# Antimicrobial Resistance Surveillance in Food Animals: Sampling Strategies and Data Interpretation


## Key Takeaways

- Surveillance objectives (prevalence, trends, early warning) dictate sampling intensity, bacterial targets (indicator organisms like *E. coli* vs. pathogens like *Salmonella*), and specimen types (fecal, cecal contents, carcass swabs).
- Sample size calculations are driven by expected prevalence, desired confidence, acceptable precision, and the design effect due to clustering within production units, with specific strategies needed for detecting rare resistance phenotypes.
- Specimen integrity is paramount; cold chain maintenance (4°C for feces) and appropriate transport media (Cary-Blair for enteric pathogens) are critical, while pooling reduces cost but sacrifices individual-level data and sensitivity for rare clones.
- Phenotypic characterization requires standardized susceptibility testing (broth microdilution or disk diffusion) interpreted against clinical breakpoints or epidemiological cut-off values (ECOFFs) to distinguish therapeutic outcomes from early resistance detection.
- Data interpretation necessitates consistent methodology across sampling rounds, analysis of multidrug resistance (resistance to ≥3 classes), and acknowledgment of limitations such as sampling bias from convenience sampling or overrepresentation of clinically affected animals.
- Escalation criteria include detecting resistance to critically important antimicrobials (e.g., carbapenems) or specific zoonotic phenotypes (e.g., ESBL-producing Enterobacterales), triggering urgent confirmation and notification to veterinary authorities.

---

Antimicrobial resistance (AMR) surveillance in food animal populations serves a dual purpose: it quantifies resistance trends within production species and generates data that inform public health risk assessment. This article provides a procedural reference for veterinary researchers designing or evaluating AMR surveillance programs in cattle, swine, poultry, and aquaculture. It covers sampling frame construction, specimen collection protocols, bacterial isolation and susceptibility testing strategies, and the statistical interpretation of resistance data, with attention to the limitations imposed by production system diversity and resource constraints.

The reader is assumed to be familiar with clinical microbiology and basic epidemiological principles. The practical questions addressed are: how many samples are needed to detect a specified resistance prevalence, which specimens and bacterial targets best answer a given surveillance objective, how should laboratory data be aggregated and reported, and what interpretive caveats apply when comparing data across regions or time periods. The article does not provide jurisdiction-specific regulatory requirements, international standards are referenced where they exist, and regional differences are flagged where they materially affect study design.

AMR in food animals is not an isolated phenomenon. Resistance genes circulate among human pathogens, livestock-associated bacteria, commensal flora, and environmental reservoirs, and homologous resistance determinants have been recovered from all of these compartments. The [One Health framework published by the World Health Organization](https://www.who.int/health-topics/one-health) explicitly links human, animal, and environmental health for zoonotic disease and AMR control, and the [Centers for Disease Control and Prevention's One Health resources](https://www.cdc.gov/one-health/index.html) provide operational guidance for cross-sector surveillance collaboration. Surveillance design must therefore consider also the target population but also the pathways by which resistance may move between populations, including foodborne transmission, occupational exposure, and environmental contamination. The burden of foodborne disease remains substantial and is influenced by dynamic factors along the production chain, which argues for sustained monitoring instead of one-time prevalence surveys. The global scale of antimicrobial use in food animal production, particularly in rapidly expanding aquaculture sectors, further underscores the need for standardized surveillance methods that permit international comparison.

## At a Glance

| Parameter | Decision or Fact |
|---|---|
| Primary surveillance objective | Determine prevalence, detect trends, or identify emerging resistance, each requires different sampling intensity |
| Target population | Herd, flock, or regional population, define by species, production stage, and antimicrobial exposure history |
| Specimen type | Fecal samples, cecal contents, carcass swabs, or environmental samples, choice depends on bacterial target and question |
| Bacterial targets | Indicator organizms (commensal *E. coli*, enterococci) versus clinical pathogens (*Salmonella*, *Campylobacter*) |
| Sample size driver | Expected prevalence, desired confidence level, acceptable precision, and design effect from clustering |
| Susceptibility testing method | Broth microdilution or disk diffusion, interpret using internationally recognized clinical breakpoints where available |
| Data reporting unit | Percent resistant, minimum inhibitory concentration distributions, or quantitative resistance gene abundance |
| Key interpretive caveat | Prevalence estimates from convenience sampling cannot be generalized to unsampled populations |

