# Purposive Sampling in Veterinary Outbreak Investigations


## Key Takeaways

- Purposive sampling in veterinary outbreak investigations prioritizes case detection, hypothesis generation, and pathogen characterization over population prevalence estimation, focusing on animals with specific clinical signs (e.g., vesicular lesions in swine vesicular disease) or epidemiological links to the index case.
- This non-probability sampling strategy is most defensible during the early phases of an outbreak, including index case identification and contact tracing, where rapid identification of affected units is critical for control measures.
- Rigorous execution demands predefined, justified selection criteria grounded in pathogen biology and outbreak patterns (e.g., temporal proximity to exposure, spatial clustering), distinguishing it from convenience sampling which lacks explicit objective linkage.
- Purposive sampling is inherently susceptible to selection bias, which cannot be statistically corrected post-collection; therefore, transparent documentation of selection rationale and limitations is paramount.
- For quantitative objectives like prevalence estimation or vaccine efficacy assessment, purposive sampling must be integrated with probability-based designs, typically in a sequential approach where purposive sampling identifies affected clusters for subsequent random sampling.
- Documentation of purposive sampling must meticulously record selection criteria, reasons for inclusion/exclusion, clinical findings, and temporal relationships to enable correct interpretation and potential transition to probability-based methods.

---

Purposive sampling is a non-probability sampling strategy in which the investigator selects units based on predefined criteria relevant to the investigation's objectives. In veterinary outbreak investigations, this approach is used when the goal is not to estimate population parameters but to detect cases, characterize an emerging pathogen, identify risk factors, or generate hypotheses under time and resource constraints. This article provides a procedural reference for veterinary researchers and field epidemiologists who must decide when purposive sampling is defensible, how to execute it rigorously, and how to integrate it with probability-based designs when the investigation requires both case detection and unbiased estimation.

The central question addressed here is practical: during an outbreak, when should you deliberately select animals, herds, or premises instead of rely on random selection? The answer depends on the investigation phase, the objective of sampling, the suspected pathogen, and the consequences of missing a case. This article covers the scientific logic of purposive sampling, its specific applications in outbreak detection and hypothesis generation, its limitations, and strategies for combining it with probability sampling to preserve inferential validity where it matters.

## At a Glance

| Parameter | Decision or Fact |
|---|---|
| Primary objective | Case detection, hypothesis generation, or characterization, not prevalence estimation |
| Sampling logic | Selection based on predefined criteria linked to exposure, clinical signs, or risk status |
| Appropriate phase | Early outbreak response, index case identification, contact tracing, emerging disease investigation |
| Inappropriate use | Prevalence estimation, regulatory certification of freedom from disease, vaccine efficacy assessment |
| Key risk | Selection bias that cannot be quantified or corrected after data collection |
| Combination strategy | Use purposive sampling for case finding, then probability sampling within detected clusters for estimation |
| Documentation requirement | Predefine selection criteria, record reasons for inclusion and exclusion, report sampling logic transparently |
| Reference standards | [WOAH terrestrial animal health standards](https://www.woah.org/en/what-we-do/standards/codes-and-manuals/terrestrial-code-online-access/) and [CDC outbreak investigation methods](https://www.cdc.gov/csels/dsepd/ss1978/index.html) |

## The Logic of Purposive Sampling in Outbreak Epidemiology

Outbreak investigations differ fundamentally from prevalence surveys. In a prevalence survey, the objective is to estimate how much disease exists in a defined population, which requires probability sampling so that every unit has a known, nonzero chance of selection. In an outbreak, the immediate objectives are to identify the causative agent, find as many cases as possible, describe the population at risk, and generate hypotheses about transmission. These objectives favour deliberate selection of informative units over random selection.

The [CDC principles of epidemiology in public health practice](https://www.cdc.gov/csels/dsepd/ss1978/index.html) describe outbreak investigation as an iterative process that begins with case confirmation and descriptive epidemiology before analytic studies are designed. During the descriptive phase, the investigator needs cases, not a representative sample. Purposive sampling maximizes the yield of cases per unit sampled, which is critical when the outbreak is explosive, the case definition is narrow, or laboratory confirmation is slow.

