria that may include P. mirabilis and P. vulgaris, it is necessary to include an antiswarming agent, which, as discussed earlier, can affect results. The method offers specific challenges in terms of quality control because the limited number of concentrations will only detect the most gross of preparation errors in standard quality control organisms and will often fail to detect the most common error—a 10-fold error in antibiotic dilutions. To overcome this, there are a number of options available: (a) assaying antibiotic dilutions prepared from stock solutions, (b) assaying agar plugs removed from a poured agar plate (78), and (c) assaying paper disks applied for a fixed interval to the surface of a poured agar plate (79). McDermott et al. (79) provided a detailed analysis of two of these methods and recommended methods with high precision.

Disk diffusion susceptibility testing has a long history, having evolved out of antibiotic diffusion from wells used for drug assay and susceptibility testing. The method in its various forms still has wide popularity owing to its ease of use and low cost compared with other methods. It has spawned many variants around the world. Unlike in dilution methods, an MIC value is not generated. Instead, in the development of the test, zone diameters must be compared with the MIC values of the same strains in order to determine which zone diameters predict which MIC values and hence which category of susceptibility.

Theoretical Aspects

All disk diffusion methods are based on the diffusion through agar of drug released from an impregnated disk. There are a large number of variables affecting this diffusion. Important features of antibiotic diffusion were worked out by Cooper and others in the 1950s (75,76,80–83). They have been clearly explained by Barry (3,84), who detailed the dynamics of zone formation and the “critical concentration,” “critical time,” and “critical population.”

When an antibiotic is placed in a well cut into the agar or in a disk applied to the agar surface, the drug commences diffusion immediately and diffuses in a decreasing gradient of concentration from the edge of the well or disk. Over a number of hours, the height of this gradient deceases from very steep initially to quite shallow as the drug continues to diffuse (85). In disk susceptibility testing, disks are applied after the surface has been inoculated with bacteria. The formation of the zone edge is thus a contest between the diffusion of the drug and the growth rate of the bacterial inoculum, including any initial lag phase after incubation commences. The critical concentration is the concentration just capable of inhibiting growth, and it is also that concentration found at the zone edge at the critical time. It is similar but not identical to the MIC as measured by dilution methods. The critical time is the time it takes for the critical concentration to be reached at what ultimately becomes the zone edge. It is generally around 3 to 4 hours under standard test conditions. The critical population is the number of bacterial cells found at the critical time at the ultimate zone edge. The relationships between these parameters are as follows:

where C_C = critical concentration, M = disk content, T₀ = critical time, D = diffusion coefficient of drug, h = depth of agar, r = zone radius, L = lag time, G = generation time, N′² = critical population at the critical time, and N₀ = number of viable cells at beginning of incubation. Although the following description is inaccurate, it is useful to simplify these relationships by imagining the zone of inhibition as a “cylinder” in which the drug is evenly distributed. The concentration of drug in this cylinder is thus the disk content (M) divided by the volume of the cylinder, namely pd²h/4, where d is the zone diameter and h is the depth of the agar and therefore. Thus, the zone diameter increases in proportion to the square root of the disk content and decreases in proportion to the square root of the agar depth. This defines the essential relationships between zone size, disk content, and agar depth and how variations in these affect zone diameters.

Disk Production

Disks for almost all antimicrobials can be obtained commercially. Even drugs that are still under development are likely to have disks available for use in the laboratory and clinical development programs. From time to time, it may be useful for a laboratory to manufacture its own disks. Strict standards must be adhered to if this is done. The same stipulations in drug sourcing and preparation of stocks as have been described in the section on agar dilution apply. Paper disks should be obtained that adhere to the same standards that apply to commercial manufacturers. In the United States, the standard is 740-E (Schleicher and Schuell, Keene, NH), and the paper used should be 30 ± 4 mg/cm²(86).

Solvents used in disk manufacture are described in a previous edition of this book (86).

Factors Influencing Zone Diameters

A large range of factors can influence the zone sizes produced. The most important are the disk content (also called potency, mass, or strength, namely, the amount of drug in the disk), the disk size, the diffusion characteristics of the drug, the depth of the agar, the growth rate of the bacterium (including the initial lag phase), the density of the inoculum, and the activity of the drug against the strain being tested. Other factors such as medium composition, pH, and the effect of additives are dealt with at the beginning of this chapter.

Disk Content

The amount of drug impregnated into the disk is somewhat arbitrary. Amounts are chosen that are likely to produce zones of moderate size (15 to 35 mm) under normal conditions. Different disk methods have often selected different disk contents based on early experience with the antibiotic during development and precedents set within antibiotic classes.

Disk contents will be subject to variation during manufacture, and regulators such as the U.S. Food and Drug Administration (FDA) have set tolerances on the true amount in the disk in the range of 90% to 125% of the label. Such small errors will have a small effect on the zone diameter because it is proportional to the square root of the disk content, which means that the possible variation in the diameter ranges from about −5% to +12%.

The difference in zone diameters with different disk contents has been exploited to determine critical concentrations (84). When three or more disks with a range of antibiotic contents are used, the resulting zone diameters, when squared, are directly proportional to the logarithms of the disk contents. The results can be plotted using linear regression, extrapolation of which to no zone yields the critical concentrations. These values will often be good approximations of the MICs as measured by other methods (3).

Disk Size

The extent of drug diffusion will obviously be affected by the width of the disk. Paper disks are now almost universally manufactured to be 6 mm wide. Further, the nature of the paper is subject to regulation, as different varieties of paper have been shown to affect the release characteristics of antibiotics (86). One disk method, that of Neo-Sensitabs produced by the Danish company Rosco Diagnostica A/S, Taastrup, Denmark (87,88), uses 9-mm disks made from hardened inert “chalk-like” substances. These larger disks result in larger zones for the same disk content.

Diffusion Characteristics of the Drug

The two crucial properties of the drug molecules are size and charge. In general, larger molecules diffuse more slowly, and the zones formed as a consequence are smaller. Large molecules demonstrating this include the glycopeptides such as vancomycin and teicoplanin (89). Strongly cationic molecules such as the polymyxins polymyxin B and colistin will also be inhibited in their diffusion owing to interaction with acid or sulfate groups on the agar polymer. Aminoglycosides are cationic to a lesser extent, and their diffusion is reduced slightly as a result (2).

Agar Depth

As noted, agar depth will naturally alter the size of the inhibition zone. Most methods have settled on a depth of 4 mm or a similar amount. This represents a balance between a smaller depth, which is likely to generate a reasonable zone size, and a larger depth, which is designed to reduce the zone size variation due to small variations in depth. A number of studies have demonstrated greater plate-to-plate variation in zone diameter when the agar depth is less than 3 mm (89a,89b).

Time between Inoculation and Disk Placement

The period of delay between the inoculation of the agar surface and the placement of the disks prior to incubation will have a significant effect on the ultimate zone diameter. This is a direct result of the concept of “the critical time” discussed earlier. Many bacteria are capable of initiating growth at room temperature, the temperature at which plates are usually inoculated. If the plates are preincubated at 35°C, the effect will be exaggerated. Studies examining this phenomenon have been used to elucidate the critical time (3). With two exceptions, disk diffusion methods recommend that the disks be applied within 15 minutes of plate inoculation.

Incubation Time

With rapidly growing bacteria, the zone diameter is formed within a few hours of commencing incubation (3). In theory, therefore, it is possible to read zone diameters when growth becomes visible. With further incubation—recommended for all organisms in all methods—there can be subtle changes in the zone diameter beyond the time when growth first becomes visible. These changes are the result of (a) delayed growth, (b) better visualization of partial inhibition, and/or (c) the delayed appearance of resistant variants. As zone diameter breakpoints are applied to species after specified incubation periods, it is not generally possible to use these values to determine susceptibility earlier. However, if the zone of inhibition is clearly in the resistant range, further incubation is only likely to make the zone smaller, and, therefore, it would be possible to categorize a strain as resistant when growth becomes visible. Incubation of the plate for a few hours longer than typically recommended (e.g., Enterobacteriaceae for 24 rather than 18 hours) will not usually influence the interpretation significantly. Some bacteria need longer intervals of incubation than 24 hours, such as H. pylori (3 days) and some strains of Yersinia pestis (48 hours).

A specific incubation duration of 24 hours is recommended in many methods for staphylococci when tested against the antistaphylococcal penicillins, usually represented by oxacillin or methicillin, and the glycopeptides, represented by vancomycin. In parallel with broth-based susceptibility testing methods, which can use 2% NaCl, this duration is used in order to maximize the expression of strains with heteroresistance. It must be used in consort with an incubation temperature of 35°C (and no greater). In the BSAC disk method, incubation at 30°C and the addition of 2% NaCl are also used, as these factors are known to enhance heteroresistance expression (90,91). Incubation for 24 hours is also recommended for enterococci when testing them against vancomycin in the CLSI method (27).

Incubation Temperature

Incubation temperatures are designed to optimize the growth of the organisms under test. Most human bacterial pathogens are adapted to optimum growth at 37°C. Most methods therefore recommend growth at 35°C to 37°C. Incubation at 30°C is known to enhance the expression of heteroresistance to methicillin and other antistaphylococcal penicillins (90) and has been recommended as part of the recently developed BSAC disk susceptibility test (12).

Inoculum Density

Inoculum density probably has a greater effect on the ultimate zone diameter than any other variable. As described by Barry (84), the critical population is one of the three critical parameters in determining zone size. Higher inoculum densities will result in bacterial numbers reaching the critical population sooner, at a time when the critical concentration has diffused less than with lower inoculum densities (84). In the case of very dense inocula, it is likely that no zone at all will be formed, even if the organism is susceptible. Inoculum density is particularly important when the bacteria produce inactivating enzymes such as β-lactamases, especially if the enzyme requires induction, as in the case of staphylococcal penicillinase. At low inoculum densities, the drug has the ability to kill the organism before sufficient enzyme is produced, and much larger zones will result.

In the development of different methods internationally, two approaches have been taken to the choice of inoculum density. Some methods have opted for simplicity and use inocula calibrated to a turbidity standard, almost always a McFarland 0.5 barium sulfate standard. This has been shown to result in quite different inoculum densities (in terms of CFUs per milliliter) for different species (74). Denser inocula (i.e., those with higher turbidity) are required for selected species such as H. pylori to ensure prompt and adequate growth (65).

Zone Edge

The features of the zone edge will be determined by the drug activity–species relationship, the resistance mechanism (if present), and occasional peculiarities of the species. Certain drug classes are notorious for producing “fuzzy” zone edges, where there is a gradual decrease in inhibition over a distance of 1 to 5 mm. This is a standard feature of sulfonamides and will be seen sometimes with DHFR inhibitors, amphenicols, and oxazolidinones. Most methods recommend that when a fuzzy zone is seen, the reader should make a judgment as to where approximately 80% inhibition occurs and define that as the zone edge. Judging 80% inhibition accurately and consistently takes practice. Studies presented at CLSI meetings have shown that linezolid zone diameters are most reproducibly read when viewed in transmitted rather than reflected light (27).