## Conceptual Foundations of AMR Surveillance

### Defining the Surveillance Objective

The design of any AMR surveillance program begins with an explicit statement of the question. Prevalence surveys estimate the proportion of isolates resistant to a given antimicrobial at a single time point. Trend surveillance repeats standardized sampling at intervals to detect changes over time. Early warning systems prioritize the detection of rare or emerging resistance phenotypes, such as carbapenemase production or colistin resistance, and require different sampling strategies than prevalence estimation. The objective determines the bacterial target, the specimen type, the sampling frame, and the statistical power required.

### The Role of Indicator Organizms

Surveillance programs commonly distinguish between indicator organizms and clinical pathogens. Commensal bacteria such as *Escherichia coli* and enterococci are present in the gastrointestinal tract of nearly all food animals, are easily cultured, and serve as a reservoir from which resistance genes can transfer to pathogenic bacteria. Indicator organizms provide a sensitive measure of antimicrobial selection pressure within a population because they are exposed to the same antimicrobials as pathogens but are not subject to the same clinical sampling biases. Pathogen surveillance, by contrast, targets organizms of direct public health or animal health significance, such as *Salmonella* and *Campylobacter*. Pathogen isolation is often complicated by intermittent shedding and low prevalence, which increases the required sample size and may necessitate enrichment procedures that affect quantitative interpretation.

### The Ecology of Resistance in Production Systems

Food animals are exposed to antimicrobials through therapeutic use, metaphylaxis, and, in some regions, growth promotion. The resulting selection pressure operates on the entire gastrointestinal microbial community, also on pathogenic species. Resistance genes can persist in commensal populations long after antimicrobial use has ceased, a phenomenon that complicates the interpretation of temporal trends. The [review of AMR in humans, livestock, and the wider environment by Woolhouse and colleagues](https://pubmed.ncbi.nlm.nih.gov/25918441/) emphasizes that farm animals are exposed to large quantities of antibiotics and act as a reservoir of resistance genes, with two-way traffic of resistant bacteria between farms and clinical settings now quantifiable through whole genome sequencing. Surveillance programs must therefore account for the possibility that resistance detected in food animals reflects historical antimicrobial use, environmental contamination, or introduction through animal movement, instead of current on-farm practices.

### Aquaculture as a Distinct Surveillance Context

Aquaculture presents unique surveillance challenges. Production occurs in open water systems where environmental exchange is continuous, and the diversity of farmed species, from shrimp to finfish, means that antimicrobial use practices and bacterial targets vary substantially. Global antimicrobial consumption in aquaculture was estimated at over 10,000 tons in 2017, with the Asia-Pacific region accounting for the overwhelming majority of use. Surveillance in aquatic systems requires sampling of water, sediment, and animal tissues, and the bacterial targets differ from terrestrial production. The [global trends in antimicrobial use in aquaculture reported by Schar and colleagues](https://pubmed.ncbi.nlm.nih.gov/33318576/) project continued growth in consumption, indicating that surveillance capacity in this sector will need to expand correspondingly.

## Sampling Strategy Design

### Defining the Sampling Frame

The sampling frame is the complete list of units from which samples will be drawn. In food animal surveillance, the unit may be an individual animal, a pen or house, a herd or flock, or a geographic region. The frame must be defined with sufficient precision that every unit has a known probability of selection. For herd-level surveillance, a list of registered production sites may serve as the frame, but such lists are often incomplete in regions with many smallholder producers. The [World Organization for Animal Health terrestrial animal health standards](https://www.woah.org/en/what-we-do/standards/codes-and-manuals/terrestrial-code-online-access/) provide guidance on surveillance system design and data quality that applies to AMR monitoring as a component of animal health surveillance.