Purposive sampling is also the method of choice when the population at risk is small or difficult to enumerate. In a backyard poultry outbreak, for example, the investigator may not have a complete list of premises. Random sampling requires a sampling frame, purposive sampling does not. The investigator can instead work through local veterinarians, feed suppliers, or community animal health workers to identify premises with sick birds. This approach trades representativeness for feasibility.

## Selection Criteria and Their Justification

The validity of purposive sampling rests entirely on the relevance of the selection criteria to the investigation's objective. Criteria must be specified before sampling begins and must be grounded in the biology of the suspected pathogen, the production system, and the outbreak's observed pattern.

Common selection criteria in veterinary outbreak investigations include:

- Clinical compatibility with the case definition, particularly when the case definition includes pathognomonic or highly suggestive signs.
- Temporal proximity to the suspected index case or to a common source exposure.
- Spatial clustering, such as premises within a defined radius of a confirmed case.
- Management or production characteriztics that plausibly affect exposure, such as open versus closed herds, grazing practices, or biosecurity protocols.
- Species, age, or physiological status groups expected to be most susceptible or most likely to shed the agent.

The [MSD Veterinary Manual professional edition](https://www.msdvetmanual.com/) provides species-specific guidance on clinical presentation and lesion patterns that can inform these criteria. For example, in a suspected swine vesicular disease outbreak, selection of animals with vesicular lesions on the snout and coronary bands is purposive and justified because those lesions are the most reliable clinical indicator. Sampling unaffected animals from the same cohort may be equally purposive if the objective is to compare exposure histories.

A critical distinction must be made between purposive sampling and convenience sampling. Convenience sampling selects whatever is easiest to access, with no explicit link to the investigation's objectives. Purposive sampling selects units because they carry information the investigation needs. The distinction matters because convenience sampling is rarely defensible in an outbreak investigation, whereas purposive sampling can be, provided the criteria are explicit and the reasoning is documented.

## Purposive Sampling in Case Finding and Contact Tracing

Case finding is the most common and most defensible use of purposive sampling in outbreak investigations. When a reportable disease is suspected, the investigator must identify all affected and exposed animals quickly to implement control measures. The [WOAH animal health surveillance standards](https://www.woah.org/en/what-we-do/animal-health-and-welfare/disease-data-collection/) emphasize that surveillance during an outbreak is directed at detecting infection, not at estimating its prevalence. Purposive sampling aligns with this objective.

In a herd outbreak, the investigator typically samples animals that are clinically affected, animals that have been in direct contact with affected animals, and animals that share a common exposure source such as contaminated feed, water, or transport. This is a form of purposive sampling known as snowball sampling, in which each detected case leads to the identification of additional candidates. The approach is efficient because transmission networks are often local and the highest-risk contacts are identifiable through husbandry records and observation.

Contact tracing requires a working case definition and a clear protocol for classifying contacts. The investigator must decide how far back in time to trace contacts, which species to include, and what constitutes sufficient exposure to warrant sampling. These decisions are inherently purposive. They should be made by a team that includes the attending veterinarian, an epidemiologist, and, where relevant, a laboratory diagnostician, and they should be reviewed as new information emerges.

## Combining Purposive and Probability Sampling

A common error in outbreak investigations is to use purposive sampling for every question the investigation must answer. Purposive sampling cannot support unbiased estimates of prevalence, incidence, or association measures. When the investigation reaches the analytic phase, the investigator needs a probability sample or a census of a defined population.

The solution is a sequential design. In the first phase, purposive sampling identifies cases and defines the affected population. In the second phase, the investigator constructs a sampling frame from the affected premises or animals identified in phase one and draws a probability sample for detailed study. This two-phase approach preserves the efficiency of purposive sampling for case detection while restoring the inferential basis for estimation.

For example, in a multi-herd outbreak of a respiratory pathogen, phase one might use purposive sampling to identify affected herds through veterinary reports and laboratory submissions. Phase two might then use stratified random sampling of animals within each affected herd to estimate within-herd prevalence and to compare risk factors between affected and unaffected herds. The [WOAH terrestrial animal health standards](https://www.woah.org/en/what-we-do/standards/codes-and-manuals/terrestrial-code-online-access/) provide guidance on the surveillance design requirements that apply when results will be used for trade or official disease status purposes, and these requirements generally mandate probability-based methods for estimation.

