# Foodborne Outbreak Source Attribution: Veterinary Contributions


## Key Takeaways

- Whole genome sequencing (WGS) is the primary tool for source attribution, enabling sub-species level discrimination of pathogen isolates from human, animal, food, and environmental sources to establish phylogenetic relatedness and infer common ancestry.
- Veterinary expertise is critical for interpreting genomic matches within the context of production systems, animal movement, and management practices, distinguishing true transmission events from coincidental genetic similarities.
- Traceback investigations, supported by veterinary records of animal identification, movement, and health status, are essential to follow contaminated food items back to their origin, with veterinarians contributing at the farm level.
- Population-level attribution models rely on representative sampling of animal reservoirs, a veterinary responsibility, to estimate the proportion of human cases originating from specific sources over time.
- Host adaptation signals, such as lineage-specific pseudogene accumulation, can facilitate source attribution by genetically specializing pathogens to particular animal species, though generalist lineages pose a greater challenge.
- Limitations in attribution include the inability of genomic similarity alone to prove transmission, the pervasive issue of sampling bias, and the need for harmonized bioinformatics pipelines to ensure data comparability across laboratories and jurisdictions.

---

Source attribution is the process of estimating which animal reservoir, food commodity, or production pathway gave rise to a human foodborne infection. For the veterinary researcher, this discipline sits at the intersection of population medicine, microbial ecology, and forensic epidemiology. The question that drives the field is deceptively simple: when a cluster of human illness occurs, which animal population was the original source of the pathogen, and through which food chain did it travel? Answering that question requires integrating genomic data from veterinary isolates, production system knowledge, and traceback evidence from the farm to the point of consumption.

This article examines the conceptual foundations and practical methods by which veterinarians contribute to source attribution. It is written for veterinary researchers and public health veterinarians who need a structured account of attribution logic, the genomic tools now used to link animal and human isolates, and the study designs that support causal inference. Clinical management of affected animals and laboratory diagnostic technique are covered elsewhere and are not treated here. The focus is on how veterinary data become attribution evidence.

The veterinary contribution is not limited to providing samples. Veterinarians interpret whether a pathogen recovered from an animal is incidental, colonizing, or causally linked to human illness. They understand production system variation, seasonal management practices, and the movement of animals between premises. This contextual knowledge determines whether a genomic match between a human isolate and an animal isolate represents a true source relationship or a coincidental similarity.

## At a Glance

| Parameter | Detail |
|---|---|
| Attribution goal | Identify the animal reservoir and food pathway responsible for a human outbreak cluster |
| Primary evidence type | Whole genome sequencing (WGS) of isolates from human, animal, food, and environmental sources |
| Core analytical logic | Phylogenetic clustering of human isolates with animal or food isolates supports a shared source |
| Key veterinary input | Contextual interpretation of animal isolates, production practices, and animal movement data |
| Major limitation | Genomic similarity does not prove transmission, epidemiological and traceback evidence are required |
| Standardizing bodies | WHO One Health framework, WOAH terrestrial animal health standards |
| Emerging challenge | Polyclonal and persistent sources show more sequence variation than point-source outbreaks |

## The Attribution Problem in Veterinary Epidemiology

Foodborne outbreak investigation has historically relied on case-control studies, food history questionnaires, and traceback of distribution records. These methods identify what people ate, but they do not always identify where the pathogen originated. Attribution adds a second layer of inference: linking the contaminated food item back to a farm, a species, or a production system.

The veterinary dimension of attribution is distinct from the public health dimension. Public health investigators ask which food vehicle caused illness. Veterinary investigators ask which animal population was the reservoir, how the pathogen entered the food chain, and what production practices permitted its amplification or spread. These questions require different data and different analytical approaches. The [WHO One Health initiative](https://www.who.int/health-topics/one-health) frames this as a cross-sectoral problem in which human, animal, and environmental health must be considered jointly, and the [CDC One Health and zoonotic disease resources](https://www.cdc.gov/one-health/index.html) similarly emphasize that presumed foodborne outbreaks are frequently transmitted through animal contact or environmental contamination instead of food alone.

