How Antimicrobial Testing Is Done in Labs: A Simple Guide 2026

Antimicrobial testing exposes a cultured bacterium or fungus to an agent and measures how the organism responds, so a lab can say whether the agent works against it. The whole process runs from a prepared sample to a reported number, and understanding each step explains why two studies can disagree about the same plant extract.

For anyone following traditional remedies through the research literature, this matters. When a paper claims a bark extract shows antimicrobial activity, that claim starts with a question about method. The extract sits on an agar plate or in a broth tube, and what the investigator measures decides what the result means.

The workflow in five steps:

1. Isolate a pure culture from the sample, then identify it.
2. Standardise the inoculum to a defined turbidity or cell density.
3. Expose the organism to graded amounts of the agent on agar or in broth.
4. Measure the response: a clear zone in millimetres, or the lowest concentration that stops visible growth.
5. Interpret the number against breakpoints set by CLSI or EUCAST, and report susceptible, intermediate or resistant.

That is the whole idea. Everything else is detail, and the detail is where the mistakes happen.

What Is Antimicrobial Testing?

Antimicrobial susceptibility testing, often shortened to AST and also called antibiotic sensitivity testing, is the laboratory process of growing a microbe from a sample, exposing it to specific antimicrobial agents, and measuring the response in order to report which agents will inhibit or kill it.

The word “antimicrobial” is broader than most people assume. Antibiotics act on bacteria. Antimicrobials also include antifungal agents, antiviral drugs and antiparasitic medicines, and each category needs its own assay. A plant extract that clears a bacterial lawn tells you nothing about whether it touches a Candida isolate, and the growth conditions differ too: fungi grow more slowly and are read at a different time point than bacteria.

It is worth separating testing from surveillance. Susceptibility testing asks a question about one isolate from one patient: does this drug inhibit this organism? Resistance surveillance asks a question about a population: what proportion of isolates in a hospital or a country are resistant right now? The second question is built out of the first, repeated across thousands of isolates, and reported as an antibiogram.

What a laboratory actually measures comes down to growth. Susceptible organisms grow; inhibited organisms do not. So every method in microbiology is a way of arranging a growing culture, adding something that might stop it, and letting the answer appear as either a gap on a plate or a change in how a tube looks.

Why Does the Choice of Test Method Matter?

Each method answers a different question, and the data point collected is not the same across methods. A zone of inhibition, a minimum inhibitory concentration and a detected resistance gene are three different kinds of evidence.

Diffusion-based methods answer a comparative question: of the agents on this plate, which holds the organism back furthest from where it was applied? That makes disc diffusion fast and cheap, and it is why routine clinical testing relies on it. What it does not give you is a concentration, and it is heavily influenced by how well a compound moves through agar.

Dilution-based methods answer a quantitative question: what is the lowest concentration of this agent that prevents visible growth? That number, the minimum inhibitory concentration, is directly comparable between studies and feeds into dosing decisions. The cost is more material and more bench time.

Molecular methods answer a different question again: is a particular resistance gene present in the DNA? A positive result confirms a mechanism but says nothing about whether the gene is expressed, and a negative result does not exclude an unknown mechanism. A gene without a phenotypic change is a real finding, not a contradiction.

Matching method to question is the whole discipline. A screening study asking whether any activity exists at all can use disc diffusion. A study ranking five extracts against each other needs MIC values, because zone sizes depend on molecular weight and solubility and will mislead the ranking. A study about a specific resistance mechanism needs molecular confirmation alongside the phenotype.

How Is a Test Sample Prepared?

Sample preparation decides how much of the eventual result you can trust, and it is where traditional-plant work most often goes wrong. A poorly prepared extract gives a clean-looking plate and a meaningless number.

Identifying the material. Plant material should be recorded with a voucher specimen deposited at a herbarium, plus collection location, date, plant part and growth habit. The same common name can refer to several unrelated species, and a study that cannot name its plant cannot be repeated by anyone else.

Extraction and standardisation. Dried material is extracted with a solvent chosen for the chemistry, commonly water, ethanol or a water-ethanol mixture, and the extract is concentrated and stored under conditions that do not degrade the compounds of interest. A crude extract is a mixture with variable composition, so its concentration is expressed by mass per volume, and the same nominal concentration can hold very different amounts of active compound between batches. Where a compound is known and available, working with the purified compound removes that variable entirely.

Coded samples. Where the agent being tested is a commercial product or a traditional preparation, samples are usually coded at the point of preparation so the person reading the plates does not know which is which. An unblinded reader measuring zones is a well-documented source of measurement bias.

Solvent compatibility. Most crude extracts are not water-soluble, so they dissolve in dimethyl sulfoxide, ethanol or a mixture. The solvent has to be checked against the growth system, because a solvent that suppresses growth or alters the medium produces a zone that has nothing to do with the extract. Every plate carrying a solvent also carries a solvent-only control at the same volume.