The decision to combine designs should be made at the outset of the investigation, not after data collection is complete. Once a purposive sample has been collected, there is no statistical procedure that can recover representativeness. The investigator who anticipates needing prevalence estimates must build the probability component into the design from the beginning.

## Field Application: Purposive Sampling in Active Outbreak Response

### The Assessment Sequence for Sampling Site Selection

When an outbreak is confirmed, the investigation team must decide where to sample first. The sequence begins with the index case report, then moves to premises-level risk assessment, and finally to within-herd or within-flock sampling strategy. Each step narrows the sampling frame and sharpens the purpose of selection.

The first decision is geographic. Purposive selection of premises should follow the epidemiological curve and the movement records, not convenience. Premises with direct contact to the index case, shared equipment, or personnel movement rank highest. Next come premises linked by common water sources, feed deliveries, or wildlife interfaces. The [WOAH terrestrial animal health code](https://www.woah.org/en/what-we-do/standards/codes-and-manuals/terrestrial-code-online-access/) provides the international framework for movement controls and surveillance zones that define the geographic boundaries within which purposive sampling operates.

Within a selected premises, the second decision is animal-level. The investigator chooses individuals that maximize diagnostic sensitivity. Animals with fever, reduced feed intake, or characteriztic lesions are sampled first. In a foot-and-mouth disease outbreak, vesicular epithelium from animals with active lesions yields the highest diagnostic return. In avian influenza, oropharyngeal and cloacal swabs from moribund birds outperform samples from recovered or subclinical birds. The [CDC principles of epidemiology](https://www.cdc.gov/csels/dsepd/ss1978/index.html) describe this logic as sampling based on the probability of exposure and the probability of detectable infection, both of which are maximized by targeted selection during the acute phase.

The third decision is temporal. Sampling must occur within the diagnostic window for the pathogen in question. For vesicular viruses, lesion epithelium remains diagnostically useful for only a few days after vesicle rupture. For influenza viruses in poultry, viral shedding peaks within the first 48 hours of clinical signs. Delayed purposive sampling of recovered animals produces false negatives and misdirects the response.

### Decision Points That Change the Sampling Approach

The correct purposive approach shifts with the outbreak phase, the production system, and the pathogen biology.

During the index case investigation, exhaustive sampling of all clinically affected animals is appropriate. The goal is pathogen detection and characterization, not prevalence estimation. Every suspect animal should be sampled because the cost of a missed diagnosis exceeds the cost of excess testing.

During the tracing phase, purposive sampling becomes selective. The investigator samples a defined number of high-risk contacts per premises, typically the most clinically suspicious animals, instead of all animals. The number sampled depends on the within-group prevalence expected for the pathogen and the diagnostic sensitivity of the test. For highly transmissible agents such as foot-and-mouth disease virus, sampling five to ten clinically affected animals per epidemiological group is usually sufficient for detection. For slower-spreading pathogens, more animals may be needed.

The production system changes the practical execution. In a commercial poultry house, purposive sampling of sick birds requires catching teams and may disturb the flock. In a backyard swine operation, the investigator can examine and sample individual animals with minimal disruption. The [MSD Veterinary Manual](https://www.msdvetmanual.com/) notes that sampling approaches must be adapted to the species and husbandry system because restraint options, biosecurity constraints, and owner cooperation vary substantially.

Patient status changes the decision. Animals that have died within the past few hours may yield better samples than live animals in some diseases because agonal viraemia concentrates pathogen in tissues. However, autolysis degrades RNA rapidly, so sampling must occur quickly after death. The investigator must weigh the diagnostic gain against the logistical cost of necropsy under field conditions.

### Structured Comparison of Purposive Sampling Strategies

| Strategy | Selection Basis | Best Use | Primary Limitation | Typical Sample Volume |
|---|---|---|---|---|
| Extreme case sampling | Animals with most severe clinical signs | Index case detection, pathogen isolation | May miss mild or atypical presentations | 5 to 10 animals per group |
| Typical case sampling | Animals with representative clinical signs | Descriptive characterization of outbreak strain | Requires clear case definition | 10 to 20 animals per group |
| Snowball sampling | Referrals from identified cases | Tracing movements through social or trade networks | Misses isolated or non-networked premises | Variable, network dependent |
| Key informant sampling | Local experts identify affected premises | Rapid geographic scoping in remote areas | Informant bias, incomplete coverage | Depends on informant reliability |
| Maximum variation sampling | Premises across different production types, ages, or regions | Understanding outbreak range and risk factors | Requires prior knowledge of variation | 3 to 5 premises per stratum |

The choice among these strategies depends on the investigation objective. If the objective is pathogen detection, extreme case sampling dominates. If the objective is describing the clinical spectrum, typical case sampling is preferred. If the objective is mapping the outbreak extent, maximum variation sampling across premises is most informative.