Attribution studies operate at two scales. Outbreak-level attribution identifies the source of a specific cluster. Population-level attribution estimates the proportion of sporadic disease caused by each reservoir over time. Both scales depend on the same underlying logic: if human isolates and animal isolates from a defined production system share a recent common ancestor, the animal population is a plausible source.

## Genomic Basis of Source Attribution

Whole genome sequencing has replaced older typing methods for most attribution work because it provides the discriminatory power needed to distinguish closely related strains. The [review of typing methods based on whole genome sequencing data](https://pubmed.ncbi.nlm.nih.gov/33829127/) describes how WGS enables comparison of genetic relatedness between bacteria at the sub-species level and is being implemented across human, veterinary, food, and environmental sectors for outbreak investigation and source attribution. The same review notes that harmonization and standardization of typing tools remain urgent needs, because data comparability between laboratories is a precondition for cross-sectoral attribution.

The analytical workflow proceeds through several stages. Raw sequencing data are assembled into genomes, then compared using single nucleotide polymorphism (SNP) analysis, core genome multilocus sequence typing (cgMLST), or whole genome multilocus sequence typing (wgMLST). The choice of method depends on the research question and the genetic diversity of the pathogen. For point-source outbreaks caused by a monoclonal strain, a small number of SNP differences between isolates is expected. For persistent or polyclonal sources, the [review of One Health approaches to whole genome sequencing](https://pubmed.ncbi.nlm.nih.gov/31316960/) notes that outbreak strains that have propagated in non-human sources often show more sequence variation than is seen in typical monoclonal point-source outbreaks, which complicates case definition and cluster detection.

### Phylogenetic Inference and Source Prediction

Phylogenetic trees constructed from WGS data place human and animal isolates in relation to each other. If human outbreak isolates cluster within a lineage that also contains isolates from a particular livestock species, that species becomes a candidate source. The strength of the inference depends on the sampling frame. If veterinary isolates are sparse or collected from only one production sector, the absence of a match does not exclude that sector as a source.

The [genomic surveillance study of Salmonella Typhimurium in the United States](https://pubmed.ncbi.nlm.nih.gov/30561314/) demonstrated that major livestock sources can be predicted from WGS data alone. The investigators used a global phylogeny and found relatively steady rates of sequence divergence in livestock lineages, which allowed inference of recent origins. They then developed a machine learning Random Forest classifier that correctly attributed seven of eight major zoonotic outbreaks during 1998 to 2013 and identified fifty genetic features sufficient for robust livestock source prediction. This work established that source attribution can be approached as a classification problem instead of only a phylogenetic one.

### Host Adaptation Signals

Some foodborne pathogens show evidence of host adaptation, meaning that lineages become genetically and phenotypically specialized to particular animal species. The Salmonella Typhimurium study identified lineage-specific pseudogene accumulation after divergence from generalizt populations, suggesting ongoing host adaptation. For attribution, host-adapted lineages are easier to trace because their genetic signal points to a specific reservoir. Generalizt lineages that circulate across multiple species present a harder problem, because a human isolate may match isolates from several animal sources equally well.

## Study Designs for Attribution

Attribution evidence comes from several study designs, each with distinct strengths and failure modes. The choice of design depends on whether the question concerns a single outbreak or the broader burden of sporadic disease.

### Outbreak Case Studies

Retrospective analysis of resolved outbreaks provides the empirical basis for attribution. The [review of significant European foodborne outbreaks](https://pubmed.ncbi.nlm.nih.gov/34197583/) describes how the European surveillance system integrates elements from public and animal health and the food chain for early detection, assessment, and control. The outbreaks described in that review, including the 2011 Escherichia coli O104:H4 outbreak linked to sprouted seeds, the 2015 Listeria monocytogenes outbreak linked to frozen corn, and the 2016 Salmonella Enteritidis outbreak linked to eggs, illustrate the range of source types that attribution must accommodate. Each outbreak required rapid sharing of sequencing and tracing data, and each exposed gaps in harmonization of bioinformatics outputs.

