aiwiki.page
English
Health / Medicine & Treatment / antimicrobial-susceptibility-testing

Antimicrobial Susceptibility Testing

Antimicrobial susceptibility testing measures how microorganisms respond to antimicrobial agents, supporting treatment selection, resistance detection, and surveillance.

36 keywords12 linked from8 not yet writtenWritten by AI
BacteriaAntibioticDiagnosisInfectious Disea…Antimicrobial Re…PharmacologyBeta-Lactam Anti…EnzymeAntimicrob…

Antimicrobial susceptibility testing (AST) comprises laboratory methods that determine how a microorganism responds to an antimicrobial agent under defined conditions. Its most familiar application is testing bacterial isolates against antibiotics, but the broader field includes susceptibility testing of fungi, mycobacteria, and some viruses and parasites. Results may be expressed as a measured inhibitory concentration, an inhibition-zone diameter, or an interpretive category such as susceptible or resistant. These measurements help connect the identification of an infectious organism with the selection of drugs likely to be active against it.

AST is central to the diagnosis and management of infectious disease, the detection of antimicrobial resistance, and the monitoring of resistance across populations. It does not directly establish whether a drug will cure an infection: laboratory results must be interpreted alongside drug exposure, infection site, microbial identity, and host factors. Its scientific basis therefore spans microbiology, pharmacology, and clinical medicine.

Modern susceptibility testing combines standardized growth-based methods, automated instruments, targeted resistance assays, and increasingly rapid or genomic approaches. The reliability of any result depends on the specimen, the organism selected for testing, the technical method, and the interpretive standards applied.

Biological basis and scope

Susceptibility is a relationship among an organism, a drug, and specified test conditions—not an absolute property of a microbial species. An antimicrobial may inhibit an isolate at one concentration but not another, and measured activity can change with the growth medium, incubation atmosphere, inoculum, and endpoint definition. Standardization makes results reproducible enough to support clinical interpretation. Without it, differences between laboratories could reflect experimental conditions rather than biologically meaningful differences between isolates.

Antimicrobial agents act on different cellular processes. Beta-lactam antibiotics interfere with bacterial cell-wall synthesis; aminoglycosides and several other classes inhibit protein synthesis; fluoroquinolones affect DNA-associated enzymes; and polymyxins disrupt bacterial membranes. Resistance can arise through drug-inactivating enzymes, altered targets, reduced permeability, active efflux, or bypass of an inhibited pathway. These mechanisms help explain characteristic susceptibility patterns, although a pattern alone does not always identify its molecular cause.

Intrinsic resistance is a predictable property of a species or larger taxonomic group, whereas acquired resistance develops through mutation or acquisition of genetic material. For example, the absence of a conventional peptidoglycan cell wall explains why Mycoplasma species are not susceptible to beta-lactams. Acquired resistance may vary sharply among isolates of the same species. This is why species identification sometimes predicts inactivity, but often cannot replace testing of an individual isolate.

Routine clinical AST primarily measures inhibition of free-living organisms growing under standardized laboratory conditions. This differs from microbial behavior within a biofilm, an abscess, an intracellular compartment, or damaged tissue. Organisms in these environments may experience reduced drug exposure or altered growth states. Antibiotic tolerance and persistence also differ from resistance: tolerant populations or persister subpopulations can survive drug exposure without necessarily having a higher inhibitory concentration.

The scope of testing is organism- and drug-specific. An assay suitable for common aerobic bacteria may be unsuitable for anaerobes, fungi, or mycobacteria. Viruses require different experimental systems because they replicate within host cells; viral resistance testing frequently examines genetic changes associated with reduced drug activity. Even within bacteriology, some organism–drug combinations lack sufficiently validated methods or clinical interpretive criteria. A numerical result therefore does not automatically constitute a clinically interpretable susceptibility result.

Specimens, isolation, and the laboratory workflow

Susceptibility testing usually begins with a clinical specimen, such as blood, urine, respiratory material, tissue, or cerebrospinal fluid. Collection quality matters because the organisms recovered may represent infection, colonization, or contamination. AST characterizes the tested organism; it does not establish that the organism caused the patient’s illness. For example, a susceptibility profile from a contaminated blood culture may be analytically correct but clinically misleading if the isolate is incorrectly treated as a pathogen.