A second phenomenon, a “heaped” zone edge, is seen occasionally, but it is a normal feature of penicillinase-producing S. aureus. It is recommended that a heaped edge be interpreted as showing that the strain is producing penicillinase.

P. mirabilis and P. vulgaris typically swarm up the disk even when susceptible to the drug. When the organism is susceptible, a thin film is observed inside a more or less distinct zone of inhibition. This film should be ignored when making zone diameter interpretations.

Disk Spacing

Although inappropriate disk spacing is a common problem, it has not been studied in any systematic way. The distance between individual disks will be determined principally by the disk contents and organism growth rate. Higher disk contents and slower growth produce larger zones. If zones are large enough, ones from adjacent disks can run into each other and reduce the chance of accurately determining the zone diameter. The CLSI method, among others, specifies the distance between adjacent disks (24 mm from center to center) to minimize the chances of zone overlap. This limits to five the number of disks that can be placed on a 90- to100-mm plate, a considerable inconvenience if six to eight drugs need to be tested, as the cost of large plates is often substantially more that the cost of the common 90- or100-mm Petri plate. The EUCAST disk method has similar recommendations, namely 6 disks for a 100-mm plate and 12 disks for a 150-mm plate (10). The additional disk for the smaller plates is possible as a number of disks in the EUCAST have strengths lower than that of CLSI. All methods agree that disks should be placed to minimize the risk of zone overlap and possible misreading.

Current Disk Susceptibility Testing Methods

Currently, there are six methods published internationally and are kept up to date through constant revision and the addition of new agents (CLSI, EUCAST, BSAC, CA-SFM, calibrated dichotomous sensitivity [CDS], and Neo-Sensitabs [Rosco Diagnostica A/S, Taastrup, Denmark]). Of these, the most widely applied are the CLSI and EUCAST methods. The other methods are largely confined in their use to single countries—BSAC in the United Kingdom, CA-SFM in France, CDS in Australia, and Neo-Sensitabs in Denmark. Methodologically, there is nothing to recommend one method over another. However, the CLSI and EUCAST processes of setting MIC breakpoints and zone diameter criteria is based on the most comprehensive data set, which includes pharmacodynamic considerations and a large amount of clinical outcome data when available. Further, these two methods are calibrated back to the now firmly established ISO reference broth microdilution method, whereas the others have not. Because zone diameter interpretive criteria are standardized for each individual method, it is not possible to either deviate from the method or use zone diameter criteria from another method unless all testing procedures and conditions are the same.

Testing of Problematic Species

Anaerobes

Two disk diffusion methods are described for anaerobes: that of the CA-SFM method (92), which uses Wilkins-Chalgren agar, and that of BSAC (12), which uses Iso-Sensitest agar supplemented with 5% horse blood and 20 mg/L of NAD. There have been several significant attempts to develop such methods in the United States (93–95), but the results have never been considered sufficiently robust for full development. In part, this relates to the prevailing view that routine testing of anaerobic bacteria is not required. Rather, it is believed that these bacteria should be tested in specific clinical circumstances or in batches to determine trends in resistance (22). BSAC has recently provided disk susceptibility testing methods for rapidly growing anaerobes on Wilkins-Chalgren agar supplemented with 5% horse blood, including interpretative criteria for a limited number of drugs (96). This effort must be considered tentative.

Campylobacter Species

After a considerable amount of work by the veterinary subcommittee of CLSI, there is now an established method for disk susceptibility testing of the principal two species (C. jejuni and Campylobacter coli) (65). The medium employed is Mueller-Hinton supplemented with 5% sheep blood. BSAC has also recently published a method using Iso-Sensitest supplemented with 5% horse blood for these two species (12). EUCAST have also recently provided a standardized disk diffusion method for these species using Mueller-Hinton-F medium (10).

Helicobacter pylori

Some authorities believe that disk susceptibility testing of H. pylori is not feasible (12,96). Their doubts are understandable given the slow-growing nature and special growth requirement of this species. However, CA-SFM has standardized a disk test on Mueller-Hinton supplemented with 10% defibrinated horse blood and developed interpretive criteria for erythromycin and ciprofloxacin (92). There have also been several preliminary studies looking at the action of macrolides and other agents against this species (97–102). There is generally excellent agreement between disk and MIC results in these methods, and testing for resistance to macrolides and metronidazole by disk only requires multilaboratory comparison and quality control development to become a fully validated method, as clinical correlates are well established (100).

Quality Control

The main features of quality control in disk susceptibility testing revolve around the regular testing of quality control strains. Each methodology has a slightly different approach and range of quality control strains. Many advocate the testing of β-lactamase–producing strains of E. coli and H. influenzae as control organisms for β-lactamase inhibitor combinations. Careful attention must be paid to the storage of quality control strains (see the earlier discussion on agar dilution). Suitable detailed guidance is provided by CLSI (27). Quality control limits are defined for the common antimicrobials and relevant control strains. Ideally, quality control tests are conducted every time that the susceptibility tests are performed. The two major methods, CLSI and EUCAST, allow the shift to weekly testing if daily testing has proven satisfactory. Usually 1 in 20 or 3 in 30 results out of range can be tolerated. Better still is the option of graphing daily results on a Shewhart chart or similar quality control chart to detect obvious trends. If there is no obvious cause for out-of-range testing, the suspect combination should be tested daily for 5 days. If all the results are within the acceptable range, normal quality control procedures can be resumed. Otherwise, a detailed analysis of the source of errors should be undertaken. Sources of error include (a) incorrect measurement or transcription of zone diameters; (b) inadequately mixed, expired, or incorrectly stored turbidity standard; (c) one or more of the materials out of date or incorrectly stored; (d) incorrect incubation temperature or atmosphere; (e) malfunctioning equipment (e.g., dispensers); (f) incorrectly stored disks; (g) altered or contaminated control strain; (h) incorrectly prepared inocula; and (i) inoculum source more than 24 hours old. Judgment is required when deciding whether patient results relevant to the out-of-range quality control tests should be reported.

Reading of Zone Diameters

Manual Methods

Good lighting is essential to proper manual reading of zone diameters. Optimally, the plate is held a few centimeters above a black, nonreflecting background illuminated with reflected light. For transparent media, zones should be measured from the back of the plate; for opaque media such as those containing blood, zones are measured on the upper agar surface in reflected light. Two typical plates are shown in Figure 2.3. All methods normally recommend that zone diameters be measured and recorded. Although measurement is time consuming, there are long-term advantages, as interpretive criteria can change, and retrospective adjustment of interpretation is possible when breakpoints change (as they do from time to time when new resistance emerges). Measuring and recording zone diameters are probably less frequently practiced (103). Most zone diameters are simple to read. The important exceptions are where the zone edge is not sharp, as noted earlier. Very faint growth inside the zone can be ignored. When there are multiple discrete colonies within the zone, either the strain has resistant mutants (and should be reported as resistant) or the inoculum was mixed. Either way, these colonies should be subcultured and the test repeated from the original plate or inoculum. Care should be taken with hemolytic streptococci not to read the zone of hemolysis rather than the zone edge.

Automated Methods

Significant advances have been made in recent years in imaging and scanning technology. As a result, there are now a small range of commercial systems that can read and interpret zone diameters. These include the Oxoid Aura (Oxoid, Basingstoke, United Kingdom), the BIOMIC V3 system (Giles Scientific, Santa Barbara, CA), and the Adagio system (Bio-Rad Laboratories, Hercules, CA). These systems have been developed over a long period and have been extensively validated. They have benefited from significant advances in scanning and imaging technology as well as in computer software. Intersubject variability in reading is virtually eliminated, although all systems allow for manual override. Zone diameter breakpoints can be adjusted to whatever method is being used. When interfaced with laboratory information systems, zone-reading systems can save reporting time and reduce transcription errors.

Establishment of Interpretive Criteria

Although formulas have been devised for the relationship between the zone diameter and the MIC of the organism (discussed earlier), in practice it turns out that breakpoint zone diameter criteria cannot really be defined using such formulas. One obstacle is that zone diameters vary continuously and MICs are measured on a discontinuous (grouped) scale of twofold dilutions. An MIC measurement implies that the strain has an MIC within the range of that particular twofold dilution; for example, a measured MIC of 16 mg/L implies that the true MIC is somewhere between 8 mg/L and 16 mg/L. Also, as can be seen from the theoretical example in Figure 2.3, there is often a broad range (scatter) of zone diameters for strains with the same MIC. Finally, it is common for both zone diameters and MICs to have “off-scale” values (less than or equal to some number or greater than some number).

“Linear” regression has often been applied in the past to the paired MIC–zone diameter data. It is linear regression only in the sense that it uses standard linear regression methods on log-transformed MIC versus untransformed zone diameter data. However, it has little value, and its validity can be questioned. First, it is not possible to perform true linear regression if there are off-scale values. Either these data must be assigned some arbitrary value or omitted. Either way, bias is introduced into the regression as a result. Second, as the MIC data are discontinuous, there are unproven assumptions that the true MICs at each point are log-normally distributed. Third, no one has yet defined a satisfactory way of “reading out” the breakpoint zone diameters from linear regression. Instead, alternative statistical methods have been developed (and are evolving) to select zone diameter criteria given previously selected MIC breakpoint criteria.

At present, there are no internationally agreed approaches. The most popular method to date is based on that originally devised by Metzler and DeHaan (104), the so-called error rate–bound method. This is based on the minimization of “errors” (discrepancies between the zone diameter and MIC) with potential clinical impact. In the modern interpretation of MIC–zone diameter discrepancies, assuming that an intermediate category is desired, there are three types of errors: minor, where the MIC is intermediate (I) and the zone diameter is either susceptible (S) or resistant (R) or where the zone diameter is I and the MIC is either S or R; major, where the MIC is S and the zone diameter is R (i.e., the zone diameter is falsely resistant), which would deter the prescriber from choosing this drug; and very major, where the MIC is R and the zone diameter is S (i.e., the zone diameter is falsely susceptible), which could cause the prescriber to choose the drug to which the organism is really resistant. Minor errors will not occur if single MIC and zone diameter breakpoints are selected. Metzler and DeHaan (104) proposed that major errors should account for no more than 5% of the sample and very major errors for no more than 1%. They made no proposal for the percentage of minor errors that could be tolerated, other than to suggest that a large intermediate range of zone diameters is “not useful.” The selection of zone diameters was based simply on any one or pair of zone diameters that gave the desired percentages of major and very major errors. Metzler and DeHaan’s (104) paper examined single MIC breakpoints and one or two zone diameter breakpoints. Brunden et al. (105) extended the observations to the more common modern situation of having or wishing to have two breakpoint criteria each for MICs and zone diameters. They also introduced more sophisticated iterative methods for zone diameter selection and applied a recommendation that the minor errors tolerated should be no more than 5%.