### Sample Size Determination

Sample size calculations for prevalence estimation require four inputs: the expected prevalence, the desired confidence level, the acceptable margin of error, and the design effect. For a simple random sample, the required number of isolates is approximately *n* = (*Z*² × *p*(1,  *p*)) / *d*², where *Z* is the standard normal deviate for the confidence level, *p* is the expected prevalence, and *d* is the half-width of the desired confidence interval. When animals are sampled in clusters, such as multiple animals from the same pen, the design effect inflates the required sample size. The design effect is estimated as 1 + (*m*,  1)ρ, where *m* is the average cluster size and ρ is the intracluster correlation coefficient, which measures the degree of similarity in resistance status among animals within the same cluster. For most enteric bacteria, ρ values between 0.1 and 0.3 are reasonable planning assumptions, but published estimates for specific production systems should be used when available.

### Detection of Rare Resistance Phenotypes

When the surveillance objective is to detect the presence of a rare resistance phenotype, such as resistance to a critically important antimicrobial at a prevalence below 1%, the sample size required for prevalence estimation becomes impractically large. Detection sampling uses a different logic: the goal is to achieve a specified probability of detecting at least one positive isolate if the phenotype is present at or above a threshold prevalence. For a prevalence of 1% and a desired detection probability of 95%, approximately 300 isolates must be tested. This calculation assumes perfect sensitivity of the laboratory method, which is rarely achieved, so additional isolates should be included to account for diagnostic sensitivity below 100%.

### Stratification and Targeted Sampling

Stratification improves precision when resistance prevalence is expected to vary across subgroups. Common strata include production type (broiler versus layer, dairy versus beef), age class, antimicrobial exposure history, and geographic region. Stratified sampling requires that the proportion of samples allocated to each stratum reflect either the population distribution or the analytical priority. Oversampling of high-risk strata, such as animals recently treated with antimicrobials, can increase the power to detect resistance but biases population-level prevalence estimates unless weighting is applied during analysis. Targeted sampling is appropriate for early warning systems but should be clearly distinguished from prevalence surveillance in the reporting of results.

## Specimen Collection and Handling

The quality of antimicrobial resistance surveillance data depends on specimen integrity from farm to laboratory. For fecal samples, collect 5 to 10 g of freshly voided feces or rectal contents into sterile containers with leak-proof seals. Avoid environmental contamination with bedding or soil, as commensal environmental bacteria can obscure the target population. For carcass swabs at abattoir level, use sponge samplers moistened with buffered peptone water over a defined surface area, typically 100 to 400 cm² depending on the tissue site. Transport media are required for swabs that will not reach the laboratory within 2 hours. Cary-Blair medium is appropriate for enteric pathogens, while Amies medium with charcoal suits most aerobic organizms.

Cold chain maintenance is critical. Fecal samples should be refrigerated at 4 °C and processed within 24 hours. Freezing at -20 °C is acceptable for retrospective analysis but reduces recovery of some Gram-negative organizms and may bias population estimates. For aquaculture samples, gill, kidney, and intestinal tissue should be collected aseptically from moribund fish instead of dead fish, since post-mortem invaders rapidly colonise tissues. Water samples require neutralising agents for residual disinfectants, typically sodium thiosulphate at 0.01% final concentration.

Pooling strategies reduce cost but sacrifice individual-level data. Pool five fecal samples from the same pen or house when the objective is herd-level prevalence estimation. Pooling is inappropriate when the objective includes detecting rare resistance phenotypes, because dilution reduces the probability of recovering low-abundance resistant clones. [The global problem of antimicrobial resistance in developing countries](https://pubmed.ncbi.nlm.nih.gov/16048717/) illustrates that resource-limited settings often require pooling to make surveillance feasible, but the trade-off in sensitivity must be documented in the surveillance protocol.

## Isolation and Phenotypic Characterization

Selective media choices determine which organizms enter the surveillance pipeline. For Escherichia coli, MacConkey agar with cefotaxime at 1 mg/L selects for extended-spectrum beta-lactamase producers, while plain MacConkey agar recovers the general commensal population. Salmonella species require pre-enrichment in buffered peptone water for 18 to 24 hours, followed by selective enrichment in Rappaport-Vassiliadis broth and plating on xylose-lysine-deoxycholate agar or chromogenic Salmonella media. Campylobacter species demand microaerophilic conditions at 42 °C on selective media such as modified charcoal-cefoperazone-deoxycholate agar.

The choice between chromogenic and conventional media affects throughput and cost. Chromogenic media allow presumptive identification directly from primary plates, reducing subculture requirements. However, they are substantially more expensive per plate and may not be justified for large-scale surveillance programs with high sample numbers. Conventional media with biochemical confirmation remain the standard for reference laboratories and for programs that require definitive species identification before antimicrobial susceptibility testing.