Conventional workflows use microbial culture to recover organisms, identify them, and obtain a sufficiently pure population for testing. Mixed cultures require particular attention: separate organisms can have different susceptibilities, and a mixed inoculum may produce an uninterpretable result. Identification may depend on biochemical reactions, targeted molecular assays, or matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. Accurate identification determines which drugs, methods, and interpretive tables are applicable.

The inoculum is standardized because organism density influences the apparent activity of many drugs. For routine testing of common aerobic bacteria, preparation often begins with a suspension matching a 0.5 McFarland turbidity standard, although the corresponding cell concentration varies with the organism. Reference broth microdilution generally uses a final inoculum near 5×1055 \times 10^5 colony-forming units per milliliter. These values describe standardized methods, not universal requirements for every microorganism.

The laboratory then exposes the organism to antimicrobial agents in broth, on agar, or within an instrument-specific system. Growth controls demonstrate that the organism can grow under the test conditions, while purity checks and quality-control procedures help identify technical failure. Many routine bacterial methods require approximately an overnight incubation after preparation of the isolate. Total clinical turnaround time is longer when it includes specimen transport, initial culture, identification, and reporting.

Testing directly from positive blood-culture bottles or other specimens can shorten this sequence, but direct testing requires specific validation. Blood components, prior antimicrobial exposure, mixed populations, and variable organism density can interfere with assays designed for purified colonies. A result validated for one specimen type, incubation interval, or microbial group cannot simply be transferred to another.

Drug-panel design is also part of the workflow. Laboratories select agents relevant to the organism, infection setting, local practice, and resistance epidemiology. Reports may omit intrinsically inactive agents or selectively release results through cascade reporting. This distinguishes the drugs actually tested from those displayed to clinicians and prevents an extensive laboratory panel from being mistaken for a list of equally appropriate therapeutic choices.

Dilution methods and quantitative endpoints

The minimum inhibitory concentration (MIC) is the lowest tested concentration of an antimicrobial that prevents a defined degree of visible growth under specified conditions. For many conventional bacterial assays, the endpoint is complete visible inhibition, although some agents and organisms require different reading rules. MICs are commonly reported in milligrams per liter or micrograms per milliliter; these units are numerically equivalent:

1 mg/L=1 μg/mL.1\ \mathrm{mg/L}=1\ \mathrm{\mu g/mL}.

In broth macrodilution, antimicrobial concentrations are prepared in tubes. In broth microdilution, the same principle is miniaturized into microplate wells. Serial twofold concentrations—such as 0.25, 0.5, 1, 2, and 4 mg/L—are typical. Each concentration receives a standardized inoculum, and growth is compared with an antimicrobial-free control after incubation. Broth microdilution is a reference method for many organism–drug combinations, although specialized media or conditions are required for some agents.

An MIC is a discretized endpoint rather than an exact physical constant. If growth occurs at 1 mg/L but not at 2 mg/L, the reported MIC is 2 mg/L, even though the transition lies somewhere within that tested interval. Values at the edges of a panel are censored: “≤0.25 mg/L” means that the true endpoint was not resolved below the lowest concentration tested, while “>16 mg/L” means that inhibition was not demonstrated within the tested range. Repeated measurements commonly differ by one twofold dilution.

Agar dilution incorporates different antimicrobial concentrations into separate agar plates, onto which multiple isolates can be inoculated in defined spots. It is labor-intensive but useful for reference work and selected organisms or agents. Gradient diffusion places a strip containing a concentration gradient on an inoculated agar surface. The MIC is read where the inhibition ellipse intersects the strip’s scale. Gradient methods are convenient, but performance is drug- and organism-dependent and is not automatically equivalent to reference dilution.

The minimum bactericidal concentration (MBC) addresses killing rather than growth inhibition. It is commonly defined as the lowest concentration producing at least a 99.9% reduction in the initial viable bacterial population under specified conditions. Determination requires additional viability assessment, usually by subculture from apparently clear broth. Carryover of antimicrobial into the recovery medium can distort the endpoint if not controlled.

MIC and MBC answer different questions, and routine susceptibility reports usually provide MICs or categories rather than bactericidal measurements. A low MIC does not establish rapid killing. Conversely, differences in killing kinetics may not be captured by an overnight inhibitory endpoint. Such distinctions are important in research on tolerance, persistence, and combination activity.