This method was developed further by CLSI, which has codified the error rate tolerances for setting zone diameter breakpoints depending on the size of the intermediate MIC range (0, 1, or 2 dilutions), compensating for both the inherent error in MIC determinations (often quoted as ±1 dilution, although this an oversimplification) and the number of strains in the different susceptibility categories (Table 2.4) (106). Even more sophisticated statistical techniques have been proposed and have been proven to work well on published data (107). Very recently, these have been implemented, along with the error rate–bound method, as an online analytics tool (108), although this has yet to be adopted by any authority.

By way of showing how these methods work, the data in Figure 2.4 would yield the following error rates using the Metzler-DeHaan method: very major, 1/132 (<1%); major, 1/132 (<1%); and minor, 3/132 (2.3%). By the CLSI method (two intermediate concentrations), the rates for those with an MIC greater than one dilution above the resistance breakpoint are as follows: very major, 0/25 (0%); and minor, 0/25 (0%). The rates for those with an MIC from the resistant to the susceptible breakpoint are as follows: very major, 1/21 (4.8%); major, 1/21 (4.8%); and minor, 1/21 (4.8%). The rates for those with an MIC less than one dilution below the susceptible breakpoint are as follows: major, 0/86 (0%); and minor, 0/86 (0%).

Rapid Disk Testing

As zone diameters for rapidly growing bacteria become visible after a few hours and after the critical time has passed (see “Theoretical Aspects” earlier in this chapter), interest has been shown in the possibility of reading disk susceptibility results earlier than the conventional 16 to 20 hours (85,109–112). High degrees of correlation with standard testing have been shown for reading at 4 to 5 hours for Enterobacteriaceae (98.9%), gram-positive cocci (98.7%), and Pseudomonas species (97.9%) (2). One quarter of the strains of P. aeruginosa will have grown insufficiently at 5 hours and will require longer incubation times (85). In the single study of Salmonella typhi, results of reading at 8 hours were 100% concordant with test results after conventional incubation (112).

In order to achieve good correlation with standard disk testing, conditions for the test need to be modified. Plates should be prewarmed to 37°C prior to inoculation; the inoculum should be adjusted to McFarland 1.0 (not 0.5); 5% blood should be added for testing all gram-positive cocci to maximize growth rates; and strains classified as intermediate but with zones close to the susceptible borderline should be reclassified as susceptible (2).

Rapid reading has not become popular despite its good performance, and it is not currently offered as an alternative by any standard-setting organization. This relates partly to workflow in the laboratory; unless plates can be set up early in the day, it is often not possible to generate the results in sufficient time for the clinician to make use of them on the same day.

Direct Disk Testing

Direct susceptibility testing involves the use of the specimen as the source of the inoculum. Few specimens are suitable for direct testing because the organism being tested in the specimen may be mixed with other pathogens and contaminants. Considerable error can occur when mixed strains are present in the disk test (113).

Studies on direct testing have been conducted principally on urine and positive blood cultures, although direct testing is feasible on a wider range of specimens, especially those obtained from sterile body sites such as cerebrospinal fluid. A single study on testing wound exudates gave unreliable results (114). The results of all such tests must be considered preliminary, and tests should be repeated by a standardized method once the organism or organisms have been isolated. The benefit of direct testing is that it can generate a preliminary result a day sooner and guide the clinician in choosing an antibiotic earlier in treatment. It is of greatest value when the clinical circumstances are urgent and/or life-threatening. In the one prospective study that compared rapid testing with conventional testing, significant benefits were seen for patient care and outcomes in the group whose results were generated rapidly (115).

Urine

The possibility of direct antimicrobial susceptibility testing of potentially infected urine was recognized many years ago (116). A number of prospective studies have been conducted examining the accuracy of direct disk diffusion testing (117–131). With few exceptions, direct susceptibility testing shows a high degree of accuracy (94% to 98% agreement with conventional testing), provided strict criteria are followed for the selection of results that can be interpreted. Problems encountered that might prevent interpretation include (a) low bacterial concentrations despite true urinary infection, (b) inaccuracies due to the nonstandardized inoculum, and (c) mixed cultures (128).

The studies in which CLSI methods were used have provided the best examples of how direct testing of urine should be undertaken (127–129,131). Specimens for direct testing should be selected using microscopy criteria indicating probable urinary tract infection: an elevated white cell count (8 to 10 cells/mm³) and an absence or a paucity of squamous epithelial cells, with or without the visible presence of bacteria. The use of a Gram stain on uncentrifuged urine also increased the accuracy of the result (131). The inoculum is prepared simply by dipping a sterile cotton-tipped swab into the urine specimen, squeezing out the excess fluid against the side of the tube, and streaking a plate of plain Mueller-Hinton agar as in the standardized test. After disk placement, the plates are incubated overnight as normal. The results are interpreted using the standard criteria for those tests where the culture confirms urinary tract infection; there is adequate growth on the medium; the cultures are not mixed (even if they are all pathogens); and the quantitative culture of the original specimen is 10⁴ CFU/mL. The issue whether to report results as susceptible when they are categorized as intermediate because of high urinary antibiotic concentrations remains unresolved. Doing so increases the “accuracy” of the direct test (127). Intermediate results are sometimes seen when the urine bacterial concentration is higher than the inoculum in the standardized test. Some authors have also noted that the accuracy of direct testing drops when resistance to an antibiotic class is prevalent (128) and have attributed this to the proximity of many zone diameters to breakpoint values when resistance is present. If all the just described criteria are followed, then direct results could be considered valid and should not require repeat testing by the conventional method (128).

Direct susceptibility testing of urine has also been successfully attempted using disk elution (132) and agar dilution breakpoint plates (130).

Blood Cultures

If blood cultures become positive, it is possible to use the blood culture broth to perform direct disk susceptibility testing. This method has been confirmed to perform well in a range of studies (133–141). Rates of overall agreement with standardized testing range from 90% to 95%. One study showed that these rates of agreement could be reached after 6 hours for gram-positive bacteria, although gram-negative bacteria required overnight incubation. One advantage of blood culture broth as a source for the inoculum is that the inoculum can be standardized turbidometrically, provided the blood is not lysed (a common problem with E. coli and S. pneumoniae), although at least one study showed good agreement by using the positive blood culture broth directly (137). There are several options for preparing the inoculum, including direct use of the blood culture broth (137), subculture of a fixed withdrawn volume and further broth incubation to reach McFarland 0.5 (136), and mixing the blood culture broth with broth or distilled water to adjust to the McFarland standard (2). A further advantage of direct blood culture broth use is that knowledge of the Gram stain characteristics of the isolate can guide the choice of a test medium.

There are some drawbacks to direct testing from blood culture bottles: the presence of high thymidine concentrations in blood culture media will carry over and antagonize the sulfonamides and trimethoprim, and the presence of sodium polyanethol sulfonate (SPS) in older blood culture media to remove antibiotics can antagonize some penicillins, aminoglycosides, and glycopeptides (2). All results generated by direct testing should be confirmed by standardized testing the next day, as bacteremia is a serious condition requiring confident antimicrobial choices.

Other Specimens

Other specimens in fluid format are also potentially capable of direct testing when infected. Direct disk testing has also been attempted with intraocular aspirates (142) and bronchial secretions (143), with some success.

Gradient diffusion susceptibility testing was pioneered by a Swedish company, AB BIODISK (Solna, Sweden), under the brand name Etest and is now owned and marketed by BioMérieux, Craponne, France. Since patent expiry, a competitor product (M.I.C.E., Thermo Fisher Scientific, Basingstoke, United Kingdom) has appeared on the market.

Gradient diffusion technology exploits the properties of antibiotic diffusion in agar to generate MIC values. It achieves this by applying a gradient of antibiotic along a carrier strip placed on the inoculated surface of an agar plate in the same way that antibiotic disks are placed. The antibiotic gradient is created on the strip by applying different concentrations of antibiotics in repeated arrays of an increasing number of small dots. When applied to the agar surface, the antibiotic diffuses into the surrounding medium in high to low amounts from one end of the strip to the other. The gradient remains stable after diffusion, and the zone of inhibition created takes the form of a pointed ellipse, from which the original Etest takes its name. The MIC is read off the scale printed on the upper side of the strip at the point where the zone edge meets the strip edge (Fig. 2.5).

The first evaluations of this product proved the Etest was comparable to conventional MIC testing (144–146). The Etest has since been evaluated against a very broad range of bacteria, including mycobacteria, as well as yeasts and molds. In general, the gradient diffusion strips produce results similar to those of conventional MIC tests against almost all bacteria. Its attraction is that it is simple to set up (as simple as disk diffusion) and takes less work, time, and materials than any agar or broth dilution method. Difficulties are encountered with some organism–antibiotic combinations, because finding the end point (where the zone intersects the strip) suffers from the same problems as zone edge interpretation in disk diffusion. Indeed, the effect is sometimes exaggerated. However, as the innovator product has undergone constant reevaluation over the 14 years since its release, the company has provided an excellent range of technical guides, charts, and other visual materials to assist the user in reading and interpreting the results correctly. Problems such as microcolonies at the intersection, swarming, paradoxical effects, hemolysis, resistant subpopulations, end points for bacteriostatic drugs, and technical errors are all dealt with and examples provided.

Like all such products, gradient diffusion strips must be properly stored and handled. The details of how to store the strips are provided by the manufacturer. Although the strips can be readily handled individually by sterile (flamed) forceps, devices are available for handling single strips in particular ways. The strips are made of plastic and are flexible enough to allow misplacement if forceps are used. As in the case of routine disks, the antibiotic will commence rapid diffusion as soon as the strip is placed. Hence, it is not possible to relocate the strip once it is placed. The length of the strips is such that one fits comfortably on a 90-mm Petri dish. With careful placement, two different antibiotics can be tested on a single 90-mm plate and is necessary with combination testing. If done, the strips must be oriented in opposite directions. Larger plates are recommended if more than one strip is to be used. A normal 150-mm Petri plate will comfortably accommodate six strips provided that the bottoms of the strips are all facing the center of the plate.

Gradient diffusion strips have the MIC scale printed on the upper side. The scale differs between the two products. On Etest strips, the scale is modified from the conventional base 2 logarithmic scale and includes intermediate “dilutions,” such as 3 mg/L between the values of 2 and 4 mg/L. These values are not true halfway points on the log₂ scale but rather rounded values, and the lines on the strip at these dilutions are placed accordingly. The scale on M.I.C.E. is arranged differently, with the intermediate values represented by a dark band rather than a line and value, requiring the reader to make some judgment about what that intermediate value should be.