Antimicrobial susceptibility testing should follow a standardized method, either broth microdilution or disk diffusion, with interpretation against clinical breakpoints where they exist. For organizms without clinical breakpoints, epidemiological cut-off values (ECOFFs) distinguish wild-type populations from those with acquired resistance mechanisms. The distinction matters: clinical breakpoints predict therapeutic outcome, while ECOFFs detect emerging resistance earlier in the population. Surveillance programs should report both values when available, because a resistant subpopulation may exist below the clinical breakpoint threshold.

## Data Management and Quality Assurance

Every isolate requires a minimum dataset: unique identifier, collection date, farm or facility identifier, production stage, sample type, anatomical site, species, and antimicrobial susceptibility results. Geographic coordinates enable spatial analysis and cluster detection. Without standardized metadata, resistance data cannot be interpreted across regions or time periods.

Internal quality control should include reference strains with known susceptibility profiles on every batch of susceptibility tests. E. coli ATCC 25922 and Staphylococcus aureus ATCC 29213 are standard controls for Gram-negative and Gram-positive organizms respectively. Participation in an external quality assurance scheme, such as those operated by national reference laboratories, identifies systematic errors in methodology or interpretation.

Laboratory information management systems should enforce data validation rules at entry. Susceptibility results outside expected ranges, missing metadata fields, and duplicate isolate identifiers should trigger automatic review. Data should be exported in a standard format, such as CSV or XML, with a defined data dictionary to facilitate sharing between institutions. [The WHO One Health framework](https://www.who.int/health-topics/one-health) explicitly requires interoperable data systems across human, animal, and environmental sectors, and standardized laboratory outputs are the foundation of that interoperability.

## Interpretation of Surveillance Data

Resistance prevalence is the proportion of isolates resistant to a given antimicrobial, expressed as a percentage. This metric is intuitive but sensitive to sampling bias. If clinical isolates are overrepresented, prevalence will overestimate resistance in the general population. Commensal indicator organizms sampled from healthy animals provide a less biased estimate of the resistance reservoir.

Temporal trends require consistent methodology across sampling rounds. Changes in media, breakpoints, or sampling sites invalidate comparisons. When breakpoints change, historical data should be re-interpreted using the new criteria where raw data permit. The [CDC One Health and zoonotic disease resources](https://www.cdc.gov/one-health/index.html) emphasize that trend analysis is only meaningful when surveillance systems remain stable over time.

Resistance patterns should be analyzed at the level of multidrug resistance, defined as resistance to three or more antimicrobial classes. Multidrug-resistant phenotypes are more clinically significant than single-drug resistance and are more likely to be associated with mobile genetic elements that facilitate horizontal transfer. The prevalence of extended-spectrum beta-lactamase-producing E. coli and methicillin-resistant S. aureus in food animals should be tracked separately from general resistance metrics, as these phenotypes have direct implications for human therapeutic options.

## Translating Data into Public Health Action

Surveillance data acquire value only when they inform decisions. The decision pathway begins with identifying resistance patterns that exceed predefined thresholds. A threshold of 20% resistance to a critically important antimicrobial in a food animal reservoir may trigger enhanced investigation, while 50% resistance may justify intervention. Thresholds should be set by the surveillance authority in consultation with public health and veterinary stakeholders, and should account for regional context and the antimicrobial class in question.

Interventions range from targeted farm-level biosecurity measures to regulatory restrictions on antimicrobial use. When resistance is concentrated in specific production stages, such as the grower phase in poultry or the finisher phase in swine, interventions can be directed accordingly. When resistance is widespread across multiple production systems, policy-level action is required.

Data dissemination must reach different audiences in different formats. Veterinary practitioners need species-specific resistance profiles to guide empirical therapy. Public health authorities need trend data to assess zoonotic transmission risk. Producers need actionable recommendations for antimicrobial stewardship. [The MSD Veterinary Manual](https://www.msdvetmanual.com/) provides species-specific clinical context that helps practitioners interpret surveillance data in the context of individual patient care, while [AVMA practice resources](https://www.avma.org/resources-tools) offer guidance on translating population-level data into practice-level decisions.