### Documentation and Data Recording During Purposive Sampling

Purposive sampling generates data that are not statistically representative, so documentation must capture the selection rationale explicitly. Each sample record should include the reason the animal or premises was selected, the clinical findings that triggered selection, and the temporal relationship between clinical onset and sampling.

The investigator should record the case definition used, the number of animals examined versus the number sampled, and the reasons for exclusion. This documentation allows later interpretation of test results and supports the transition to probability-based sampling if the investigation shifts toward prevalence estimation.

Standardized forms reduce omission errors. A field form for purposive sampling should include premises identification, geographic coordinates, production type, animal identification, clinical signs with onset date, sample type, sample preservation method, and the specific selection criterion. Photographs of lesions and gross pathology should accompany written descriptions where possible.

The [AVMA practice resources](https://www.avma.org/resources-tools) emphasize that contemporaneous record keeping is a professional obligation, also a research convenience. In outbreak investigations, records may be scrutinised by regulatory authorities, insurers, or legal counsel. Incomplete documentation of sampling rationale undermines the credibility of the entire investigation.

### Transitioning from Purposive to Probability Sampling

Purposive sampling answers the question "Is this pathogen present and where is it spreading?" It does not answer "What proportion of the population is infected?" The transition to probability sampling occurs when the investigation shifts from detection and containment to surveillance and freedom-from-disease demonstration.

The trigger for transition is operational. Once the outbreak is contained and the immediate risk is managed, the investigation team must estimate the extent of spread. This requires a sampling frame, random selection, and sample size calculations based on expected prevalence and desired confidence. The [WOAH animal health surveillance standards](https://www.woah.org/en/what-we-do/animal-health-and-welfare/disease-data-collection/) specify the confidence levels and detection thresholds expected for demonstrating freedom from infection in different disease contexts.

A common hybrid approach uses purposive sampling to identify affected groups and then applies stratified random sampling within those groups to estimate prevalence. This two-phase design preserves the diagnostic sensitivity of purposive selection while adding the statistical validity of probability sampling for the quantitative component.

The investigator must document the point at which the sampling approach changed and the justification for the change. This transparency allows downstream analysts to interpret the combined dataset correctly and avoids the common error of treating purposively collected data as if it were randomly sampled.

### Species-Specific and System-Specific Adaptations

In dairy cattle herds, purposive sampling of bulk tank milk from suspect premises provides a low-cost screening tool, but individual animal sampling is required for confirmation. In beef cattle on extensive range, mustering costs may exceed laboratory costs, so the investigator should sample the first animals gathered that meet the clinical case definition instead of insisting on a fixed number.

In commercial poultry, the unit of sampling is often the house instead of the individual bird. Purposive selection of the most affected house within a multi-house farm is standard practice, followed by sampling of sick birds within that house. In free-range layers, access to outdoor areas changes exposure patterns, and the investigator must sample birds from different range zones.

In aquaculture, purposive sampling of moribund fish from the inflow and outflow ends of a cage or pond can reveal whether the pathogen is entering or leaving the system. This spatial purposive sampling provides information that random sampling would dilute.

In wildlife, purposive sampling of found-dead animals or animals showing abnormal behavior is often the only feasible approach. The [MSD Veterinary Manual](https://www.msdvetmanual.com/) notes that wildlife disease surveillance relies heavily on convenience and purposive sampling because random sampling of free-ranging populations is rarely practical. The investigator must acknowledge the resulting detection bias when interpreting negative results.

The correct choice of sampling approach always depends on the question being asked, the pathogen biology, the production system, and the resources available. Purposive sampling is not a compromise. It is a deliberate, defensible strategy for the detection and tracing phases of outbreak investigation, and it becomes the foundation upon which later probability-based sampling is built.