### Population-Level Attribution Models

For sporadic disease, attribution models estimate the fraction of human cases originating from each reservoir. These models typically use pathogen subtype distributions in animal reservoirs and human cases, then apply Bayesian or frequentist methods to estimate source proportions. The quality of these estimates depends on representative sampling of animal reservoirs, which is a veterinary responsibility. Sampling that overrepresents one production sector or one geographic region will bias the attribution estimates.

### Longitudinal Surveillance

Attribution improves when veterinary surveillance is continuous instead of outbreak-driven. Routine collection and sequencing of isolates from healthy animals, slaughter plants, and retail meat creates a reference database against which human clinical isolates can be compared. The [review of dairyborne disease in developing countries](https://pubmed.ncbi.nlm.nih.gov/33076183/) notes that rigorous evidence on the full burden of foodborne disease is lacking in many regions, and that the most credible estimates come from the WHO Global Burden of Foodborne Disease initiative. In settings without routine veterinary surveillance, attribution relies on weaker evidence from outbreak reports and risk assessment.

## The Veterinary Contribution to Traceback

Traceback is the process of following a contaminated food item backward through distribution channels to its origin. Veterinarians contribute at the production end of the traceback chain. They document animal movements between premises, identify management practices that could introduce or amplify pathogens, and interpret whether conditions on a farm are consistent with the pathogen being present.

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 animal health surveillance and trade-related disease control that frame these activities. Veterinary records of animal identification, movement, and health status become evidentiary documents during an outbreak investigation. The quality of these records determines whether traceback can reach the farm level or stops at a distribution center.

A critical veterinary judgment is whether a pathogen recovered from an animal on a suspected farm is the outbreak strain or an unrelated strain. This requires comparing isolates from the farm with the human outbreak strain using the same genomic methods applied to human isolates. A match supports the farm as the source. The absence of a match does not exclude the farm, because the pathogen may have been present earlier and eliminated, or the sampling may have missed the affected animals.

## Limitations and Uncertainties

Attribution evidence has inherent limits that the veterinary researcher must recognize. Genomic similarity between human and animal isolates establishes a plausible link but does not prove transmission. The pathogen may have moved from humans to animals, or both populations may have acquired it from a common environmental source. The [review of fresh produce-associated listeriosis outbreaks](https://pubmed.ncbi.nlm.nih.gov/26818997/) illustrates this problem: Listeria monocytogenes outbreaks linked to produce involved serotypes and epidemic clones that had previously been associated with animal-derived foods, and the 2011 cantaloupe outbreak involved multiple strains in a single common-source event. These findings complicate the assumption that a single reservoir or a single strain explains an outbreak.

Sampling bias is a persistent concern. Veterinary isolates are more likely to be collected from clinically affected animals, from large operations, or from sectors with active surveillance programs. Pathogen populations in healthy carriers, small farms, or non-surveyed sectors are underrepresented. Attribution models that do not account for this bias will systematically overattribute cases to well-sampled reservoirs.

Finally, the evidence base for attribution in low- and middle-income countries is limited. The [dairyborne disease review](https://pubmed.ncbi.nlm.nih.gov/33076183/) notes that the global burden estimate for dairy products corresponds to approximately four percent of the global foodborne disease

## Practical Attribution Workflow in Veterinary Field Investigation

The veterinary contribution to source attribution begins when a foodborne cluster is recognized and ends when the animal reservoir, production pathway, or point of contamination is identified with sufficient confidence to support control measures. The sequence below reflects current practice in integrated surveillance systems, where [whole genome sequencing bridges human, animal, food, and environmental sectors](https://pubmed.ncbi.nlm.nih.gov/31316960/).