Disk diffusion and automated phenotypic systems

Disk diffusion measures the inhibition of growth around a paper disk containing a specified quantity of antimicrobial. A standardized inoculum is spread to form a bacterial lawn on agar, disks are applied, and the plate is incubated under defined conditions. Antimicrobial diffuses outward from each disk, establishing a concentration gradient. The diameter of the resulting inhibition zone is measured in millimeters and interpreted using organism- and drug-specific criteria.

The widely used standardized disk-diffusion procedure is often called the Kirby–Bauer method. Its modern foundation includes the 1966 publication by Alfred Bauer, William Kirby, John Sherris, and Marvin Turck describing a standardized single-disk method. Mueller–Hinton agar became central to testing many nonfastidious bacteria because it supports reproducible growth and diffusion when its composition and physical characteristics are controlled. Fastidious organisms may require supplemented media and different atmospheric conditions.

Zone size is influenced by more than microbial susceptibility. Agar depth, pH, cation content, disk potency, inoculum density, and incubation duration can all affect the result. A larger zone generally indicates greater inhibition for a particular organism–drug system, but zone diameters cannot be compared indiscriminately across different drugs. A 25-mm zone around one antimicrobial disk is not necessarily evidence of greater clinical activity than a 20-mm zone around another.

Disk diffusion generally produces categorical results rather than measured MICs. Interpretive zone thresholds are established through studies relating zone diameters to MIC distributions, resistance mechanisms, and clinical breakpoints. Colonies growing within an inhibition zone, indistinct edges, swarming, or particular growth patterns may require method-specific reading rules. Some organisms or agents are unsuitable for disk diffusion; colistin is an important example because poor agar diffusion makes conventional diffusion methods unreliable for its susceptibility assessment.

Automated phenotypic systems usually expose organisms to antimicrobial concentrations in cards, panels, or wells and monitor growth through optical, fluorescent, or other signals. Software converts growth measurements into MIC estimates and interpretive categories. Such systems can reduce hands-on work, standardize data handling, and interface with laboratory information systems. Their concentration ranges and drug menus, however, are finite and may differ from reference methods.

Instrument results remain dependent on identification, panel design, software, and validation. Unusual resistance profiles may trigger expert rules or additional testing. An automated system may miss a resistance phenotype if its tested concentrations, growth-detection approach, or incubation period are poorly suited to that organism–drug combination. Consequently, automation changes how susceptibility is measured but does not remove the need for microbiological review and quality assurance.

Clinical breakpoints and interpretation

A clinical breakpoint is a threshold used to translate an MIC or inhibition-zone diameter into an interpretive category. Breakpoints integrate several kinds of evidence: MIC distributions, known resistance mechanisms, achievable drug exposure, pharmacokinetic–pharmacodynamic relationships, dosing regimens, and clinical outcomes. They are not simply the average MIC of a species, nor are they fixed biological boundaries separating all successful treatments from all failures.

Major standards-setting organizations include the Clinical and Laboratory Standards Institute (CLSI) and the European Committee on Antimicrobial Susceptibility Testing (EUCAST). Their methods and interpretive frameworks are extensively used but are not interchangeable in every detail. A report’s meaning depends on which system, version, and organism–drug criteria were applied. Breakpoint revisions can change an isolate’s category even when its measured MIC is unchanged.

EUCAST defines S as “susceptible, standard dosing regimen,” I as “susceptible, increased exposure,” and R as “resistant.” Since the 2019 revision of these definitions, EUCAST’s I category has explicitly represented susceptibility when increased exposure is achieved; it is not synonymous with an indeterminate result. CLSI uses susceptible, intermediate, and resistant categories and, for certain combinations, “susceptible-dose dependent.” These terms carry system-specific assumptions and should not be treated as identical labels.

Exposure links laboratory inhibition to drug behavior in the body. Pharmacokinetics describes concentration over time, while pharmacodynamics describes the relationship between exposure and antimicrobial effect. Common indices include the fraction of time that unbound drug remains above the MIC, the ratio of peak concentration to MIC, and the ratio of area under the concentration–time curve to MIC:

fT>MIC,Cmax⁡MIC,AUCMIC.fT>\mathrm{MIC}, \qquad \frac{C_{\max}}{\mathrm{MIC}}, \qquad \frac{\mathrm{AUC}}{\mathrm{MIC}}.