## Recognized Failure Modes and Early Detection

Purposive sampling in outbreak investigations fails in predictable ways. The most consequential failure is selection bias that goes unrecognised because the investigation team mistakes a purposive sample for a representative one. This occurs when case finding concentrates on the most accessible premises, the most cooperative owners, or the most visibly affected animals. The resulting sample overrepresents severe clinical presentations and underrepresents subclinical infections, which distorts the apparent case fatality rate and the estimated speed of spread.

A second failure mode is premature saturation of the sampling frame. When investigators continue purposive sampling after new selections stop yielding novel information, they waste resources and may reinforce existing assumptions. This is detected by tracking the yield of new findings per sampling unit. When three consecutive selections produce no new premises, no new risk factors, and no new case types, the purposive phase has served its purpose.

A third failure is the loss of the sampling rationale during the investigation. Teams that begin with explicit criteria for selecting premises often drift toward convenience as fatigue sets in. The corrective action is to record the selection justification for every sampled unit at the time of sampling, not retrospectively. The documentation standards used in structured surveillance programs provide a useful template for this discipline, as described in the [World Organization for Animal Health surveillance standards](https://www.woah.org/en/what-we-do/animal-health-and-welfare/disease-data-collection/).

A fourth failure mode is the failure to recognize when purposive sampling has generated a hypothesis that requires a different design to test. Purposive sampling identifies candidate risk factors and putative transmission pathways. It cannot quantify their relative contribution. Teams that attempt to draw prevalence estimates or relative risk measures from purposive samples produce numbers that appear precise but have no defined sampling frame.

| Observation | Likely cause | Discriminating check |
|---|---|---|
| Case fatality appears unusually high | Purposive sample overrepresents clinically severe cases | Compare clinical spectrum in sampled premises against all reported premises |
| New selections add no novel premises or risk factors | Sampling saturation reached | Review last three sampling justifications for new information |
| Sampling criteria change mid-investigation | Convenience drift | Audit selection logs for deviations from stated criteria |
| Investigators report risk ratios from purposive data | Design mismatch | Confirm whether a probability-based follow-up sample was drawn |
| Repeated sampling of same premises | Failure to update sampling frame | Cross-check premises list against new case reports |

## Common Errors and Corrective Actions

Less experienced investigators often confuse purposive sampling with haphazard sampling. Purposive sampling requires predefined criteria applied deliberately. Haphazard sampling selects whatever is convenient. The distinction matters because only purposive sampling produces a defensible rationale for why particular premises were chosen. The corrective action is to write the selection criteria before the first field visit and to require that every sampled unit satisfies at least one stated criterion.

A second common error is the belief that a larger purposive sample automatically improves validity. Sample size in purposive sampling relates to information saturation, not statistical power. Adding more units that resemble the existing sample does not reduce bias. The corrective action is to diversify selection criteria instead of increase the number of units selected under the same criteria.

A third error is the failure to document non-selected premises. Investigators should record which premises were considered but not sampled and why. This information is essential for later transition to probability sampling and for interpreting the generalizability of findings. The [Centers for Disease Control and Prevention principles of epidemiology](https://www.cdc.gov/csels/dsepd/ss1978/index.html) emphasize that outbreak investigations require explicit accounting of the population at risk, which includes premises that were not sampled.

A fourth error is the use of purposive sampling to estimate vaccine efficacy or diagnostic test performance. These parameters require defined denominators and random or systematic selection. Purposive sampling cannot support such estimates, and attempts to do so produce misleading confidence intervals.

## Limitations of the Evidence Base

The published literature on purposive sampling in veterinary outbreak investigations is largely descriptive. Studies that use purposive sampling, such as those documenting ethnoveterinary knowledge, report their sampling approach but rarely evaluate its performance against alternative designs. For example, the [ethnoveterinary survey in the Jimma zone of Ethiopia](https://pubmed.ncbi.nlm.nih.gov/24679045/) used purposive sampling to identify knowledgeable informants, and the [ethnobotanical study in Hawassa Zuria District](https://pubmed.ncbi.nlm.nih.gov/31126296/) applied stratified purposive sampling to select key informants. These studies demonstrate the utility of purposive sampling for knowledge documentation, but they do not provide comparative data on bias or efficiency in outbreak contexts.