The limitations of surveillance data must be acknowledged in all reports. Sampling frames that exclude smallholder farms, backyard flocks, or artisanal aquaculture operations will miss resistance reservoirs that differ systematically from commercial production. [Antimicrobial resistance in humans, livestock and the wider environment](https://pubmed.ncbi.nlm.nih.gov/25918441/) notes that farm animals are exposed to enormous quantities of antibiotics and act as a reservoir of resistance genes, but the two-way traffic of resistance between farm and clinic is only beginning to be quantified. Surveillance reports should therefore state their coverage explicitly and avoid extrapolating beyond the sampled population.

## Reporting Frameworks and Thresholds

| Metric | Definition | Interpretation | Action Threshold |
|--------|------------|----------------|------------------|
| Resistance prevalence | Proportion of isolates resistant to a specific antimicrobial | Baseline resistance burden in the sampled population | 10% for critically important antimicrobials |
| Multidrug resistance prevalence | Proportion of isolates resistant to 3 or more antimicrobial classes | Indicates co-selection and mobile genetic element involvement | 5% in commensal indicators |
| Extended-spectrum beta-lactamase prevalence | Proportion of E. coli or Salmonella with ESBL phenotype | Direct zoonotic transmission risk | 2% in food animal isolates |
| Temporal trend slope | Change in resistance prevalence per year | Detects emerging resistance before it becomes established | Positive slope over 2 consecutive years |
| Spatial cluster | Geographic concentration of resistant isolates | Identifies point sources or regional practices driving resistance | Any statistically significant cluster |

Thresholds in this table are illustrative starting points. Each surveillance program should calibrate its own thresholds based on baseline data, antimicrobial importance, and regional public health priorities. The [WOAH terrestrial animal health code](https://www.woah.org/en/what-we-do/standards/codes-and-manuals/terrestrial-code-online-access/) provides international standards for surveillance design and reporting that should be consulted when establishing a new program.

Reporting frequency depends on the decision cycle. Annual reports suffice for trend monitoring. Quarterly reports are appropriate during intervention periods or when emerging resistance is detected. Outbreak alerts require immediate communication through established rapid reporting channels. Each report should include the sampling frame, methodology, quality assurance results, and a clear statement of limitations.

## Recognized Complications and Failure Modes

Surveillance systems fail in characteriztic patterns. The most common is silent selection bias introduced at the specimen collection stage. When samples are drawn exclusively from clinically affected animals, the resulting data describe resistance in diseased populations, not the baseline carriage in healthy production groups. This distinction matters because clinical isolates overrepresent resistant clones selected by recent therapy and underrepresent the reservoir of resistance genes circulating in the absence of disease. Detection of this failure requires comparing clinical isolate data with periodic screening of healthy animals from the same cohort, a practice many programs omit for cost reasons.

A second failure mode is temporal misalignment between antimicrobial use data and resistance data. Resistance emerges weeks to months after selection pressure, and the lag varies by bacterial species, drug class, and production stage. Programs that pair a single month of usage data with a single month of resistance data will misattribute causation. The corrective approach is to analyze resistance trends against cumulative exposure over the preceding quarter, at minimum, and to acknowledge that the lag itself is a study parameter instead of a nuisance variable.

Laboratory-driven failure modes include the use of different susceptibility testing methods across participating laboratories without harmonised quality control. Zone diameter interpretations from disk diffusion are not directly interchangeable with minimum inhibitory concentration values from broth microdilution, and inter-laboratory variation can exceed the biological signal being measured. Participation in an external quality assurance scheme, with blinded repeat isolates, is the only reliable early detection mechanism. A program that cannot demonstrate acceptable inter-laboratory reproducibility should not publish pooled resistance percentages.

## Common Errors in Program Design

Less experienced investigators frequently underpower their surveillance for the resistance phenotypes that matter most. A sample size calculated to detect a 10% change in ampicillin resistance in *Escherichia coli* will not detect the emergence of carbapenemase-producing organizms at 0.5% prevalence, yet the latter carries greater public health significance. The solution is to calculate sample size separately for each target phenotype and to accept that rare phenotypes require either substantially larger sampling frames or targeted enrichment strategies, as discussed in earlier sections of this article.