### Initial Triage and Case Definition

The first decision is whether the cluster warrants veterinary investigation at all. A veterinary epidemiologist should be engaged when the outbreak strain has a known animal reservoir, when the food vehicle is plausibly of animal origin, or when the case questionnaire reveals exposure to livestock, poultry, or their products. For produce-associated outbreaks, veterinary input is still relevant because [fresh produce can be contaminated by animal manure, irrigation water, or direct animal intrusion](https://pubmed.ncbi.nlm.nih.gov/26818997/).

The case definition must be flexible. [Outbreak strains that persist in animal reservoirs often show more sequence variation than typical monoclonal point-source outbreaks](https://pubmed.ncbi.nlm.nih.gov/31316960/), so a narrow single-nucleotide polymorphism threshold may exclude true cases. The veterinary epidemiologist should review the genomic cluster definition with the laboratory and agree on a threshold that accommodates the expected diversity of the suspected reservoir species.

### Farm-Level Sampling Strategy

Once the suspect commodity or production system is identified, the sampling plan must target the point in the production chain where contamination is most likely to have entered. The plan changes with the species and system:

- **Dairy herds**: sample bulk tank milk, milking equipment surfaces, teat skin, and bedding. Bulk tank culture detects shedding but not the individual animal, follow-up individual cow sampling is required when the herd is small or when culling decisions are needed.
- **Poultry flocks**: sample caecal contents at slaughter, drag swabs from the house, and feed. Vertical transmission through the breeder flock must be considered when the same strain appears across multiple houses.
- **Swine operations**: sample fecal pats, lairage pens, and carcass surfaces after slaughter. The lairage environment is a common cross-contamination point.
- **Produce fields**: sample irrigation water, soil, and animal intrusion evidence such as scat. The 2011 cantaloupe outbreak demonstrated that [multiple strains can be implicated in a single common-source outbreak](https://pubmed.ncbi.nlm.nih.gov/26818997/), so sampling must not assume clonality.

Sample numbers should be guided by expected prevalence and detection sensitivity. For a herd-level prevalence of 5%, sampling 60 animals detects at least one positive with 95% confidence. For environmental samples, collect at least 10 to 20 sites per production unit, prioritizing wet areas, drainage points, and surfaces with visible organic material.

### Genomic Data Interpretation at the Farm Level

The veterinary epidemiologist must interpret genomic results in the context of the farm's history. A key distinction is whether the isolate from the farm is the outbreak strain or a closely related but distinct strain. [Whole genome sequencing provides highly discriminative power for comparing genetic relatedness between bacteria even at the sub-species level](https://pubmed.ncbi.nlm.nih.gov/33829127/), but the interpretation of that relatedness requires epidemiological context.

A farm isolate that differs from the human outbreak strain by more than the expected within-outbreak diversity may represent an older introduction, a different lineage, or a false association. The decision to implicate a farm depends on:

1. Genomic distance consistent with a recent common ancestor
2. Temporal correlation between farm contamination and human cases
3. Plausible transmission pathway from farm to food to human
4. Absence of an alternative source

### Attribution Decision Framework

| Evidence Category | Strong Support | Moderate Support | Weak Support |
|---|---|---|---|
| Genomic match | Isolates differ by ≤ 5 SNPs | Isolates differ by 6 to 20 SNPs | Isolates differ by > 20 SNPs |
| Temporal link | Farm contamination precedes or coincides with human cases | Farm contamination detected after human cases but within outbreak period | No temporal data available |
| Exposure pathway | Documented movement of product from farm to outbreak venue | Plausible pathway but gaps in traceback records | No pathway identified |
| Prevalence in farm | Outbreak strain is dominant type in farm samples | Outbreak strain present but minority type | Strain detected only after extensive sampling |
| Alternative sources | Ruled out by traceback and environmental testing | Partially ruled out | Not investigated |

Thresholds for SNP distances vary by pathogen and should be established from the local baseline diversity. The table provides a decision aid, not a fixed rule.