Expert opinion differs on the acceptable proportion of a purposive sample within a mixed design. Some authorities argue that purposive sampling should be confined to the hypothesis-generation phase and that any quantitative claim requires probability sampling. Others accept purposive sampling for descriptive outbreak reports when probability sampling is logistically impossible, provided the limitations are stated explicitly. This disagreement reflects the absence of empirical studies comparing outbreak investigation outcomes under different sampling strategies.

The evidence base is also limited by publication bias. Investigations that used purposive sampling and reached clear conclusions are more likely to be published than those that failed or produced ambiguous results. This makes it difficult to assess the true failure rate of purposive sampling in field conditions.

## Referral, Consultation, and Reporting Thresholds

Referral to specialist services becomes appropriate when the investigation exceeds the capacity of the local team. This includes situations where laboratory confirmation is required, where the outbreak spans multiple jurisdictions, or where the causal agent is not identifiable through routine diagnostic panels. The [MSD Veterinary Manual](https://www.msdvetmanual.com/) provides species-specific guidance on diagnostic approaches and differential considerations that can help determine when local resources are insufficient.

Regulatory reporting obligations vary by jurisdiction and by pathogen. The [World Organization for Animal Health terrestrial animal health code](https://www.woah.org/en/what-we-do/standards/codes-and-manuals/terrestrial-code-online-access/) sets international standards for notification of listed diseases. Investigators should consult their national veterinary authority early in the investigation, particularly when the clinical presentation is consistent with a notifiable disease. Delayed reporting can compromise control efforts and may have trade implications.

Consultation with an epidemiologist is warranted when the investigation team lacks formal training in sampling design or when the outbreak has unusual features such as an unknown source, an atypical host range, or an unexplained pattern of spread. The [American Veterinary Medical Association practice resources](https://www.avma.org/resources-tools) include guidance on professional collaboration and referral that applies to epidemiological consultation as well as clinical referral.

The decision to escalate from a local investigation to a regional or national response should be made on the basis of predefined criteria: the number of affected premises, the potential for rapid spread, the zoonotic potential of the agent, and the availability of local resources. Purposive sampling can support the initial characterization of the outbreak, but escalation decisions should be guided by the full picture of reported cases, not by the purposive sample alone.

## Frequently Asked Questions

### How Do I Justify Purposive Sampling When a Supervisor Expects Random Sampling?

Explain that purposive sampling serves a different inferential goal. Probability sampling supports prevalence estimation and statistical generalization to a defined population. Purposive sampling supports analytic generalization, where the objective is to identify cases, describe transmission pathways, or characterize an emerging pathogen. In an outbreak, the priority is often case detection and hypothesis generation, not unbiased prevalence measurement. Reference the [CDC principles of epidemiology](https://www.cdc.gov/csels/dsepd/ss1978/index.html) to frame the distinction between descriptive and analytic objectives. Document the selection criteria, the reasoning behind them, and the specific outbreak question each criterion addresses. This documentation allows a reviewer to assess whether the sampling logic matches the investigation objective.

### What Is the Minimum Number of Animals or Herds I Should Sample Purposively?

There is no universal minimum. The number depends on the outbreak size, the suspected source, the incubation period, and the heterogeneity of the population. For case finding, sample until you stop identifying new epidemiologically relevant information, a saturation point analogous to qualitative research practice. For source tracing, sample every epidemiologically linked unit until the transmission chain is resolved. For hypothesis generation, sample across the full range of clinical presentations, production stages, and management systems. If you are sampling to rule out a specific source, the sample size must be large enough to detect infection at the expected prevalence, which requires moving to probability-based methods. Consult the [WOAH terrestrial animal health code](https://www.woah.org/en/what-we-do/standards/codes-and-manuals/terrestrial-code-online-access/) for species-specific surveillance expectations.

### How Should I Adjust Purposive Sampling When Working with Wildlife or Free-Ranging Populations?

Wildlife investigations require sampling where animals are accessible, which constrains the selection frame. Use purposive sampling to target high-risk groups such as juveniles, scavengers, or animals near index case locations. Combine this with convenience sampling at carcass sites, feeding stations, or water sources. Document the accessibility bias explicitly in the record. For example, an [ethnoveterinary survey in Kenya](https://pubmed.ncbi.nlm.nih.gov/23044218/) used purposive and snowball methods to reach knowledgeable respondents in dispersed pastoral communities, illustrating how purposive approaches adapt to field constraints. In wildlife, the absence of detection does not confirm absence of disease. State this limitation clearly in any report and recommend follow-up probability sampling where feasible.