A related error is the conflation of prevalence with incidence. Cross-sectional sampling measures the proportion of samples yielding a resistant isolate at one time point. It does not measure the rate of new acquisitions. For intervention assessment, incidence-based designs with repeated sampling of the same production units are required. Programs that rely solely on annual cross-sectional surveys cannot distinguish a reduction in new infections from a reduction in duration of colonisation.

Misclassification of surveillance purpose is another recurring problem. Data collected for early detection of emerging resistance cannot simultaneously serve as a robust estimate of population prevalence without design compromise. Early detection favours high-risk sampling, such as diagnostic submissions or known hotspot units. Prevalence estimation favours probability-based sampling of the general population. Attempting both with a single sampling frame produces data that are adequate for neither purpose. The program must declare its primary objective and accept the corresponding interpretive limitations.

## Limitations of the Evidence Base

The evidence supporting AMR surveillance design in food animals carries substantial geographic bias. High-income countries with established programs contribute most published data, while regions with the highest antimicrobial consumption intensity, particularly parts of Asia, remain under-represented in accessible literature. Global estimates of antimicrobial use in aquaculture illustrate the scale of this disparity, with the Asia-Pacific region accounting for the overwhelming majority of consumption and China alone contributing more than half of the global total [Global trends in antimicrobial use in aquaculture](https://pubmed.ncbi.nlm.nih.gov/33318576/). Surveillance capacity in these regions is improving but has not kept pace with production growth.

Expert opinion still differs on the choice of indicator organizm. Some programs prioritize *E. coli* as a universal enteric indicator, while others argue that species-specific pathogens such as *Salmonella* or *Campylobacter* provide more actionable data for human health risk assessment. The evidence does not currently resolve this debate, and the choice remains a function of the surveillance objective, the production system, and the downstream users of the data. A pragmatic position is to maintain both an indicator organizm for trend analysis and a pathogen-specific component for risk assessment, accepting the additional laboratory cost.

The relationship between antimicrobial use in food animals and human clinical resistance is contested in its magnitude, though not in its existence. Homologous resistance genes have been identified across human pathogens, livestock flora, and environmental bacteria, and whole genome sequencing continues to quantify two-way traffic between farm and clinic [Antimicrobial resistance in humans, livestock and the wider environment](https://pubmed.ncbi.nlm.nih.gov/25918441/). The unresolved question is the proportional contribution of food animal reservoirs relative to human-to-human transmission and environmental spread. Surveillance programs should therefore frame their outputs as contributions to a broader One Health picture instead of as standalone risk assessments.

## Escalation and Referral Criteria

Certain findings warrant immediate escalation beyond routine reporting. Detection of resistance to critically important antimicrobial classes where food animal use is restricted or prohibited, such as carbapenems or glycopeptides, requires urgent laboratory confirmation, repeat testing, and notification to the relevant veterinary authority. The same applies to resistance phenotypes with documented zoonotic transmission potential in organizms such as methicillin-resistant *Staphylococcus aureus* or extended-spectrum beta-lactamase-producing Enterobacterales.

Regulatory reporting obligations vary by jurisdiction and production species. Veterinarians should consult their national veterinary authority and the [WOAH terrestrial animal health standards](https://www.woah.org/en/what-we-do/standards/codes-and-manuals/terrestrial-code-online-access/) for notifiable resistance findings and trade-related reporting requirements. Where a surveillance program identifies a resistance trend that threatens therapeutic options for a production species, referral to a veterinary clinical pharmacologist or diagnostic laboratory with specialised resistance testing capacity is appropriate.

Laboratory involvement is warranted when phenotypic results are ambiguous, when confirmatory genotypic testing is required, or when an isolate shows an unusual resistance pattern that may indicate a novel mechanism. Reference laboratories with whole genome sequencing capacity should be engaged for such isolates, as the public health significance of a novel resistance determinant cannot be assessed from phenotypic data alone.