### Case Example: A Polyclonal Egg-Associated Outbreak

A multinational cluster of Salmonella Enteritidis infections was linked to eggs through case-control studies. The outbreak was unusual because [the strain was polyclonal, a pattern more consistent with a persistent reservoir than a point-source contamination event](https://pubmed.ncbi.nlm.nih.gov/34197583/). The veterinary investigation traced the eggs to a single laying hen operation with multiple houses.

Sampling revealed the outbreak strain in two of six houses, but with different SNP profiles in each house. The genomic diversity suggested the strain had been circulating in the flock for several weeks before detection. The investigation identified the likely entry point as contaminated feed delivered three weeks before the first human cases. The feed mill was sampled and yielded the same strain, confirming the attribution chain.

The control measures included depopulation of the affected houses, feed mill sanitation, and enhanced biosecurity. The investigation succeeded because the veterinary team sampled broadly across the production system instead of focusing only on the eggs, and because the genomic data were interpreted with awareness that [persistent polyclonal outbreaks require different case definitions than monoclonal point-source events](https://pubmed.ncbi.nlm.nih.gov/31316960/).

### Documentation and Data Sharing

The veterinary investigation must produce a written record that supports regulatory action and future reference. The record should include:

- Sampling dates, locations, and methods
- Laboratory accession numbers and genomic data files
- SNP distances between farm isolates and human outbreak strains
- Herd health history, including vaccination and antimicrobial use
- Movement records for animals, feed, and personnel
- Corrective actions taken and verification sampling results

The record must be shareable across sectors. [Rapid sharing of sequencing and tracing data is essential for detecting and investigating foodborne illnesses](https://pubmed.ncbi.nlm.nih.gov/34197583/), and the veterinary report is often the document that links the agricultural and public health responses. Data formats should follow the standards used by national reference laboratories, and the report should state explicitly which analyzes were performed and which software versions were used.

### Species-Specific Adjustments

The workflow changes when the suspect species is not a typical food animal. Wildlife reservoirs require different sampling methods and may involve regulatory constraints on animal handling. Companion animals are rarely implicated in foodborne outbreaks but can serve as bridging hosts between livestock and humans, particularly for Salmonella and Campylobacter. In developing countries, [dairyborne disease accounts for a substantial share of the foodborne burden](https://pubmed.ncbi.nlm.nih.gov/33076183/), but laboratory capacity and traceback infrastructure may be limited, requiring the veterinary investigator to rely more heavily on epidemiological questionnaires and targeted sampling of high-risk points in the informal milk chain.

The correct choice of sampling strategy, genomic threshold, and control measure depends on the production system, the pathogen, and the available laboratory capacity. The veterinary epidemiologist must adapt the framework to the local context instead of applying a fixed protocol.

## Recognized Failure Modes in Attribution Work

Attribution fails in characteriztic patterns. The most common is overmatching, where a farm isolate shares core genome similarity with a human outbreak strain but differs in accessory content or in single nucleotide polymorphisms that fall below the routine cluster threshold. Whole genome sequencing offers high discriminatory power, but the interpretation of relatedness depends on the chosen bioinformatics pipeline and the comparator set [Typing methods based on whole genome sequencing data](https://pubmed.ncbi.nlm.nih.gov/33829127/). A cluster defined at 10 allele differences by one scheme may split at five by another. Early detection requires running both core genome multilocus sequence typing and single nucleotide polymorphism phylogenies in parallel and inspecting the support values at the nodes that separate farm from human isolates.

A second failure mode is sampling bias in the farm environment. Sampling only clinically affected animals misses the carrier state that characterizes many foodborne pathogens, particularly Salmonella in swine and cattle. The corrective action is to sample the production environment, feed, water, and pooled feces from healthy cohorts, and to collect samples before antimicrobial therapy is initiated. A third mode is temporal mismatch. The farm isolate may be genetically indistinguishable from the outbreak strain but collected months after the exposure window, reflecting persistence instead of causation. Discriminating checks include reviewing movement records, feed deliveries, and the introduction of new stock against the outbreak timeline.