The greatest advantage of the gradient diffusion strips is that they provide a laboratory with the ability to conveniently perform an MIC determination on a single strain without the full setup required for a conventional MIC test. If the clinical circumstances demand it, a quantitative susceptibility value can be easily generated. As an example, suppose a serious invasive infection such as meningitis or septicemia has been caused by S. pneumoniae but the categorical susceptibility results (susceptible, intermediate, resistant) provide insufficient information to guide clinical antibiotic decision making. A gradient diffusion test or tests can also be used to validate unusual resistances generated by conventional testing.

Gradient Agar Plates

The concept of creating a single agar plate with a gradient of antibiotic concentration is an old one (147,148). Gradient plates provide a range of concentrations in a single plate and have been used principally to compare the susceptibilities of a range of strains on a single plate, especially when the MIC differences are likely to be small. More recently, gradient plates have also been used to select for resistant variants. When a gradient plate is used for this purpose, the strain being examined must be exposed to subinhibitory concentrations in one part of the plate (149).

In the most common method for preparing a gradient plate, a square Petri plate (100 mm by 100 mm usually) is tilted at a 12° angle while kept precisely level along the perpendicular axis (150–151). Agar (~25 mL per 100-mm plate) containing the antibiotic of interest at a known concentration is poured into the tilted plate and allowed to set. The plate is then set flat and precisely level, and a second identical volume of agar containing no antibiotic is poured over the top. The plate is ready to use after this second layer has set. Strains under test are usually streaked in a linear fashion starting from the lowest to the highest concentration (from the thin to the thick end of the lower layer).

The emergence of low-level resistance to vancomycin in a number of countries has led to the reintroduction of a research tool now called population analysis profiling. The technique was successfully applied to demonstrate the emergence of resistance during treatment of P. aeruginosa in an animal model with single agents (152). It was later applied to the characterization of methicillin resistance in S. aureus(153). Its application to the study of reduced vancomycin susceptibility in S. aureus began with Hiramatsu et al. (154), who successfully applied this technique to strains with homogeneous and heterogeneous reduced susceptibility. Refinements have been made to the technique in an attempt to minimize false positivity by comparing the area under the curve (AUC) of the population profile (log-transformed CFU/mL) with that of the original heteroresistant strain (Mu3) isolated in Japan (150). AUC ratios of 0.9 comparing the test strain with Mu3 were considered positive for heteroresistance. A ratio of 1.3 was used in a later study to define strains with homogeneous reduced resistance (155). There is ongoing uncertainty about the sensitivity and specificity for detecting hetero- and homogeneous reduced glycopeptide susceptibility in staphylococci using this test. It is very labor-intensive and consumes significant resources and thus is likely to remain a research or reference tool (156).

The artificial nature of in vitro susceptibility testing frequently raises questions about the interpretation and meaning of results for patient management. Certainly, there are significant differences between the in vitro and in vivo situations on a range of factors such as (a) fixed versus varying drug concentrations, (b) no protein binding versus protein binding, (c) 16 to 24 hours of drug exposure versus days for treatment courses, (d) fixed versus variable physiologic conditions such as pH and eH, (e) absence versus presence of host immune response, (f) optimum growth conditions versus restricted growth in vivo, and (g) low versus high inocula at the start of drug exposure. It is likely, therefore, that the MIC in vitro is different from the MIC in vivo in some, many, or all situations.

The differences between the morphologic and physiologic characteristics of bacterial growth in vitro and in vivo can be stark, suggesting that in vitro simulation of conditions might aid in getting in vitro responses to better mimic in vivo responses to therapy (157,158). Tests on solid surfaces may approximate the state of organisms in vivo closer. Standard laboratory susceptibility tests do not detect the range of antibacterial effects of an antimicrobial, such as sub-MIC effects (157,159), postantibiotic effects (160,161), postantibiotic sub-MIC effects (162), leukocyte enhancement effects (163), and supra-MIC effects.

Yet there is evidence that in vitro testing as conducted in the routine laboratory is predictive of response to therapy. Although there are very few studies that have examined test results versus patient outcomes in the routine setting (164), those that have been done, together with clinical experience and the findings of some prospective clinical studies, suggest that there is a high correlation. The question is why such correlation is possible.

The first part of the answer is that although MICs and their susceptibility testing correlates are artificial and indeed somewhat arbitrary, this does not matter. The yard and the meter are also arbitrary units of measurement, yet no one would deny their universal value for making measurements and, more importantly, comparing of measurements. Provided that there are MIC methods that are simple and reproducible and have low degrees of scatter, MICs can be used as a basis for comparing in vitro activity with the outcomes of treatment.

Second, if reproducible correlations can be shown between MICs (whatever values are generated) and bacteriologic and clinical outcomes, concerns about their poor reflection of in vivo activity can be bypassed. In essence, MICs become a convenient “yardstick” for predicting treatment outcomes. Prediction of outcomes has always been the goal of in vitro susceptibility testing. It is only then necessary to demonstrate that routine methods themselves correlate with MICs obtained using standardized methods.

Correlations between MICs and outcomes are now being generated through the emergence of the science of antimicrobial pharmacodynamics (166–169). By the use of in vitro and animal models, it has been possible to define the key pharmacodynamic parameters for a broad range of antimicrobial classes, including the β-lactams (166,170), the aminoglycosides (170,171), the macrolides (172,173), the quinolones (174,175), and the glycopeptides (166,176), and to provide guidance on the polymyxins (177), the tetracyclines (179), the lincosamides (178), and the oxazolidinones (166). Extrapolation of these findings to human infections has so far not revealed major discrepancies. Extrapolation is frequently necessary, as it is unethical to conduct clinical studies in which some patients must fail therapy in order to confirm pharmacodynamic correlates. Occasionally, failure to take into account pharmacodynamics has led to poor outcomes (180,181). Data on pharmacodynamics are now being applied as part of the development of breakpoint values.

Third, another goal of susceptibility testing is to detect the presence of resistance, defined microbiologically as the acquisition of a mutation or external genetic material or the induction of natural enzymes with an effect on the phenotype. Resistance will be reflected by an increase in MIC to above normal levels. All susceptibility testing methods include the concept of detecting abnormal phenotypes, although many have set breakpoints that will not detect all abnormal phenotypes. This reflects the tension between microbiologic and pharmacodynamic breakpoints, a tension that is not yet fully resolved. In a range of standardized methods, supplementary tests are recommended to increase the chances of detecting abnormal phenotypes, including β-lactamase detection tests and screening plates. How closely abnormal phenotypes that are below the breakpoints correlate with clinical outcomes is largely unknown, but the conservative position is to interpret them as resistant (or at least intermediate) until it is known that they are susceptible clinically.

In clinical practice, combinations of antimicrobials are often used (a) to ensure adequate coverage of potential pathogens in empirical therapy, (b) to increase the likelihood of clinical success by achieving synergy between agents, (c) to reduce the risk of selection of resistance during treatment, and (d) to boost the efficacy of a primary agent that tests as resistant. Of these, the one of greatest interest to the laboratory is synergy, for which a great variety of methods have been developed over the decades, most using broth systems (182) but some using agar. Important for the laboratory, there are actually only a few instances where true synergy has been shown in clinical practice (apart from old combination agents such as trimethoprim-sulfamethoxazole), the most important of these being enterococcal endocarditis (183).

Agar-Based Methods for Detection of High-Level Aminoglycoside Resistance

Cure rates of enterococcal endocarditis are higher when a cell wall–active agent (penicillin or glycopeptides) is combined with an aminoglycoside, even though Enterococcus species have low-level intrinsic resistance to this class. The two agents most often favored by treating clinicians are gentamicin and, when there is high-level resistance to gentamicin, streptomycin. When there is high-level aminoglycoside resistance due to the presence of an acquired resistance mechanism, the synergistic effect is abolished. Endocarditis isolates of Enterococcus species should therefore be tested for the presence of high-level resistance to gentamicin and streptomycin. Two agar-based methods are described by CLSI: a disk diffusion method performed on Mueller-Hinton agar using high-potency disks of gentamicin (120 µg) and streptomycin (300 µg) and an agar screening plate method performed on two different brain–heart infusion agar plates containing gentamicin 500 µg/mL and streptomycin 2,000 µg/mL (69). EUCAST describes a disk diffusion method using a different potency disk for gentamicin and with interpretive criteria (184). Other disk diffusion methods provide details of similar tests.

Disk Approximation Methods

The most time-honored agar-based method for detecting the drug interactions is that of disk approximation. This is a phenotypic test that still has application for the detection of resistance mechanisms such as inducible resistance to clindamycin, extended-spectrum β-lactamases, and some carbapenemases (27,69). When testing for the presence of synergy or antagonism, the disk approximation test will provide visual evidence of these types of interactions (Figs. 2.6 and 2.7). The test can be applied to any organism that can be grown on agar plates, including those with the modifications and additives mentioned at the beginning of this chapter. Figure 2.6 shows the three different types of interaction that may be observed: indifferent, synergistic, and antagonistic. Note the two different types of synergistic interaction that may be seen. This is because interactions are critically dependent on both disk potency and separation distance. Unfortunately, there are no simple rules for determining the optimum combination of potency and separation distance of disks. This limits the value of this test to that of being indicative of a type of interaction, but it is unable to quantify the interaction further.

Disk Combination Methods

A variation on the disk approximation method is to use two disks containing the drugs individually plus a disk containing both drugs. After incubation with the test strain, the zone diameters of the three disks are compared. Interpretation is more difficult than the disk approximation method: A larger zone with the combination disk than with either single drug disk alone excludes antagonism but does not readily distinguish between indifference and synergy. A smaller zone with the combination disk than with either single drug disk alone suggests antagonism.

Gradient Diffusion Combination Methods

More recently, with the emergence of highly multiresistant organisms in clinical practice, there has been renewed interest in examining antimicrobial interactions using the application of gradient diffusion strips (184). Most interest has focused on multiresistant A. baumannii (184–195), but the method has also been used with Enterococcus faecium (196,197), staphylococci (198,199), P. aeruginosa (195,200–202), Enterobacteriaceae (195,203), and Stenotrophomonas maltophilia (195,204).

The features of this method are illustrated in Figure 2.8. The MICs are determined with the two gradient diffusion strips alone on the agar medium most appropriate for the test strain. The following day, two new strips are placed at right angles to other such that they intersect at the marks on their individual MICs. After a further day’s incubation, when synergy is present, there will be a new zone of inhibition between the two strips, and the new lower MICs for both agents can be read off the two strips. The results are then interpreted using the same fractional inhibitory concentrations employed in checkerboard broth microdilution procedures (182).

A variation on this method has also been described. It takes advantage of the fact that an antimicrobial agent can be incorporated into the medium on which the test organism and the gradient diffusion strips are placed, making it possible to test the interaction of three antimicrobial agents at once (Fig. 2.9) (187).