Safe handling. Culturing clinical isolates happens in a biosafety level 2 laboratory, where containment protects the worker and the environment. Teaching laboratories and plant-extract screening work sits somewhere below that, and the honest position is that a student teaching lab is not a reference laboratory. Its results are fine for method training and not fine for clinical reporting.

How Antimicrobial Testing Is Done in Labs: From Sample to Result

Here is the end-to-end path, from study design to interpretation. The exact numbers are published convention rather than something to reproduce at a bench, but the order of the steps is what matters.

1. Define the question and design the study

Before anything is weighed, the study states what it is trying to establish: screening for any activity, ranking against a reference agent, testing for synergy, or assessing one isolate. Each of those needs a different design, and the design determines the method, the controls and the replication.

2. Select and culture the organisms

Organisms come from reference collections such as the American Type Culture Collection, from clinical isolates, or from the target species of interest. Each is grown to purity, because a mixed culture on a plate produces ambiguous zones. Common target organisms include Staphylococcus aureus, Escherichia coli and Pseudomonas aeruginosa, with Streptococcus pneumoniae and Streptococcus pyogenes appearing often in work on respiratory and throat preparations.

3. Standardise the inoculum

For disc diffusion the culture is adjusted to a defined turbidity against a 0.5 McFarland standard, which is the reference for a bacterial suspension used in routine AST. A denser inoculum suppresses zones; a lighter one inflates them, and this single step causes more inter-laboratory variation than any other.

4. Set up the assay with controls

Every plate carries controls. A growth control confirms the organism grew. A sterility control confirms the medium and the diluent carried no contamination. A solvent control shows what the extraction solvent alone does. A positive control, usually a standard antibiotic at a known concentration, confirms the system can detect a known effect.

5. Incubate

Routine bacterial disc diffusion plates are read after roughly 18 to 24 hours of incubation at about 37 degrees Celsius, with the plate inverted to reduce condensation on the agar surface. Fungal work runs longer and cooler. An assay that does not grow within a reasonable window usually points to an inoculum problem rather than an active agent.

6. Measure and record

Zones are measured in millimetres with a ruler or caliper, normally across the full diameter including the disc. Dilution assays are read as the last well without visible growth, with turbidity and subculture used to separate inhibition from killing.

7. Interpret against breakpoints

The measurement only becomes a result once it is compared with a threshold. CLSI, the Clinical and Laboratory Standards Institute, and EUCAST, the European Committee on Antimicrobial Susceptibility Testing, both publish breakpoints, and each reports categories differently. Comparing a zone from one system against a table from the other produces a wrong answer.

8. Report with context

A useful report states the strains, the identification method, the sample batch, the method, the media, the control results, the number of replicates, the endpoints, the units and the variability. The reporting section below returns to this in full.

How Antimicrobial Testing Is Done in Labs for Plant Extracts

Extract testing follows the same logic, but the delivery step changes and one well-known trap appears that clinical testing avoids.

Extract-loaded discs and agar well diffusion are the common formats. In disc loading, sterile discs are impregnated with a fixed volume of extract at graded concentrations and placed on an inoculated plate, following the convention in the published literature of loading discs with 10 microlitres of test substance dissolved in solvent. Agar well diffusion punches wells in the agar and loads extract into them, which suits extracts that do not diffuse well from a disc. Agar well diffusion is often preferred when diffusion is poor, because a zone from a well tells you the compound moved at all, and absence of a zone may simply mean poor diffusion.

Replicates are not optional. Published zone results for botanical extracts are typically reported as a mean with a standard deviation, such as 34.6 millimetres plus or minus 0.57, and triplicate plates are the usual minimum. The variability is reported precisely because crude extracts vary in composition and in diffusion rate, and a single plate tells you almost nothing.

Broth dilution MIC and MBC are run alongside disc diffusion for exactly this reason. Researchers across the plant-extract literature repeatedly add MIC determination because disc diffusion is time-efficient and cost-effective but insufficient for ranking extracts. Minimum bactericidal concentration is found by subculturing the clear wells from the MIC assay onto antibiotic-free agar and recording the lowest concentration with no growth on the plate, which distinguishes an agent that stops growth from one that kills.

Combined-effect assays are a live research area. Checkerboard and combination designs test an extract alongside a standard antibiotic, and recent work has reported synergy in that format. Streptococcus pyogenes is reported as resistant to many plant extracts across studies, which makes it a useful in-house negative control when screening botanicals.