A fourth failure is misattribution of the food vehicle. Outbreak strains that circulate in animal reservoirs can contaminate produce through water, manure, or wildlife, and the veterinary investigation may identify a plausible animal source that is not the proximate vehicle [Whole Genome Sequencing: Bridging One-Health Surveillance of Foodborne Diseases](https://pubmed.ncbi.nlm.nih.gov/31316960/). The farm finding is then a reservoir signal, not a food vehicle signal, and the traceback must continue up the supply chain.

| Observation | Likely cause | Discriminating check |
|---|---|---|
| Farm isolate matches human cluster at core genome level | True common source or shared ancestral lineage | Compare accessory genome content and SNP distances, inspect phylogenetic tree topology |
| Farm isolate matches but from a different time window | Persistent contamination or unrelated lineage | Align collection dates with outbreak exposure period, review movement records |
| Human isolates are polyclonal | Multiple sources, or a source with a diverse population | Perform source attribution on each subcluster separately |
| Produce implicated but animal reservoir positive | Environmental transmission from manure, water, or wildlife | Sample irrigation water and soil, genotype wildlife and domestic animal isolates |
| No farm isolate matches | Sampling gap or non-animal source | Expand sampling to lairage, transport, and feed mill, consider imported product |

## Common Errors in Field Investigation

Less experienced investigators often collect too few isolates per farm. A single positive sample cannot distinguish a point contamination event from an established within-herd infection. Collect at least five to ten isolates per positive farm across different pens, age groups, and sample types before concluding that the farm is the source. A second error is overinterpreting a negative result. Absence of the outbreak strain in a small sample set does not exclude the farm, particularly for pathogens with intermittent shedding. The corrective action is to calculate the detection probability for the sampling scheme used and to report it in the investigation record.

A third error is treating the first matching isolate as the answer. The veterinary contribution is strongest when it rules out alternative farms with equal genomic matches. Compare all candidate farms using the same sampling intensity and the same bioinformatics parameters. A fourth error is neglecting the interval between farm collection and human exposure. Processed products may be stored, distributed, and consumed weeks after slaughter, and the farm isolate must be compatible with that lag.

## Evidence Gaps and Divergent Expert Opinion

The evidence base for genomic source attribution is strongest for Salmonella Typhimurium, where machine learning classifiers trained on livestock lineages have retrospectively attributed a majority of major outbreaks [Zoonotic Source Attribution of Salmonella enterica Serotype Typhimurium Using Genomic Surveillance Data, United States](https://pubmed.ncbi.nlm.nih.gov/30561314/). For other pathogens and for non-typhimurium serovars, the predictive accuracy is less well established. Expert opinion differs on whether host-adapted lineages can be distinguished reliably from generalizt populations that merely circulate in livestock. The accumulation of pseudogenes and metabolic changes in some livestock lineages suggests host adaptation, but the rate of such change varies and may confound attribution models [Zoonotic Source Attribution of Salmonella enterica Serotype Typhimurium Using Genomic Surveillance Data, United States](https://pubmed.ncbi.nlm.nih.gov/30561314/).

Attribution in low- and middle-income countries is further constrained by limited sequencing capacity and by dairy systems where milk is pooled across many smallholders, making farm-level traceback impractical [MILK Symposium review: Foodborne diseases from milk and milk products in developing countries-Review of causes and health and economic implications](https://pubmed.ncbi.nlm.nih.gov/33076183/). Expert opinion diverges on whether genomic surveillance should be prioritized over basic hygiene interventions in these settings. The harmonisation of typing tools across laboratories remains incomplete, and comparisons between jurisdictions can be unreliable until standardized pipelines are adopted [Typing methods based on whole genome sequencing data](https://pubmed.ncbi.nlm.nih.gov/33829127/).

## Escalation and Referral Criteria

Referral to a reference laboratory is warranted when the farm isolate is a suspected match but the bioinformatics evidence is equivocal, when the pathogen is unusual or has not been isolated previously from the production system, or when the investigation requires methods beyond the capacity of the submitting laboratory. Specialist consultation with a veterinary epidemiologist is indicated when the outbreak is prolonged, polyclonal, or spans multiple farms or jurisdictions.