  1.  Acar JF, Goldstein FW. Disk susceptibility testing. In: Lorian V, ed. Antibiotics in laboratory medicine. 3rd ed. Baltimore: Lippincott Williams & Wilkins, 1991:17–52.

  2.  Acar JF, Goldstein FW. Disk susceptibility testing. In: Lorian V, ed. Antibiotics in laboratory medicine. 4th ed. Baltimore: Lippincott Williams & Wilkins, 1996:1–51.

  3.  Barry AL. Procedure for testing antibiotics in agar media: theoretical considerations. In: Lorian V, ed. Antibiotics in laboratory medicine. Baltimore: Lippincott Williams & Wilkins, 1980:1–23.

  4.  Clinical and Laboratory Standards Institute. Susceptibility testing of mycobacteria, nocardiae and other aerobic actinomycetes—second edition. Wayne, PA: Clinical and Laboratory Standards Institute, 2011. CLSI document M24-A2.

  5.  Clinical and Laboratory Standards Institute. Methods for antimicrobial susceptibility testing for human mycoplasmas. Wayne, PA: Clinical and Laboratory Standards Institute, 2011. CLSI document M43-A.

  6.  Clinical and Laboratory Standards Institute. Method for antifungal disk diffusion susceptibility testing of yeasts—second edition. Wayne, PA: Clinical and Laboratory Standards Institute, 2009. CLSI document M44-A2.

  7.  Clinical and Laboratory Standards Institute. Method for antifungal disk diffusion susceptibility testing of nondermatophyte filamentous fungi. Wayne, PA: Clinical and Laboratory Standards Institute, 2010. CLSI document M51-A.

  8.  McHugh DJ. A guide to the seaweed industry. In: FAO Fisheries Technical Paper 441. Rome, Italy: Food and Agriculture Organization of the United Nations, 2003.

  9.  Amisen R, Galatas F. Production, properties and uses of agar. In: HcHugh DJ, ed. Production and utilization of products from commercial seaweeds. Rome, Italy: Food and Agriculture Organization of the United Nations, 1987. http://www.fao.org/docrep/X5822E/X5822E00.htm. Accessed March 1, 2013.

 10.  European Committee on Antimicrobial Susceptibility Testing. Antimicrobial susceptibility testing—EUCAST disk diffusion method, version 2.1. http://www.eucast.org/fileadmin/src/media/PDFs/EUCAST_files/Disk_test_documents/Manual_v_2.1_EUCAST_Disk_Test.pdf. Accessed March 1, 2013.

 11.  Bell SM, Pham JN, Nguyen TT. Antibiotic susceptibility testing by the CDS method. http://web.med.unsw.edu.au/cdstest/GTF_CDS_site/WebPages/HomeLevel/ManualFrames.htm. Accessed March 1, 2013.

 12.  British Society for Antimicrobial Chemotherapy. BSAC methods for antimicrobial susceptibility testing version 11.1. May 2012. http://bsac.org.uk/wp-content/uploads/2012/02/Version-11.1-2012-Final-.pdf. Accessed March 1, 2013.

 13.  Bauer AW, Kirby WMM, Sherris JC, et al. Antibiotic sensitivity testing by a standardized single disk method. Am J Clin Path 1966;45:493–496.

 14.  Bridson EY, ed. The Oxoid manual. 9th ed. Basingstoke, United Kingdom: Oxoid Limited, 2006.

 15.  Members of the SFM Antibiogram Committee. Comité de l’Antibiogramme de la Société Française de Microbiologie report 2003. Int J Antimicrob Agents 2003;21:364–391.

 16.  Clinical and Laboratory Standards Institute. Protocols for evaluating dehydrated Mueller-Hinton agar—second edition. Wayne, PA: Clinical and Laboratory Standards Institute, 2006. CLSI document M6-A2.

 17.  International Organization for Standardization. Criteria for acceptable lots of dehydrated Mueller-Hinton agar and broth for antimicrobial susceptibility testing. Geneva: International Organization for Standardization. ISO 16782 March 8, 2013 draft.

 18.  European Committee for Antimicrobial Susceptibility Testing of the European Society for Clinical Microbiology and Infectious Diseases. Determination of minimum inhibitory concentrations (MICs) of antibacterial agents by agar dilution. Clin Microbiol Infect 2000;6:509–515.

 19.  Toohey M, Francis G, Stingemore N. Variation in Iso-Sensitest agar affecting β-lactam testing. ASIGnation 1990;1(6):5–8.

 20.  Swedish Reference Group on Antibiotics. Dilution methods. http://www.srga.org/RAFMETOD/Basmet2.htm. Accessed March 1, 2013.

 21.  Andrews JM. Determination of minimum inhibitory concentrations. J Antimicrob Chemother 2001;48(Suppl 1):5–16.

 22.  Clinical and Laboratory Standards Institute. Methods for antimicrobial susceptibility testing of anaerobic bacteria: approved standard—fifth edition. Wayne, PA: Clinical and Laboratory Standards Institute, 2012. CLSI document M11-A8.

 23.  Hecht DW, Lederer L. Effect of the choice of medium on the results of in vitro susceptibility testing of eight antimicrobials against the Bacteroides fragilis group. Clin Infect Dis 1995;20(Suppl 2):S346–S349.

 24.  Roe DE, Hecht DW, Finegold SM, et al. Multilaboratory comparison of anaerobe susceptibility results using three different agar media. Clin Infect Dis 2001;35(Suppl 1):S40–S46.

 25.  Roe DE, Hecht DW, Finegold SM, et al. Multilaboratory comparison of growth characteristics for anaerobes using five different media. Clin Infect Dis 2001;35(Suppl 1):S36–S45.

 26.  Waterworth PM. Sulphonamides and trimethoprim. In: Reeves DS, Phillips I, Williams JD, et al, eds. Laboratory methods in antimicrobial chemotherapy. Edinburgh, United Kingdom: Churchill Livingstone, 1978:82–84.

 27.  Clinical and Laboratory Standards Institute. Performance standards for antimicrobial disk susceptibility tests: approved standards—eleventh edition. Wayne, PA: Clinical and Laboratory Standards Institute, 2012. CLSI document M2-A11.

 28.  Kenny MA, Pollock HM, Minshew BH, et al. Cation components of Mueller-Hinton agar affecting testing of Pseudomonas aeruginosa susceptibility to gentamicin. Antimicrob Agents Chemother 1980;17:55–62.

 29.  Casillas E, Kenny MA, Minshew BH, et al. Effect of ionized calcium and soluble magnesium on the predictability of the performance of Mueller-Hinton agar susceptibility testing of Pseudomonas aeruginosa with gentamicin. Antimicrob Agents Chemother 1981;19:987–992.

 30.  Fuchs PC, Barry AL, Brown S. Daptomycin susceptibility tests: interpretive criteria, quality control, and the effect of calcium on in vitro tests. Diagn Microbiol Infect Dis 2000;38:51–58.

 31.  Fernández-Mazarrasa C, Mazarrasa O, Calvo J, et al. High concentrations of manganese in Mueller-Hinton agar increase MICs of tigecycline determined by Etest. J Clin Microbiol 2009;47:827–829.

 32.  Veenemans J, Mouton JW, Kluytmans JA, et al. Effect of manganese in test media on in vitro susceptibility of Enterobacteriaceae and Acinetobacter baumannii to tigecycline. J Clin Microbiol 2012;50:3077–3079.

 33.  Girardello R, Bispo PJM, Yamanaka TM, et al. Cation concentration variability of four distinct Mueller-Hinton agar brands influences polymyxin B susceptibility results. J Clin Microbiol 2012;50:2414–2418.

 34.  Waterworth PM. The aminoglycosides. In: Reeves DS, Phillips I, Williams JD, et al, eds. Laboratory methods in antimicrobial chemotherapy. Edinburgh: Churchill Livingstone, 1978:85–87.

 35.  D’Amato RF, Thornsberry C, Baker CN, et al. Effect of calcium and magnesium on the susceptibility of Pseudomonas species to tetracycline, gentamicin, polymyxin B, and carbenicillin. Antimicrob Agents Chemother 1975;7:596–600.

 36.  Reller LB, Schoenknecht FD, Kenny MA, et al. Antibiotic susceptibility testing of Pseudomonas aeruginosa: selection of a control strain and criteria for magnesium and calcium content of media. J Infect Dis 1974;130:454–463.

 37.  Washington JA, Snyder RJ, Kohner PC, et al. Effect of cation content of agar on the activity of gentamicin, tobramycin and amikacin against Pseudomonas aeruginosa. J Infect Dis 1978;137:103–111.

 38.  Cooper GL, Louie A, Baltch AL, et al. Influence of zinc on Pseudomonas aeruginosa susceptibilities to imipenem. J Clin Microbiol 1993;31:2366–2370.

 39.  Daly JS, Dodge RA, Glew RH, et al. Effect of zinc concentration in Mueller-Hinton agar on susceptibility of Pseudomonas aeruginosa to imipenem. J Clin Microbiol 1997;35:1027–1029.

 40.  Atmaca S. Effect of zinc concentration in Mueller-Hinton agar on susceptibility of Pseudomonas aeruginosa to meropenem. J Med Microbiol 1998;47:653.

 41.  Cooke P, Heritage J, Kerr K, et al. Different effects of zinc ions on the in vitro susceptibilities of Stenotrophomonas maltophilia to imipenem and meropenem. Antimicrob Agents Chemother 1996;40:2909–2910.

 42.  Berkman S, Henry RJ, Housewright RD. Studies on streptomycin, I: factors influencing the activity of streptomycin. J Bacteriol 1947;53:567.

 43.  Medeiros AA, O’Brien TF, Wacker WEC, et al. Effect of salt concentration on the apparent in vitro susceptibility of Pseudomonas and other Gram-negative bacilli to gentamicin. J Infect Dis 1971;124(Suppl):S59–S66.

 44.  Rosenblatt JE, Schoenknecht F. Effect of several components of anaerobic incubation on antibiotic susceptibility results. Antimicrob Agents Chemother 1972;1:433–440.

 45.  Bemer-Melchior P, Juvin M-E, Tassin S, et al. In vitro activity of the new ketolide telithromycin compared to those of the macrolides against Streptococcus pyogenes: influences of resistance mechanisms and methodological factors. Antimicrob Agents Chemother2000;44:2999–3002.

 46.  Dibb WL, Digranes A, Bottolfson KL. Effects of carbon dioxide upon the in vitro activity of erythromycin. Acta Pathol Microbiol Immunol Scand 1986;94:173–176.

 47.  Goldstein EJC, Sutter VL, Kwok Y-Y, et al. Effect of carbon dioxide on in vitro susceptibility of anaerobic bacteria to erythromycin. Antimicrob Agents Chemother 1981;20:705–708.

 48.  Johnson MM, Hill SL, Piddock LJV. Effect of carbon dioxide on testing of susceptibilities of respiratory tract pathogens to macrolides and azalide antimicrobial agents. Antimicrob Agents Chemother 1999;43:1862–1865.