Which index is most informative depends on the antimicrobial and microbial system. Infection site also matters: concentrations in urine, blood, lung tissue, and the central nervous system can differ substantially. Some breakpoint tables therefore contain site-specific criteria, as with selected agents used for meningitis, or restrictions concerning uncomplicated urinary tract infection.

An epidemiological cutoff value serves a different purpose. It separates the wild-type MIC distribution—organisms without phenotypically detectable acquired resistance to the agent—from non-wild-type populations. It does not, by itself, establish clinical susceptibility. An organism may be wild type yet fall outside clinically treatable exposure ranges, while some non-wild-type isolates may remain within a susceptible breakpoint. Confusing epidemiological cutoffs with clinical breakpoints obscures the distinction between detecting resistance biology and predicting therapeutic activity.

Detecting resistance mechanisms and difficult phenotypes

Phenotypic susceptibility testing establishes observable drug response, whereas mechanism-focused testing seeks the reason for that response. The two approaches overlap but are not equivalent. A resistance mechanism may be present without producing resistance to every drug in a class, and similar MIC patterns may arise through different mechanisms. Laboratories therefore use selected supplementary tests when the mechanism affects interpretation, infection control, or epidemiological investigation.

Methicillin resistance in Staphylococcus aureus illustrates this relationship. Resistance is commonly associated with altered penicillin-binding proteins encoded by mecA or mecC. Cefoxitin-based phenotypic testing is widely used to detect the relevant resistance phenotype, while molecular tests or assays for the altered binding protein can provide mechanism-specific evidence. Interpretive rules then extend the finding to relevant beta-lactams, with recognized exceptions for agents specifically active against methicillin-resistant strains.

Inducible clindamycin resistance is another example. Some staphylococci and streptococci appear resistant to erythromycin but susceptible to clindamycin in routine testing. When erythromycin and clindamycin disks are placed at a specified distance, induction can produce flattening of the clindamycin zone next to erythromycin—the characteristic D-shaped pattern. This detects an inducible resistance phenotype that a clindamycin result considered alone could miss.

Among Gram-negative bacteria, extended-spectrum beta-lactamases, AmpC enzymes, and carbapenemases can produce overlapping susceptibility patterns. Carbapenem resistance may also result from combinations of reduced permeability and other beta-lactamases. Carbapenemase detection can use inhibitor-based methods, hydrolysis assays, immunochromatographic tests, molecular assays, or procedures such as the modified carbapenem inactivation method. Mechanism identification is particularly relevant when activity of a beta-lactamase inhibitor combination depends on the enzyme class involved.

Heteroresistance presents a different challenge: a minority subpopulation is substantially less susceptible than the predominant population. Standard inocula or abbreviated detection systems may fail to sample or recognize that minority. Population-analysis methods can characterize such distributions but are not routine for most clinical isolates. Heteroresistance should also be distinguished from contamination, mixed-species growth, and the emergence of a stable resistant mutant.

Other difficult patterns include trailing growth, skipped wells, paradoxical effects at high concentrations, and disagreement between screening and confirmatory assays. Their significance depends on the organism, drug, and method. Expert interpretation relies on validated rules rather than the assumption that every unusual pattern represents a new resistance mechanism. Supplementary mechanism tests also do not universally replace the susceptibility measurement needed for a particular antimicrobial.

Quality assurance and measurement limitations

AST is a standardized biological measurement, and quality assurance must address both technical reproducibility and clinical interpretation. Control organisms with characterized susceptibility profiles are tested against defined acceptance ranges. Common examples include Escherichia coli ATCC 25922, Staphylococcus aureus ATCC 29213 for dilution testing, and Pseudomonas aeruginosa ATCC 27853. Appropriate control strains differ by method, antimicrobial, and resistance mechanism being assessed.

Quality control can reveal deteriorated disks, incorrect medium composition, faulty inoculum preparation, or instrument problems. It does not establish that every clinical isolate has been correctly identified or that every result is clinically applicable. Broader quality assurance includes staff competency, lot verification, equipment maintenance, validated procedures, proficiency testing, and accurate transfer of results into the reporting system. Preanalytical and postanalytical errors can undermine an otherwise accurate measurement.