Regulatory reporting obligations vary by jurisdiction and by pathogen. Reportable conditions should be notified to the competent authority at the earliest point of suspicion, not after laboratory confirmation. International trade implications may arise when a notifiable pathogen is detected in a food animal population, and the [WOAH terrestrial animal health standards](https://www.woah.org/en/what-we-do/standards/codes-and-manuals/terrestrial-code-online-access/) provide the framework for such reporting. The [CDC One Health and zoonotic disease resources](https://www.cdc.gov/one-health/index.html) describe the cross-sector coordination expected during multistate or multinational investigations. When the investigation crosses human, animal, and environmental sectors, the [WHO One Health initiative](https://www.who.int/health-topics/one-health) provides the collaborative structure for aligning case definitions and data sharing.

## Frequently Asked Questions

### How Much Does Whole-Genome Sequencing Cost for Routine Attribution Work, and Who Bears That Cost?

Sequencing costs have fallen substantially, but the full attribution workflow includes library preparation, sequencing, bioinformatic analysis, and secure data storage. Public health laboratories in many regions absorb these costs during outbreak investigations, and veterinary diagnostic laboratories increasingly offer sequencing as a fee-for-service option. For a single farm investigation, the cost may be justified when human cases are linked to a suspected animal source. When resources are limited, targeted gene-based typing of a smaller isolate set can provide preliminary clustering information at lower cost. [Whole genome sequencing approaches for foodborne pathogens](https://pubmed.ncbi.nlm.nih.gov/33829127/) describe a range of bioinformatic tools with varying complexity, so laboratories can match analytical depth to available expertise and budget.

### What Can a Practitioner Do When Sequencing Is Unavailable or Delayed?

When sequencing is not immediately accessible, conventional serotyping and pulsed-field gel electrophoresis remain useful for excluding obvious mismatches between human and animal isolates. These methods lack the discriminatory power of whole-genome sequencing, so negative or weak matches should be interpreted cautiously. [Genomic surveillance approaches for Salmonella source prediction](https://pubmed.ncbi.nlm.nih.gov/30561314/) demonstrate that sequence-based methods can attribute outbreaks to livestock sources with high accuracy, a performance level that phenotypic typing cannot match. In the interim, focus on structured epidemiological data: collection dates, animal movements, feed sources, and on-farm biosecurity records. These data remain essential regardless of the typing method and can be shared with reference laboratories that will perform sequencing later. Document all sample handling and storage conditions to preserve isolate viability for retrospective analysis.

### How Does Attribution Differ for Dairy Operations Compared with Confined Feeding Operations?

Dairy operations present distinct attribution challenges because of continuous animal movement, commingling at collection points, and bulk tank sampling that represents multiple animals. A positive bulk tank sample identifies herd-level exposure but not the individual shedder. In confined feeding operations, cohort structure and all-in-all-out management simplify source tracing because isolates can be linked to a defined group. [Dairyborne disease burden in developing countries](https://pubmed.ncbi.nlm.nih.gov/33076183/) notes that dairy products contribute a substantial share of animal-source foodborne disease globally, which underscores the need for attribution methods suited to this production system. For dairy, repeated bulk tank sampling over time and sampling of individual animals in the milking string can narrow the source. For confined operations, sampling by pen or barn section and correlating with feed delivery records often resolves the question more quickly.

### What Records Should a Veterinary Practice Maintain to Support Future Attribution Investigations?

Maintain a permanent log of every sample submitted for typing, including the animal identification, collection date, anatomical site, farm location, and clinical context. Record the laboratory, the method used, and the resulting type designation. Keep a separate log of isolate storage locations and freezer coordinates. [One Health surveillance frameworks](https://www.cdc.gov/one-health/index.html) emphasize that cross-sector data sharing depends on consistent record keeping across human, animal, and environmental sectors. Photograph sample collection sites and label containers with permanent markers that resist freezer conditions. When isolates are shared with public health laboratories, document the chain of custody and any restrictions on data release. These records become critical when an outbreak is recognized months after the original samples were collected.