 49.  Retsema JA, Brennan LA, Girard AE. Effects of environmental factors on the in vitro potency of azithromycin. Eur J Clin Microbiol Infect Dis 1991;10:834–842.

 50.  Spangler SK, Jacobs MR, Appelbaum PC. Effect of CO₂ on susceptibilities of anaerobes to erythromycin, azithromycin, clarithromycin, and roxithromycin. Antimicrob Agents Chemother 1994;38:211–216.

 51.  Bouchillon SK, Johnson JL, Hoban DJ, et al. Impact of carbon dioxide on the susceptibility of key respiratory tract pathogens to telithromycin and azithromycin. J Antimicrob Chemother 2005;56:224–227.

 52.  Johnson J, Bourchillon S, Pontani D. The effect of carbon dioxide on susceptibility testing of azithromycin, clarithromycin and roxithromycin against isolates of Streptococcus pneumoniae and Streptococcus pyogenes by broth microdilution and Etest: Artemis Project–first-phase study. Clin Microbiol Infect 1999;5:327–330.

 53.  Spangler SK, Appelbaum PC. Oxyrase, a method which avoids CO₂ in the incubation atmosphere for anaerobic susceptibility testing of antibiotics affected by CO₂. J Clin Microbiol 1993;31:460–462.

 54.  Jansen JE, Bremmelgarrd A. Susceptibility testing of 7 antibiotics against anaerobic bacteria: comparison of 2 different media and carbon dioxide concentrations. Acta Pathol Microbiol Scand B 1987;95:65–73.

 55.  Hansen SL, Swomley P, Drusano G. Effect of carbon dioxide and pH on the susceptibility of Bacteroides fragilis group to erythromycin. Antimicrob Agents Chemother 1991;19:335–336.

 56.  Reynolds AV, Hamilton-Miller JM, Brumfitt W. Diminished activity of gentamicin under anaerobic or hypercapnic conditions. Lancet 1976;1(7957):447–449.

 57.  Traub WH, Leonhard B. Antibiotic susceptibility tests with fastidious and nonfastidious bacteria reference strains: effects of aerobic versus hypercapnic incubation. Chemotherapy 1995;41:18–33.

 58.  Ward PB, Palladino S, Looker JC, et al. P-nitro-phenylglycerol in susceptibility testing media alters the MICs of antimicrobials for Pseudomonas aeruginosa. J Antimicrob Chemother 1993;31:489–496.

 59.  Ward PB, Palladino S, Looker JC. P-nitrophenylglycerol in susceptibility testing media alters the MICs of antimicrobials for aerobic Gram-negative bacilli. J Antimicrob Chemother 1993;31:803–805.

 60.  Winstanley T, Edwards C, Limb D, et al. Evaluation of a surfactant, Dispersol LN, as an anti-swarming agent in agar dilution susceptibility testing. J Antimicrob Chemother 1994;33:353–356.

 61.  Liu M-C, Lin S-B, Chien H-F, et al. 10’(Z),13’(E)-heptadecadienylhydroquinone inhibits swarming and virulence factors and increases polymyxin B susceptibility in Proteus mirabilis. PLoS One 2012;7:e45563.

 62.  Clinical and Laboratory Standards Institute. Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically; approved standard—ninth edition. Wayne, PA: Clinical and Laboratory Standards Institute, 2012. CLSI document M7-A9.

 63.  Murray CK, Walter EA, Crawford S, et al. Abiotrophia bacteremia in a patient with neutropenic fever and antimicrobial susceptibility testing of Abiotrophia isolates. Clin Infect Dis 2001;32:e140–e142.

 64.  Touhy MJ, Procop GW, Washington JA. Antimicrobial susceptibility of Abiotrophia adiacens and Abiotrophia defectiva. Diagn Microbiol Infect Dis 2000;38:189–191.

 65.  Clinical and Laboratory Standards Institute. Methods for antimicrobial dilution and disk diffusion susceptibility tests for testing of infrequently isolated or fastidious bacteria—second edition. Wayne, PA: Clinical and Laboratory Standards Institute, 2010. CLSI document M45-A2.

 66.  Acar JF, Goldstein FW, Lagrange P. Human infections caused by thiamine- or menadione-requiring Staphylococcus aureus. J Clin Microbiol 1978;8:142–147.

 67.  Tapsall JW, Wilson E, Harper J. Thymine dependent strains of Escherichia coli selected by trimethoprim-suphamethoxazole during therapy. Pathology 1974;6:161–167.

 68.  Clark RB, Lewinski MA, Loeffelholz MJ, et al. Cumitech 31A, verification and validation of procedures. In: Sharp SE, ed. Clinical microbiology laboratory. Washington, DC: ASM Press, 2009.

 69.  Clinical and Laboratory Standards Institute. Performance standards for antimicrobial susceptibility testing: fourteenth informational supplement. Wayne, PA: Clinical and Laboratory Standards Institute, 2013. CLSI document M100-S23.

 71.  Ericsson HM, Sherris JC. Antibiotic sensitivity testing: report of an international collaborative study. Acta Pathol Microbiol Scand 1971;217(Suppl B):1–90.

 72.  Courvalin P, Soussy J-C, eds. Technical recommendations for in vitro susceptibility testing. 1996 Report of the Comité de l’Antibiogramme de la Société Française de Microbiologie. Clin Microbiol Infect 1996;2(Suppl 1):S11–S34.

 73.  Rousseau D, Harbec PS. Delivery volumes of the 1- and 3-mm pins of a Cathra replicator. J Clin Microbiol 1987;25:1311.

 74.  Moosdeen F, Williams JD, Secker A. Standardization of inoculum size for disc susceptibility testing: a preliminary report of a spectrophotometric method. J Antimicrob Chemother 1988;21:439–443.

 75.  Cooper KE, Gillespie WA. The influence of temperature on streptomycin inhibition zones in agar cultures. J Gen Microbiol 1952;7:1–7.

 76.  Cooper KE, Linton AH. The importance of the temperature during the early hours of incubation of agar plates in assays. J Gen Microbiol 1952;7:8–17.

 77.  Working Party on Antibiotic Sensitivity Testing of the British Society for Antimicrobial Chemotherapy. A guide to sensitivity testing. J Antimicrob Chemother 1991;27(Suppl D):1–50.

 78.  Franklin JC. Quality control in agar dilution sensitivity testing by direct assay of the antibiotic in solid medium. J Clin Path 1980;33:93–95.

 79.  McDermott SN, Hartley TF. New datum handling methods for the quality control of antibiotic solutions and plates used in the antimicrobial susceptibility test. J Clin Microbiol 1989;27:1814–1825.

 80.  Cooper KE. Theory of antibiotic inhibition zones in agar media. Nature 1955;176:510–511.

 81.  Cooper KE, Linton AH, Sehgal SN. The effect of inoculum size on inhibition zones in agar media using staphylococci and streptomycin. J Gen Microbiol 1958;18:670–687.

 82.  Cooper KE. The theory of antibiotic inhibition zones. In: Kavanaugh F, ed. Analytical microbiology. New York: Academic Press, 1964:1–86.

 83.  Humphrey JH, Lightbown J. A general theory for plate assay of antibiotics with some practical applications. J Gen Microbiol 1952;7:129–143.

 84.  Barry AL. The antimicrobic susceptibility test: principles and practices. Philadelphia: Lea and Febiger, 1976.

 85.  Lorian V. A five hour disc antibiotic susceptibility test. In: Lorian V, ed. Significance of medical microbiology in the care of patients. Baltimore: Lippincott Williams & Wilkins, 1977:203–212.

 86.  Shungu D. Chemical and physical properties of antibiotics: preparation and control of antibiotic susceptibility disks and other devices containing antibiotics. In: Lorian V, ed. Antibiotics in laboratory medicine. 4th ed. Baltimore: Lippincott Williams & Wilkins, 1995:766–792.

 87.  Casals JB, Gylling Pedersen O. Tablet sensitivity testing: a comparison of methods. Acta Pathol Microbiol Scand 1972;80:806–816.

 88.  Rosco Diagnostica A/S. Users guide: Neo-Sensitabs™ susceptibility testing, 2011. https://rosco.docontrol.com. Accessed March 1, 2013.

 89.  Cavenaghi LA, Biganzoli E, Danese A, et al. Diffusion of teicoplanin and vancomycin in agar. Diag Microbiol Infect Dis 1992;15:253–258.

89a.  Barry AL, Fay GD. The amount of agar in antimicrobic disk susceptibility test plates. J Clin Pathol 1973;59:196–198.

89b.  Davis WW, Stout TR. Disc plate method of microbiological antibiotic assay, I: Factors influencing variability and error. Appl Microbiol 1971;22:659–665.

 90.  Annear D. Full expression of methicillin resistance in Staphylococcus aureus. J Antimicrob Chemother 1985;15:253–254.

 91.  Baker CN, Huang MB, Tenover FC. Optimizing testing of methicillin-resistant Staphylococcus species. Diagn Microbiol Infect Dis 1994;19:167–170.

 92.  Comité de l’Antibiogramme de la Société Française de Microbiologie. Recommandations 2011. http://www.sfm-microbiologie.org/UserFiles/file/CASFM/casfm_2011.pdf. Accessed March 1, 2013.

 93.  Barry AL, Fuchs PC, Gerlach EH, et al. Multilaboratory evaluation of an agar diffusion disk susceptibility test for rapidly growing anaerobic bacteria. Rev Infect Dis 1990;12(Suppl 2):S210–S217.

 94.  Horn R, Bourgault A-M, Lamothe F. Disk diffusion susceptibility testing of the Bacteroides fragilis group. Antimicrob Agents Chemother 1987;31:1596–1599.

 95.  Sutter VL, Kwok YY, Finegold SM. Standardized antimicrobial disc testing of anaerobic bacteria, I: susceptibility of Bacteroides fragilis to tetracycline. Appl Microbiol 1973;3:188–193.

 96.  King A. Recommendations for susceptibility tests on fastidious organisms and those requiring special handling. J Antimicrob Chemother 2001;48(Suppl 1):77–80.

 97.  Hachem CY, Clarridge JE, Reddy R, et al. Antimicrobial susceptibility testing of Helicobacter pylori. Diagn Microbiol Infect Dis 1996;24:37–41.

 98.  Iovene MR, Romano M, Pilloni AP, et al. Prevalence of antimicrobial resistance in eighty clinical isolates of Helicobacter pylori. Chemotherapy 1999;45:8–14.

 99.  Midolo PD, Bell JM, Lambert JR, et al. Antimicrobial resistance testing of Helicobacter pylori: a comparison of Etest and disk diffusion methods. Pathology 1997;29:411–414.

100.  McNulty C, Owen R, Tompkins D, et al. Helicobacter pylori susceptibility testing by disc diffusion. J Antimicrob Chemother 2002;49:601–609.