| Observation | Likely cause | Discriminating check |
|---|---|---|
| Resistance percentages differ markedly between laboratories | Methodological variation in susceptibility testing | Review external quality assurance results and harmonise protocols |
| Resistance trend rises after a use reduction intervention | Temporal lag between use and resistance | Extend the analysis window to cumulative exposure over the prior quarter |
| Rare resistance phenotype appears in one region only | True emergence or sampling artefact | Repeat sampling with targeted enrichment and confirm at reference laboratory |
| Clinical isolates show higher resistance than healthy animal screening | Selection bias toward diseased populations | Compare sampling frames and consider adding baseline carriage screening |
| Pooled data show no trend despite suspected resistance emergence | Underpowered for rare phenotypes | Recalculate sample size for the specific phenotype of interest |

## Frequently Asked Questions

### How can a surveillance program be designed when laboratory capacity and funding are severely limited?

Prioritize a narrow, well-defined objective over broad coverage. A program that reliably isolates one indicator organizm, such as commensal *Escherichia coli*, from a small, stratified sample will produce more interpretable data than an under-resourced attempt at multi-pathogen surveillance. Use selective media that require minimal specialised equipment, and consider pooling samples from several animals within a single pen to reduce processing costs. Establish a partnership with a regional or national reference laboratory for confirmatory testing and quality assurance. The global burden of resistance is heavily concentrated in settings where surveillance infrastructure is weakest, so practical, incremental approaches are more valuable than deferred ideal designs [Antimicrobial resistance in developing countries. Part I: recent trends and current status](https://pubmed.ncbi.nlm.nih.gov/16048717/).

### What should be done when the target pathogen cannot be isolated from the planned sample matrix?

First verify that the isolation method is appropriate for the matrix and the organizm. Enrichment steps, selective supplements, and incubation conditions may need adjustment. If isolation continues to fail, reassess the sampling frame. The target organizm may be present at very low prevalence, or the chosen matrix may not be the optimal reservoir. Consider switching to a different indicator organizm that is more readily cultured from that matrix, or add an enrichment step to increase sensitivity. Document the failure and the methodological changes made, because negative isolation data are still informative for prevalence estimation. Surveillance programs in livestock remain relatively poor compared with human systems, and reporting methodological obstacles transparently helps the wider community improve future designs [Antimicrobial resistance in humans, livestock and the wider environment](https://pubmed.ncbi.nlm.nih.gov/25918441/).

### How does surveillance design differ for aquaculture compared with terrestrial food animals?

Aquaculture presents distinct sampling challenges because the production unit is water, not an individual animal. Sampling water, sediment, and fish skin or gut contents captures different compartments of the bacterial population. Antimicrobial use in aquaculture is highly variable by species group and region, with the Asia-Pacific region accounting for the overwhelming majority of global consumption [Global trends in antimicrobial use in aquaculture](https://pubmed.ncbi.nlm.nih.gov/33318576/). Pooled water samples are often more practical than individual fish sampling, but they dilute the signal from clinically relevant isolates. Define the sampling frame by pond or cage instead of by animal, and stratify by production stage. Seasonal temperature variation affects both bacterial load and antimicrobial degradation, so time-series sampling should account for these cycles.

### What records must be kept to make surveillance data interpretable?

Each isolate requires a minimum dataset: species, sample type, collection date, farm or production unit identifier, production stage, and antimicrobial use history for that group of animals. Without antimicrobial use data, resistance prevalence cannot be linked to selection pressure. Record laboratory methods, including the antimicrobial panel, breakpoints used, and quality control results for reference strains. Store isolates in a manner that permits later confirmation or whole genome sequencing. The One Health framework explicitly requires data that can be shared across human, animal, and environmental sectors [WHO One Health Initiative](https://www.who.int/health-topics/one-health), so use standardized data fields and controlled vocabularies from the outset. Inconsistent record keeping is a common cause of program failure that no amount of downstream analysis can repair.

### How should results be communicated to producers or farm managers who are not trained in microbiology?

Frame the results around production outcomes and risk, not around laboratory jargon. Explain that resistance in commensal bacteria is an early warning signal, not necessarily a diagnosis of disease. Show trends over time instead of single time point values, and compare the farm to regional benchmarks if those exist. Be explicit about what the data do not show, for example that a resistant commensal isolate does not prove treatment failure in a sick animal. The MSD Veterinary Manual provides species-specific guidance on interpreting diagnostic results in clinical context [MSD Veterinary Manual, Professional Edition](https://www.msdvetmanual.com/). Offer concrete management adjustments, such as revisiting biosecurity protocols or reviewing antimicrobial choice protocols, and schedule a follow-up discussion once the next sampling round is complete.