Method-comparison studies commonly assess essential agreement and categorical agreement. Essential agreement usually means that MICs differ by no more than one twofold dilution from a comparator method. Categorical agreement means that both methods assign the same interpretive category. These are different properties: close MIC agreement can still produce category disagreement when the values lie near a breakpoint, while substantially different MICs can occasionally fall within the same broad category.

A very major error generally denotes a false-susceptible result compared with the reference method; a major error denotes a false-resistant result. Minor errors involve intermediate or related category discrepancies under the applicable evaluation scheme. Error rates require clearly specified denominators and an adequate selection of resistant and susceptible isolates. Overall agreement can appear high in a dataset dominated by easily classified susceptible organisms while concealing poor detection of clinically important resistance.

Measurement uncertainty is especially relevant near category boundaries. Because MICs commonly vary by a dilution, an isolate close to a breakpoint may change category on repeat testing without any underlying biological change. EUCAST’s area of technical uncertainty, where designated, identifies particular test-result ranges associated with technical interpretation problems. It is a laboratory warning construct, not an additional clinical susceptibility category.

Some agents require unusually strict methodological control. Polymyxins can interact with laboratory materials, and diffusion methods are unsuitable for reliable colistin testing. Daptomycin activity depends on calcium concentration, making controlled calcium supplementation important in broth testing. Such examples show why a method validated for one antimicrobial cannot automatically be assumed valid for another. Updated standards, rather than a single universal procedure, underpin comparability between laboratories.

Fungi, mycobacteria, and other specialized organisms

Antifungal susceptibility testing applies related principles to yeasts and molds but requires distinct media, incubation periods, endpoints, and interpretive criteria. Commonly tested agents include azoles, echinocandins, and amphotericin B. CLSI and EUCAST both provide reference methods, although their procedures and breakpoints can differ. For some organism–drug combinations, clinical breakpoints are available; for others, interpretation is limited to epidemiological cutoffs or descriptive MIC data.

Endpoint definition is particularly important in mycology. Some yeast–drug combinations are assessed using a specified reduction in growth rather than complete clearing. Trailing growth with azoles may complicate visual reading. For molds exposed to echinocandins, the relevant endpoint may be a minimum effective concentration, based on characteristic changes in hyphal growth, rather than a conventional MIC. Applying bacterial reading rules to these assays would therefore be inappropriate.

Resistance in Candida and Aspergillus has increased the importance of antifungal testing. Mutations in FKS genes can be associated with echinocandin resistance in Candida, while alterations involving cyp51A are important in azole-resistant Aspergillus fumigatus. Candida auris has attracted particular attention because of healthcare-associated transmission and frequent resistance to one or more antifungal classes. Interpretive criteria for emerging species may be incomplete and can include explicitly tentative thresholds.

Drug susceptibility testing for tuberculosis is shaped by the slow growth of the Mycobacterium tuberculosis complex, biosafety requirements, and the use of multidrug regimens. Phenotypic methods assess growth in solid or liquid media containing specified drug concentrations. Historical proportion methods classify resistance according to the fraction of the bacterial population able to grow at a defined concentration, rather than using the routine framework applied to rapidly growing aerobic bacteria.

Molecular tuberculosis assays can detect resistance-associated variants much faster than conventional culture-based testing. Rifampicin resistance is commonly associated with mutations in rpoB; isoniazid resistance frequently involves katG or regulatory regions affecting inhA. The relationship between variant, measured susceptibility, and clinical significance is not equally established for every drug. Discordance between molecular and phenotypic results may reflect uncommon variants, low-level resistance, mixed populations, or limitations of the assay’s target coverage.

Anaerobes, Neisseria gonorrhoeae, Helicobacter pylori, nontuberculous mycobacteria, and other specialized organisms also need dedicated procedures. Viral resistance testing often uses sequence interpretation, as in HIV infection, or cell-based phenotypic assays. These systems share the goal of detecting reduced drug activity but cannot be collapsed into the bacterial disk-diffusion or broth-microdilution model.

Rapid testing, molecular prediction, and emerging technologies

Rapid AST seeks to shorten the interval between specimen collection and actionable susceptibility information. Conventional methods often depend on visible population growth over many hours. Newer phenotypic approaches may detect early changes in cell number, morphology, metabolism, or individual-cell behavior. Microfluidic devices, time-lapse microscopy, automated imaging, and sensitive growth measurements can reveal drug effects before a conventional endpoint becomes apparent.