101.  Smith C, Perkins J, Tompkins D. Conparison of Etest and disc diffusion for detection of antibiotic resistance in Helicobacter pylori. Microbiol Digest 1997;14:21–23.

102.  Weiss K, Laverdiere M, Restieri C. Comparison of 10 antibiotics against clinical strains of Helicobacter pylori by three different techniques. Can J Gastroenterol 1998;12:181–185.

103.  Shah PM. Recording zone diameters—a rare phenomenon in microbiology laboratories? [letter]. Int J Antimicrob Agents 2004;23:208.

104.  Metzler CM, DeHaan RM. Susceptibility tests of anaerobic bacteria: statistical and clinical considerations. J Infect Dis 1974;130:588–594.

105.  Brunden MN, Zurenko GE, Kapik B. Modification of the error-rate bounded classification scheme for use with two MIC break points. Diagn Microbiol Infect Dis 1992;15:135–140.

106.  Clinical and Laboratory Standards Institute. Development of in vitro susceptibility testing criteria and quality control parameters—third edition. Wayne, PA: Clinical and Laboratory Standards Institute, 2008. CLSI document M23-A3.

107.  Craig BA. Modeling approach to diameter breakpoint determinations. Diagn Microbiol Infect Dis 2000;36:193–202.

108.  DePalma G, Craig BA. dBETS—diffusion Breakpoint Estimation Testing Software. http://glimmer.rstudio.com/dbets/dBETS/. Accessed March 1, 2013.

109.  Barry AL, Joyce LJ, Adams AP, et al. Rapid determination of antimicrobial susceptibility for urgent clinical situations. Am J Clin Path 1973;59:693–699.

110.  Boyle VJ, Faucher ME, Ross RW. Rapid modified Kirby Bauer susceptibility test with single high concentration antimicrobial discs. Antimicrob Agents Chemother 1973;3:418–424.

111.  Kluge RM. Accuracy of Kirby-Bauer susceptibility tests read at 4, 8 and 12 hours of incubation: comparison with reading at 18 to 20 hours. Antimicrob Agents Chemother 1975;8:139–145.

112.  Saha SK, Darmstadt GL, Baqui AH, et al. Rapid identification and antibiotic susceptibility testing of Salmonella enterica serovar typhi isolated from blood: implications for therapy. J Clin Microbiol 2001;39:3583–3585.

113.  Shahidi M, Ellner PD. Effect of mixed cultures on antibiotic susceptibility testing. Appl Microbiol 1969;18:766–770.

114.  Ellner PD, Johnson E. Unreliability of direct susceptibility testing on wound exudates. Antimicrob Agents Chemother 1976;9:355–356.

115.  Doern GV, Vautour R, Gaudet M, et al. Clinical impact of rapid in vitro susceptibility testing and bacterial identification. J Clin Microbiol 1994;32:1757–1762.

116.  Bacteriology Committee of the Association of Clinical Pathologists. Report on antibiotic sensitivity test trial organized by the Bacteriology Committee of the Association of Clinical Pathologists. J Clin Pathol 1965;18:1–5.

117.  Perez JR, Gillenwater JY. Clinical evaluation of testing immediate antibiotic disk sensitivities in bacteriuria. J Urol 1973;110:452–456.

118.  Dornbusch K, Nord C-E, Olsson B, et al. Antibacterial susceptibility testing by the dip-slide technique: a methodological evaluation. Chemotherapy (Basel) 1976;22:190–202.

119.  Hollick GE, Washington JA. Comparison of direct and standardized disk diffusion susceptibility testing of urine cultures. Antimicrob Agents Chemother 1976;9:804–809.

120.  Waterworth PM, Del Piano MD. Dependability of sensitivity tests in primary culture. J Clin Pathol 1976;29:179–184.

121.  Dornbusch K, Lindeberg B, Nord C-E, et al. Bacteriuria diagnosis and antibiotic susceptibility testing in a group practice by dip-slide techniques. Chemotherapy (Basel) 1979;25:227–232.

122.  Källenius G, Dornbusch K, Hallander HO, et al. Comparison of direct and standardized antibiotic susceptibility testing in bacteriuria. Chemotherapy (Basel) 1981;27:99–105.

123.  Scully PG, O’Shea B, Flanagan KP, et al. Urinary tract infection in general practice: direct antibiotic sensitivity testing as a potential diagnostic method. Irish J Med Sci 1990;159:98–100.

124.  Blue AP, Gordon DL. Is primary sensitivity testing on urine samples valid? Pathology 1991;23:149–152.

125.  Jenschke WJ, Trevino E, Vaqnce PH, et al. Comparison of direct urine and standard Kirby-Bauer sensitivities. In: Abstracts of the 93rd general meeting of the American Society for Microbiology. Washington, DC: American Society for Microbiology; 1993. Abstract C-190.

126.  Mukerjee C, Reiss-Levy E. Evaluation of direct disc diffusion susceptibility testing for bacteriuria using diluted urine against the standard CDS method. Pathology 1994;26:201–207.

127.  Oakes AR, Badger R, Grove DI. Comparison of direct and standardized testing of infected urine for antimicrobial susceptibilities by disk diffusion. J Clin Microbiol 1994;32:40–45.

128.  Johnson JR, Tiu FS, Stamm WE. Direct antimicrobial susceptibility testing for acute urinary tract infection in women. J Clin Microbiol 1995;33:2316–2323.

129.  Gillenwater JY, Clark MM. Tentative direct antimicrobial susceptibility testing in urine. J Urol 1996;156:149–153.

130.  Bronnestam R. Direct antimicrobial susceptibility testing in bacteriuria. APMIS 1999;107:437–444.

131.  Breteler KB, Rentenaar RJ, Verkaart G, et al. Performance and clinical significance of direct antimicrobial susceptibility testing of urine from hospitalized patients. Scand J Infec Dis 2011;43:771–776.

132.  Heinze PA, Thrupp LD, Anselmo CR. A rapid (4–6 hour) urine-culture system for direct identification and direct antimicrobial susceptibility testing. Am J Clin Pathol 1979;71:177–183.

133.  Johnson JE, Washingon JA. Comparison of direct and standardized antimicrobial susceptibility testing of positive blood cultures. Antimicrob Agents Chemother 1976;10:211–214.

134.  Wegner DL, Mathis CR, Neblett TR. Direct method to determine the antibiotic susceptibility of rapidly growing blood pathogens. Antimicrob Agents Chemother 1976;9:861–862.

135.  Fay D, Oldfather JE. Standardization of direct susceptibility test for blood cultures. J Clin Microbiol 1979;9:347–350.

136.  Mirrett S, Reller LB. Comparison of direct and standard antimicrobial disk susceptibility testing for bacteria isolated from blood. J Clin Microbiol 1979;10:482–487.

137.  Coyle MB, McGonagle LA, Plorde JJ, et al. Rapid antimicrobial susceptibility testing of isolates from blood cultures by direct inoculation and early reading of disk diffusion tests. J Clin Microbiol 1984;20:473–477.

138.  Edelmann A, Pietzcker T, Wellinghausen N. Comparison of direct disk diffusion and standard microtitre broth dilution testing of blood culture isolates. J Med Microbiol 2007;56(Pt 2):202–207.

139.  Tan TY, Ng LS, Kwang LL. Evaluation of disc susceptibility tests performed directly from positive blood cultures. J Clin Pathol 2008;61:343–346.

140.  Bennett K, Sharp SE. Rapid differentiation of methicillin-resistant Staphylococcus aureus and methicillin-susceptible Staphylococcus aureus from blood cultures by use of a direct cefoxitin disk diffusion test. J Clin Microbiol 2008;46:3836–3838.

141.  Cuellar-Rodriguez JM, Ponce-de-León A, Quiroz-Mejia R, et al. Rapid detection of ESBL-producing gram-negative bacteria isolated from blood: a reasonable and reliable tool for middle and low resource countries. Rev Invest Clin 2009;61:306–312.

142.  Miño de Kaspar H, Neubauer AS, Molnar A, et al. Rapid direct antibiotic susceptibility testing in endophthalmitis. Ophthalmology 2002;109:687–693.

143.  Kontopidou F, Galani I, Panagea T, et al. Comparison of direct antimicrobial susceptibility testing methods for rapid analysis of bronchial secretion samples in ventilator-associated pneumonia. Int J Antimicrob Agents 2011;38:130–134.

144.  Brown DF, Brown L. Evaluation of the Etest, a novel method of quantifying antimicrobial activity. J Antimicrob Chemother 1991;27:185–190.

145.  Jorgensen JH, Howell AW, Maher LA. Quantitative antimicrobial susceptibility testing of Haemophilus influenzae and Streptococcus pneumoniae by using the E-test. J Clin Microbiol 1991;29:109–114.

146.  Hunt GS, Citron DM, Claros MC, et al. Comparison of Etest to broth microdilution method for testing Streptococcus pneumoniae susceptibility to levofloxacin and three macrolides. Antimicrob Agents Chemother 1996;40:2413–2415.

147.  Hunt DE, Sandham HJ. Improved agar gradient-plate technique. Appl Microbiol 1969;17:329–330.

148.  Szybalski W. Microbial selection, I: gradient plate technique for study of bacterial resistance. Science 1952;116:46–48.

149.  Carsenti-Etesse H, Roger PM, Dunais B, et al. Gradient plate method to induce Streptococcus pyogenes resistance. J Antimicrob Chemother 1999;44:439–443.

150.  Wootton M, Howe RA, Hillman R, et al. A modified population analysis profile (PAP) method to detect hetero-resistance to vancomycin in Staphylococcus aureus in a UK hospital. J Antimicrob Chemother 2001;47:399–403.

151.  Liu Y, Li J, Du J, et al. Accurate assessment of antibiotic susceptibility and screening resistant strains of a bacterial population by linear gradient plate. Sci China Life Sci 2011;54:953–960.

152.  Gerber AU, Vastola AP, Brandel J, et al. Selection of aminoglycoside-resistant variants of Pseudomonas aeruginosa in an in vivo model. J Infect Dis 1982;146:691–697.

153.  Berge-Bachi B, Strassle A, Kayser FH. Characterization of an isogenic set of methicillin-resistant and susceptible mutants of Staphylococcus aureus. Eur J Clin Microbiol 1986;5:697–701.

154.  Hiramatsu K, Arikata N, Hanaki H, et al. Dissemination in Japanese hospitals of strains of Staphylococcus aureus heterogeneously resistant to vancomycin. Lancet 1997;350:1670–1673.

155.  Walsh TR, Bolmström A, Qwärnström A, et al. Evaluation of current methods for detection of staphylococci with reduced susceptibility to glycopeptides. J Clin Microbiol 2001;39:2439–2444.