### When should a surveillance finding trigger escalation to regulatory authorities or a formal outbreak investigation?

Escalation is warranted when a resistance phenotype of public health significance appears where it was previously absent, when prevalence of a critical resistance pattern rises sharply between consecutive sampling rounds, or when a zoonotic pathogen with a clinically important resistance profile is isolated. Examples include extended-spectrum beta-lactamase producing Enterobacterales, methicillin-resistant *Staphylococcus aureus*, or fluoroquinolone resistant *Salmonella* from a food animal. The World Organization for Animal Health provides international standards for notification and trade related disease control [WOAH terrestrial animal health code](https://www.woah.org/en/what-we-do/standards/codes-and-manuals/terrestrial-code-online-access/). Consult national veterinary authorities early, even if the finding is equivocal, because they can advise on confirmatory testing and epidemiological follow-up. Document the isolate, the laboratory results, and the farm context before making the call.

## Related Clinical & Scientific Guides

* [Wildlife Disease Surveillance: Designing and Implementing a One Health Program](/knowledge/veterinary-medicine/veterinary-public-health/wildlife-disease-surveillance-designing-implementing-one-health-program)
* [Biosecurity Risk Assessment for Livestock Operations: A Practical Framework](/knowledge/veterinary-medicine/veterinary-public-health/biosecurity-risk-assessment-livestock-operations-practical-framework)
* [Rabies Post-Exposure Prophylaxis in Veterinary Personnel](/knowledge/veterinary-medicine/veterinary-public-health/rabies-post-exposure-prophylaxis-in-veterinary-personnel)


## References and Further Reading

- [Antimicrobial resistance in developing countries. Part I: recent trends and current status.](https://pubmed.ncbi.nlm.nih.gov/16048717/). 2005.
- [Antimicrobial Resistance: A Growing Serious Threat for Global Public Health.](https://pubmed.ncbi.nlm.nih.gov/37444780/). 2023.
- [Pet animals as reservoirs of antimicrobial-resistant bacteria.](https://pubmed.ncbi.nlm.nih.gov/15254022/). 2004.
- [Global trends in antimicrobial use in aquaculture.](https://pubmed.ncbi.nlm.nih.gov/33318576/). 2020.
- [Food-borne diseases - the challenges of 20 years ago still persist while new ones continue to emerge.](https://pubmed.ncbi.nlm.nih.gov/20153070/). 2010.
- [Antimicrobial resistance in humans, livestock and the wider environment.](https://pubmed.ncbi.nlm.nih.gov/25918441/). 2015.
- [WHO One Health Initiative](https://www.who.int/health-topics/one-health). WHO.
- [CDC One Health and Zoonotic Disease Resources](https://www.cdc.gov/one-health/index.html). CDC.
- [MSD Veterinary Manual, Professional Edition](https://www.msdvetmanual.com/). MSD Veterinary Manual.

## Related Articles

- [Antimicrobial Resistance Surveillance in Wildlife: Methods and Gaps](/knowledge/veterinary-medicine/veterinary-public-health/antimicrobial-resistance-surveillance-in-wildlife-methods-and-gaps)
- [One Health Approach to Antimicrobial Resistance: A Veterinary Perspective](/knowledge/veterinary-medicine/veterinary-public-health/one-health-approach-to-antimicrobial-resistance-a-veterinary-perspective)
- [Antimicrobial Stewardship in Food Animal Practice: Metrics and Benchmarks](/knowledge/veterinary-medicine/veterinary-public-health/antimicrobial-stewardship-in-food-animal-practice-metrics-and-benchmarks)
- [One Health Surveillance Data Integration: Challenges and Solutions](/knowledge/veterinary-medicine/veterinary-public-health/one-health-surveillance-data-integration-challenges-and-solutions)
- [Antimicrobial Resistance in Wildlife: Environmental Reservoir and Public Health Risk](/knowledge/veterinary-medicine/veterinary-public-health/antimicrobial-resistance-wildlife-environmental-reservoir-public-health-risk)

> This article is educational professional reference material for veterinary audiences. It is not a substitute for veterinary diagnosis, individual clinical judgment, current product labeling, or applicable regulatory requirements.