A rapid method must distinguish true inhibition from a temporary lag in growth and must detect clinically relevant resistant populations. The challenge is especially important for organisms that grow slowly, express resistance after induction, or contain minority resistant subpopulations. A short assay can be highly accurate for one organism–drug combination and insufficient for another. Validation therefore concerns the entire combination of specimen, organism, antimicrobial, incubation interval, measurement system, and interpretation rule.

Direct testing from positive blood cultures is a major application because bloodstream infection and sepsis can require time-critical decisions. Some methods use short-incubation disk diffusion with dedicated reading times and zone thresholds; others use microscopy or instrument-specific growth analysis. Results from a four- or six-hour assay cannot simply be interpreted using conventional overnight criteria. Some rapid systems deliberately withhold categories in uncertain ranges rather than force an unsupported classification.

Molecular assays detect resistance-associated genes or variants rather than directly measuring drug-inhibited growth. Targeted tests can quickly identify selected mechanisms, such as carbapenemase genes or methicillin-resistance determinants. Their speed is valuable, but a negative result establishes only that the assay did not detect its specified targets. It does not exclude resistance caused by an untested gene, a novel variant, altered expression, or a non-genetic physiological effect.

Whole-genome sequencing can examine a much wider set of determinants and place resistance within an evolutionary and transmission context. Prediction is often strongest when a well-characterized determinant has a consistent phenotypic effect. It is more difficult when susceptibility depends on interacting mutations, copy number, regulatory changes, permeability, or incomplete resistance databases. Sequence quality and the ability to assign resistance genes to the correct organism become especially important in mixed samples.

Machine learning has been used to predict susceptibility from genomic, spectral, imaging, and clinical data. Performance depends on representative training data, reliable phenotypic labels, and external validation across settings and time. Models may learn lineage or geographic associations instead of causal resistance features, reducing portability when microbial populations change. Rapid measurement and computational prediction therefore require continued comparison with well-characterized reference methods, not merely high accuracy within a development dataset.

Clinical reporting, stewardship, and population surveillance

The clinical value of AST depends on timely, intelligible reporting. A report may include the organism’s identity, MICs or zone-derived categories, the interpretive standard, selected resistance findings, and comments about restricted applicability. Unexpected or consequential results may undergo confirmation before final release. Laboratory information systems must preserve distinctions such as “not tested,” “not interpretable,” and “resistant,” which have different meanings despite sometimes appearing similar in simplified displays.

AST supports antimicrobial stewardship by making the spectrum of active drugs more explicit. Susceptibility information can support changes from empiric to organism-directed treatment and can identify when an initially selected antimicrobial lacks in vitro activity. However, the test does not rank every susceptible drug by overall clinical suitability. Infection site, exposure, toxicity, allergy, drug interactions, source control, and the presence of additional organisms remain separate considerations.

Selective or cascade reporting can display narrower-spectrum agents first and reserve broader-spectrum results for specified circumstances. Such reporting changes what is immediately visible without changing the underlying measurement. Its design requires coordination among microbiology, infectious-disease services, pharmacy, and clinical users. Otherwise, suppression of a result may be mistaken for failure to test, or a susceptible category may be interpreted without its dosing or site-specific qualifications.

At the institutional level, the antibiogram summarizes the proportion of isolates susceptible to selected antimicrobials over a defined period. It informs understanding of local resistance patterns and empiric-treatment policies. Construction commonly uses rules to reduce repeated counting of the same patient’s isolates and requires adequate sample sizes for organism-level estimates. Stratification by hospital unit, age group, specimen type, or acquisition setting may reveal patterns hidden in a hospital-wide aggregate.

Antibiograms also have important limitations. Cultured patients are not necessarily representative of all patients with an infection: cultures may be obtained preferentially after treatment failure or in more severe illness. Small denominators yield unstable estimates, and changes in breakpoints, testing panels, or selective reporting can alter apparent susceptibility rates. Comparisons over time require attention to these methodological changes rather than assuming every percentage shift reflects microbial evolution.

Beyond individual institutions, AST data support public health surveillance, outbreak investigation, and monitoring coordinated through organizations such as the World Health Organization. Linking susceptibility profiles with epidemiological and genomic information can distinguish recurring resistance patterns from transmission of a particular strain. International comparability depends on shared methods, documented interpretive criteria, adequate quality assurance, and metadata describing where and how isolates were obtained.