156.  Howden BP, Davies JK, Johnson PD, et al. Reduced vancomycin susceptibility in Staphylococcus aureus, including vancomycin-intermediate and heterogeneous vancomycin-intermediate strains: resistance mechanisms, laboratory detection, and clinical implications. Clin Microbiol Rev 2010;23:99–139.

157.  Lorian V. Low concentrations of antibiotics. J Antimicrob Chemother 1985;15(Suppl A):15–26.

158.  Lorian V. Differences between in vitro and in vivo studies. Antimicrob Agents Chemother 1988;32:1600–1601.

159.  Odenholt I. Pharmacodynamic effects of subinhibitory antibiotic concentrations. Int J Antimicrob Agents 2001;17:1–8.

160.  Craig WA. Post-antibiotic effects in experimental animal models: relationship to in-vitro phenomena and the treatment of infections in man. J Antimicrob Chemother 1993;31(Suppl D):149–158.

161.  MacKenzie FM, Gould IM. The post-antibiotic effect. J Antimicrob Chemother 1993;31(Suppl D):519–537.

162.  Cars O, Odenholt-Tornqvist I. The post-antibiotic sub-MIC effect in vitro and in vivo. J Antimicrob Chemother 1993;31(Suppl D):159–166.

163.  Pruul H, McDonald PJ. Damage to bacteria by antibiotics in vitro and its relevance to antimicrobial chemotherapy. J Antimicrob Chemother 1988;21:695–698.

164.  Lorian V, Burns L. Predictive value of susceptibility tests for the outcome of antibacterial therapy. J Antimicrob Chemother 1990;25:175–181.

166.  Craig WA. Basic pharmacodynamics of antibacterials with clinical applications to the use of β-lactams, glycopeptides and linezolid. Infect Dis Clin North Am 2003;17:479–501.

167.  Drusano GL. Antimicrobial pharmacodynamics: critical interactions of “bug and drug.” Nat Rev Microbiol 2004;2:289–300.

168.  Drusano GL. Pharmacokinetics and pharmacodynamics of antimicrobials. Clin Infect Dis 2007;45(Suppl 1):S89–S95.

169.  Ambrose PG, Bhavnani SM, Rubino CM, et al. Pharmacokinetics-pharmacodynamics of antimicrobial therapy: it’s not just for mice anymore. Clin Infect Dis 2007;44:79–86.

170.  McNabb JJ, Bui KQ. β-lactam pharmacodynamics. In: Nightingale CH, Murakawa T, Ambrose PG, eds. Antimicrobial pharmacodynamics in theory and clinical practice. New York: Marcel Dekker, 2002:99–123.

171.  Turnidge JD. Pharmacodynamics and dosing of aminoglycosides. Infect Dis Clin North Am 2003;17:503–528.

172.  Nightingale CH, Mattoes HM. Macrolide, azalide and ketolide pharmacodynamics. In: Nightingale CH, Murakawa T, Ambrose PG, eds. Antimicrobial pharmacodynamics in theory and clinical practice. New York: Marcel Dekker, 2002:205–220.

173.  Maglio D, Nicolau DP, Nightingale CH. Impact of pharmacodynamics on dosing of macrolides, azalides and ketolides. Infect Dis Clin North Am 2003;17:563–577.

174.  Owens RC Jr, Ambrose PG. Pharmacodynamics of quinolones. In: Nightingale CH, Murakawa T, Ambrose PG, eds. Antimicrobial pharmacodynamics in theory and clinical practice. New York: Marcel Dekker, 2002:155–178.

175.  Ambrose PG, Bhavnani SM, Owens RC. Clinical pharmacodynamics of quinolones. Infect Dis Clin North Am 2003;17:529–543.

176.  Ross GH, Wright DH, Rotschafer JC, et al. Glycopeptide pharmacodynamics. In: Nightingale CH, Murakawa T, Ambrose PG, eds. Antimicrobial pharmacodynamics in theory and clinical practice. New York: Marcel Dekker, 2002:177–204.

177.  Hermsen ED, Sullivan CJ, Rotschafer JC. Polymyxins: pharmacology, pharmacokinetics, pharmacodynamics and clinical applications. Infect Dis Clin North Am 2003;17:545–562.

178.  Lamp K, Lacy MK, Freeman C. Metronidazole, clindamycin, and streptogramin pharmacodynamics. In: Nightingale CH, Murakawa T, Ambrose PG, eds. Antimicrobial pharmacodynamics in theory and clinical practice. New York: Marcel Dekker, 2002:221–246.

179.  Cunha BA, Mattoes HM. Tetracycline pharmacodynamics. In: Nightingale CH, Murakawa T, Ambrose PG, eds. Antimicrobial pharmacodynamics in theory and clinical practice. New York: Marcel Dekker, 2002:247–257.

180.  Harding I, MacGowan AP, White LO, et al. Teicoplanin therapy for Staphylococcus aureus septicaemia: relationship between pre-dose serum concentrations and outcome. J Antimicrob Chemother 2000;45:835–841.

181.  Snydman DR, Cucheral GJ Jr, McDermott L, et al. Correlation of various in vitro testing methods with clinical outcomes in patients with Bacteroides fragilis group infections treated with cefoxitin: a retrospective analysis. Antimicrob Agents Chemother 1992;36:540–544.

182.  Pillai SK, Moellering RC Jr, Eliopoulos GM. Antimicrobial combinations. In: Lorian V, ed. Antibiotics in laboratory medicine. Philadelphia: Lippincott Williams & Wilkins, 2005:365–424.

183.  Turnidge J. Drug-drug combinations. In: Vinks A, Derendorf H, Mouton J, eds. Fundamentals of antimicrobial pharmacokinetics and pharmacodynamics. New York: Springer, 2013:153–198.

184.  European Committee on Antimicrobial Susceptibility Testing. Breakpoint tables for interpretation of MICs and zone diameters, version 3.1. http://www.eucast.org/fileadmin/src/media/PDFs/EUCAST_files/Disk_test_documents/Manual_v_2.1_EUCAST_Disk_Test.pdf. Accessed March 1, 2013.

185.  White RL, Burgess DS, Manduru M, et al. Comparison of three different in vitro methods of detecting synergy: time-kill, checkerboard, and E test. Antimicrob Agents Chemother 1996;40(8):1914–1918.

186.  Bonapace C, White R, Friedrich L, et al. Evaluation of antibiotic synergy against Acinetobacter baumannii: comparison with Etest, time kill and checkerboard methods. Diagn Microbiol Infect Dis 2000;38:43–50.

187.  Wareham DW, Bean DC. In-vitro activity of polymyxin B in combination with imipenem, rifampicin and azithromycin versus multidrug resistant strains of Acinetobacter baumannii producing Oxa-23 carbapenemase. Ann Clin Microbiol Antimicrob 2006;5:10.

188.  Tan TY, Ng LS, Tan E, et al. In vitro effect of minocycline and colistin combinations on imipenem-resistant Acinetobacter baumannii clinical isolates. J Antimicrob Chemother 2007;60:421–423.

189.  Pankey GA, Ashcraft DS. The detection of synergy between meropenem and polymyxin B against meropenem-resistant Acinetobacter baumannii using Etest and time-kill assay. Diagn Microbiol Infect Dis 2009;63:228–232.

190.  Gordon NC, Png K, Wareham DW. Potent synergy and sustained bactericidal activity of a vancomycin-colistin combination versus multidrug-resistant strains of Acinetobacter baumannii. Antimicrob Agents Chemother 2010;54:5613–5622.

191.  Sopirala M, Mangino JE, Gebreyes WA, et al. Synergy testing by Etest, microdilution checkerboard, and time-till methods for pan-drug-resistant Acinetobacter baumannii. Antimicrob Agents Chemother 2010;54:4678–4683.

192.  Tan TY, Lim TP, Lee WH, et al. In vitro synergy in extensively drug-resistant Acinetobacter baumannii: the effect of testing by time-kill, checkerboard, and Etest methods. Antimicrob Agents Chemother 2011;55:436–438.

193.  Wareham DW, Gordon NC, Hornsey M. In vitro activity of teicoplanin combined with colistion versus multidrug-resistant strains of Acinetobacter baumannii. J Antimicrob Chemother 2011;66:1047–1051.

194.  Miyasaki Y, Morgan MA, Chan RC, et al. In vitro activity of antibiotic combinations against multidrug-resistant strains of Acinetobacter baumannii and the effects of their antibiotic resistance determinants. FEMS Microbiol Lett 2012;328:26–31.

195.  Hornsey M, Longshaw C, Phee L, et al. In vitro activity of televancin in combination with colistin versus Gram-negative bacterial pathogens. Antimicrob Agents Chemother 2012;56:3080–3085.

196.  Pankey G, Ashcraft D, Patel N. In vitro synergy of daptomycin plus rifampicin against Enterococcus faecium resistant to both linezolid and vancomycin. Antimicrob Agents Chemother 2005;49:5166–5168.

197.  Pankey GA, Ashcraft DS. In vitro synergy of telavancin and rifampin against Enterococcus faecium resistant to both linezolid and vancomycin. Ochsner J 2013;13:61–65.

198.  Tsuji BT, Rybak MJ. Etest synergy testing of clinical isolates of Staphylococcus aureus demonstrating heterogeneous resistance to vancomycin. Diagn Microbiol Infect Dis 2006;54:73–77.

199.  Hellmark B, Unemo M, Nilsdottir-Augustinsson A, et al. In vitro antimicrobial synergy testing of coagulase-negative staphylococci isolated from prosthetic joint infections using Etest and with a focus on rifampicin and linezolid. Eur J Clin Microbiol Infect Dis 2010;29:591–595.

200.  Chachanidze V, Curbelo-Irizarry A, Ashcraft D, et al. In vitro synergy of levofloxacin plus piperacillin/tazobactam against Pseudomonas aeruginosa. Interdiscip Perspect Infect Dis 2009;2009:948934.

201.  Samonis G, Maraki S, Karageogopoulos DE, et al. Synergy of fosfomycin with carbapenems, colistin, netilmicin, and tigecycline against multidrug-resistant Klebsiella pneumoniae, Escherichia coli, and Pseudomonas aeruginosa. Eur J Clin Microbiol Infect Dis 2012;31:695–701.

202.  He W, Kaniga K, Lynch AS, et al. In vitro Etest synergy of doripenem with amikacin, colistin, and levofloxacin against Pseudomonas aeruginosa with defined carbapenem resistance mechanisms as determined by the Etest method. Diagn Microbiol Infect Dis 2012;74:417–419.

203.  Pankey GA, Ashcraft DS. Detection of synergy using the combination of polymyxin B with either meropenem or rifampin against carbapenemase-producing Klebsiella pneumoniae. Diagn Microbiol Infect Dis 2011;70:561–564.

204.  Church D, LLloyd T, Peirano G, et al. Antimicrobial susceptibility and combination testing of invasive Stenotrophomonas maltophilia isolates. Scand J Infect Dis 2013;45:265–270.