Here is the trap. Zone size in disc diffusion depends on diffusion through agar, and small molecules diffuse faster than large ones. A standard antibiotic in a disc gives a large, sharply defined zone; a crude polyphenolic extract gives a smaller, ragged zone partly because the active molecules are simply diffusing more slowly. Comparing those two numbers as if they were the same quantity is the most common error in the extract literature, and it is why a promising zone result does not survive without an MIC value beside it.

Which Laboratory Methods Are Commonly Used?

There are six method families you will meet in almost any microbiology course or paper.

• Kirby-Bauer disc diffusion — qualitative screening on Mueller-Hinton agar, read as a zone in millimetres.
• Broth microdilution — quantitative, gives an MIC in micrograms per millilitre.
• Agar dilution — quantitative, useful for agents that do not dissolve reliably in broth.
• Gradient diffusion (Etest strips) — quantitative MIC from a strip with a concentration gradient, without a full panel.
• Automated susceptibility analysers — instruments such as VITEK 2, BD Phoenix and Microscan read panels and report MICs with expert-system interpretation.
• Molecular methods — PCR, DNA microarrays and MALDI-TOF identify resistance mechanisms or organisms rather than measuring a response.

Method	What it measures	Result type	Typical turnaround	Where it is used
Disc diffusion (Kirby-Bauer)	Zone of inhibition around an agent	Qualitative, categorical	18 to 24 hours	Routine clinical laboratories, screening studies
Broth microdilution	Lowest concentration with no visible growth	Quantitative MIC	18 to 24 hours	Reference testing, research ranking, extract work
Agar dilution	Lowest concentration with no growth on agar	Quantitative MIC	18 to 72 hours	Agents poorly soluble in broth, some reference work
Gradient diffusion (Etest)	Where growth stops along a concentration gradient	Quantitative MIC	18 to 24 hours	Single agents, confirmatory and difficult agents
Automated analyser	Growth in wells read optically	Quantitative MIC and category	8 to 24 hours	High-throughput clinical testing
PCR or other molecular assay	Presence of a resistance gene	Genotypic, presence or absence	1 to 8 hours	Outbreak screening, confirmation, rapid methods

Two columns in that table explain the practical split. Turnaround is what makes a method usable for a patient waiting, and result type is what makes it usable for a comparison between two extracts.

Choosing between them is less arbitrary than it looks. Disc diffusion wins when the question is which of several agents is worth following up, because one plate answers it in under a day for a handful of antibiotic discs. An automated analyser wins when volume is the constraint: it runs panels overnight and returns MICs in a shift. A gradient strip fills the gap for a single difficult agent where a full panel is not justified, and it reports an MIC rather than a category, which is useful when a result looks unexpectedly resistant.

Two things are worth separating from the choice. Agar dilution exists mainly because some agents will not dissolve cleanly in broth, not because it is a better assay. And molecular methods sit outside the ranking entirely: they answer a mechanism question and produce evidence of a different kind, which is why a laboratory rarely offers them as a replacement for phenotypic testing.

A brief history explains how the current set developed. Fleming worked in the 1920s on a diffusion idea, observing that mould could inhibit surrounding bacteria. Paper discs replaced the culture-dish approach, and the World Health Organization confirmed disc diffusion as a reference method in 1966. Gradient strips followed in 1980, automated analysers and MALDI-TOF expanded through the 2000s, PCR-based resistance detection was published in 2001, and rapid phenotypic systems in microfluidic formats are now in clinical development, including single-cell approaches aimed at urinary tract isolates. Those documents are revised regularly, and the standards themselves are not open access.

How Are MIC, MBC, and Zone-of-Inhibition Results Read?

MIC is the minimum inhibitory concentration: the lowest concentration of an agent that prevents visible growth under defined conditions. MBC is the minimum bactericidal concentration: the lowest concentration that kills the organism rather than only pausing growth. A wide gap between them usually means the agent is bacteriostatic under those conditions.

These are different units of information. A MIC of 32 micrograms per millilitre and a zone of 18 millimetres cannot be compared, because one is a concentration and the other is a distance shaped by diffusion, disc mass and agar composition.

To measure and record a zone of inhibition:

1. Measure the full diameter in millimetres, passing the ruler or caliper across the clear zone through the centre of the disc, not the radius and not the edge of the disc.
2. Take three measurements at roughly 120 degree intervals if the zone is uneven, or repeat the measurement across a replicate plate.
3. Measure the control zone first. If a known standard antibiotic gives no zone on a plate where the sample gave a small one, the plate or the inoculum is at fault rather than the sample.
4. Compare against the breakpoint table for that organism, that drug and that infection site, and record the category rather than only the number.

Worked example: a clear zone is measured at 22 millimetres including the disc. That figure alone is not a result. It becomes one only when it is looked up in the correct breakpoint table and reported as susceptible, intermediate or resistant, with the strain, method and controls attached.