### How Should a Veterinarian Explain a Source Attribution Finding to a Producer?

Explain the distinction between association and causation. A matching sequence type between a farm isolate and a human case indicates a plausible link, not proof that the farm caused the outbreak. Describe the investigative process in terms of probability and converging evidence, including epidemiological data and traceback findings. [One Health approaches to outbreak investigation](https://pubmed.ncbi.nlm.nih.gov/31316960/) show that outbreaks are often solved through combined evidence from human, animal, and environmental sectors. Emphasize that attribution findings guide control measures, not blame. Discuss the producer's legal obligations regarding notifiable diseases and food safety programs, and recommend consulting the [AVMA practice resources](https://www.avma.org/resources-tools) for guidance on professional communication and liability considerations. Offer to review biosecurity and monitoring plans that can reduce future risk.

### When Should a Veterinary Practice Refer an Attribution Investigation to a Higher Authority?

Refer when the investigation crosses jurisdictional boundaries, when human cases are involved, or when the suspected source involves commercial distribution networks beyond the individual farm. Refer also when the practice lacks the laboratory capacity or bioinformatic expertise to interpret sequencing data reliably. [European multistate outbreak investigations](https://pubmed.ncbi.nlm.nih.gov/34197583/) illustrate that large outbreaks require rapid data sharing across regions and sectors, which individual practices cannot coordinate. Refer when wildlife or environmental contamination is suspected, because these investigations require specialised sampling strategies and regulatory authority. Finally, refer when the producer is uncooperative or when legal proceedings are anticipated. Early referral preserves evidence integrity and prevents duplication of effort.

## 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

- [Typing methods based on whole genome sequencing data.](https://pubmed.ncbi.nlm.nih.gov/33829127/). 2020.
- [Zoonotic Source Attribution of Salmonella enterica Serotype Typhimurium Using Genomic Surveillance Data, United States.](https://pubmed.ncbi.nlm.nih.gov/30561314/). 2019.
- [Whole Genome Sequencing: Bridging One-Health Surveillance of Foodborne Diseases.](https://pubmed.ncbi.nlm.nih.gov/31316960/). 2019.
- [A Review of Significant European Foodborne Outbreaks in the Last Decade.](https://pubmed.ncbi.nlm.nih.gov/34197583/). 2021.
- [Fresh Produce-Associated Listeriosis Outbreaks, Sources of Concern, Teachable Moments, and Insights.](https://pubmed.ncbi.nlm.nih.gov/26818997/). 2016.
- [MILK Symposium review: Foodborne diseases from milk and milk products in developing countries-Review of causes and health and economic implications.](https://pubmed.ncbi.nlm.nih.gov/33076183/). 2020.
- [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

- [Foodborne Outbreak Investigation: Veterinary Roles in Traceback and Source Attribution](/knowledge/veterinary-medicine/veterinary-public-health/foodborne-outbreak-investigation-veterinary-roles-traceback-source-attribution)
- [Foodborne Pathogen Outbreak Investigation: Veterinary Roles](/knowledge/veterinary-medicine/veterinary-public-health/foodborne-pathogen-outbreak-investigation-veterinary-roles)
- [Wildlife Disease Outbreak Investigation: A Veterinary Field Guide](/knowledge/veterinary-medicine/veterinary-public-health/wildlife-disease-outbreak-investigation-a-veterinary-field-guide)
- [Diagnostic Approaches for Foodborne Pathogens in Veterinary Samples](/knowledge/veterinary-medicine/veterinary-public-health/diagnostic-approaches-foodborne-pathogens-veterinary-samples)
- [Foodborne Zoonoses: Comparative Pathogenesis and Clinical Manifestations in Animals and Humans](/knowledge/veterinary-medicine/veterinary-public-health/foodborne-zoonoses-comparative-pathogenesis-clinical-manifestations-animals-humans)

> 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.