Two breakpoint details catch out readers. First, CLSI and EUCAST use different scales: EUCAST defines susceptible and resistant around a single cutoff with an intermediate explanatory note, while CLSI historically used three categories, and an intermediate result has a different meaning in each system. Second, breakpoints depend on the site of infection, which is why the penicillin breakpoints for Streptococcus pneumoniae differ between meningitis and infections elsewhere in the body.

What Controls Show That a Test Is Reliable?

A control is a known answer included in the run. Six types cover most of what can go wrong.

• Sterility control — a well or plate with no inoculum, showing the medium and diluent were clean.
• Growth control — the organism on plain media with no agent, showing it grows.
• Solvent control — the extraction solvent alone at the volume used, showing the solvent is not creating or hiding a zone.
• Positive control — a standard agent at a known effective concentration, showing the assay can detect activity that is genuinely there.
• Negative control — an organism reported to be resistant, showing the assay does not produce zones for free.
• Reference control — reference strains such as American Type Culture Collection isolates tested in the same run to confirm the system behaves as expected.

A positive control that fails voids the run, and that is the habit most worth copying from clinical labs. The other two are quality assurance rather than interpretation: replicate plates to estimate variability, documented media batches, calibrated pipettes, and operators working from a single written protocol.

Results from one operator should be reproducible by another. Where they are not, the usual culprits are inoculum density, disc placement, agar depth, reading time and reading light. Those five explain most of the disagreement reported between laboratories.

How Should Results Be Recorded and Reported?

A result without its context cannot be interpreted later, and this is the part of the literature most often done thinly.

Identity. Strain designations and their source, the identification method used, and for a plant, the voucher specimen reference with collection details.

Sample. Batch identity, extraction and concentration basis, storage, solvent and solvent percentage, and the code under which it was read.

Method. Method name and standard, media and their lot numbers, inoculum standard, disc or extract loading amount, concentration range tested, incubation conditions and reading time.

Controls. Which controls ran, their results, and whether reference strains fell inside acceptable ranges.

Data. Replicates, the mean with standard deviation, units in millimetres or micrograms per millilitre, the endpoint definition for MIC and MBC, and the breakpoint table used with its version.

Analysis and limits. The statistical method, the variability, and an honest statement of what the design cannot show. Version numbers matter because CLSI and EUCAST documents are revised, and a result read against a superseded table may not match the current category.

This matters beyond the bench. Clinical laboratories pool results into antibiograms, which hospitals use to guide empiric prescribing before any individual culture returns. A result that arrives without its method, its controls and its breakpoint version cannot safely feed that process.

What Can Antimicrobial Testing Not Prove?

An inhibition zone or an MIC proves that an agent, at a defined concentration, under defined laboratory conditions, stopped a defined organism from growing. Every stronger claim needs different evidence.

It does not prove safety for human use. A plate has no concept of a dose, a route, a liver or a pregnancy.

It does not prove clinical effectiveness. In a patient, the drug must reach the site of infection, achieve an effective concentration there, and stay there long enough. A biofilm, an abscess or an intracellular organism all change the answer, and a laboratory plate shows none of that.

It does not prove broad-spectrum activity. Six strains is a screen, not a spectrum. Wider claims need a representative panel.

It does not prove bioavailability, or that a compound reaches blood concentrations anywhere near the in vitro MIC.

Between an inhibition result and a treatment claim sit further studies in order of demand: cytotoxicity testing in human cells, acute and repeat-dose toxicity work, pharmacokinetics, and only then preclinical and clinical studies.

Two failure modes are common in how results are written up, and both are avoidable. The first is a sample that behaves well in broth and poorly on a plate, or the reverse, when a compound binds a carrier protein in one system or precipitates out of the other. The second is an extract that looks selective against one organism and is equally mild against everything on the panel, meaning the clear zone reflects a mild general effect rather than anything worth pursuing.

Antimicrobial stewardship sits alongside this the whole way, since reducing unnecessary use is one of the levers that slows resistance. A traditional preparation with a promising MIC still belongs in a lab, not in a self-treatment plan, and anyone weighing one for personal use should raise it with a doctor or pharmacist first.

Frequently Asked Questions

Conclusion: Start with the Right Method

How antimicrobial testing is done in labs comes down to a sequence most people can hold in their head: isolate a pure culture, standardise the inoculum, expose it to graded amounts of the agent, measure the response, and interpret that measurement against published breakpoints. The workflow is stable, and the method choice is where a study is won or lost.

So start with the question, not the equipment. Decide whether you need a screen, a ranking or a mechanism, pick the method that produces that kind of data, and build in the controls before the first plate is poured. For medicinal plant work, run MIC alongside disc diffusion, report triplicate zones with variability, and keep the claim no larger than the data.