### What Records Must I Keep When Using Purposive Sampling During an Outbreak?

Record the outbreak definition, the selection criteria, and the rationale for each criterion. Document who selected the sampling units, when, and under what operational constraints. For each sampled animal or premise, record the location, identification, clinical status, sampling date, and test results. Note any animals that were eligible but not sampled and the reason for exclusion. Keep a log of deviations from the planned protocol, including substitutions made because of access, safety, or animal welfare concerns. The [AVMA practice resources](https://www.avma.org/resources-tools) emphasize that contemporaneous records support both clinical defensibility and professional accountability. These records also allow a statistician to assess later whether probability-based follow-up sampling is feasible.

### How Do I Explain Purposive Sampling to a Producer or Farm Manager?

Use concrete language tied to the outbreak goal. State that you are deliberately selecting animals that are most likely to reveal how the disease entered and spread, instead of randomly picking animals that may tell you little. Give an example: if calves are affected but adult cows are not, you will sample calves across multiple pens and also sample the shared colostrum source. Emphasize that this approach speeds up the investigation and reduces the cost of testing. Acknowledge that purposive sampling cannot produce a precise prevalence estimate, and explain that a random sample may follow if that becomes necessary. The [MSD Veterinary Manual](https://www.msdvetmanual.com/) provides species-specific background that can help you frame the explanation in terms the producer already understands.

### What Should I Do When the Ideal Sampling Strategy Exceeds My Available Resources?

Prioritize sampling units that yield the highest diagnostic value per unit cost. Target animals with acute clinical signs, recent onset, or direct contact with the index case. Pool samples within epidemiologically defined groups where laboratory protocols permit. Reduce the number of tests by testing pooled samples first and resolving positive pools individually. If laboratory capacity is the constraint, store properly preserved samples and test them in order of epidemiological priority. If personnel are the constraint, train local staff to collect and label samples under your supervision. Document every resource-driven deviation from the ideal protocol, because these deviations affect interpretation. The [WOAH animal health surveillance standards](https://www.woah.org/en/what-we-do/animal-health-and-welfare/disease-data-collection/) describe how surveillance outputs depend on sampling design and resource allocation.

## Related Clinical & Scientific Guides

* [Evaluating Veterinary Surveillance System Attributes](/knowledge/veterinary-medicine/veterinary-epidemiology/evaluating-veterinary-surveillance-system-attributes)
* [Network Analysis for Infectious Disease Spread in Animal Populations](/knowledge/veterinary-medicine/veterinary-epidemiology/network-analysis-infectious-disease-spread-animal-populations)
* [Randomized Controlled Trials in Veterinary Field Settings](/knowledge/veterinary-medicine/veterinary-epidemiology/randomized-controlled-trials-veterinary-field-settings)


## References and Further Reading

- [Ethnoveterinary medicines in four districts of Jimma zone, Ethiopia: cross sectional survey for plant species and mode of use.](https://pubmed.ncbi.nlm.nih.gov/24679045/). 2014.
- [Ethnobotanical study of medicinal plants in the Hawassa Zuria District, Sidama zone, Southern Ethiopia.](https://pubmed.ncbi.nlm.nih.gov/31126296/). 2019.
- [A survey of plants and plant products traditionally used in livestock health management in Buuri district, Meru County, Kenya.](https://pubmed.ncbi.nlm.nih.gov/23044218/). 2012.
- [Traditional Arabic Palestinian ethnoveterinary practices in animal health care: A field survey in the West Bank (Palestine).](https://pubmed.ncbi.nlm.nih.gov/26869545/). 2016.
- [Medicinal plant knowledge of the Bench ethnic group of Ethiopia: an ethnobotanical investigation.](https://pubmed.ncbi.nlm.nih.gov/19912633/). 2009.
- [Blame and shame in the veterinary profession: barriers and facilitators to reporting significant events.](https://pubmed.ncbi.nlm.nih.gov/30837292/). 2019.
- [WOAH Animal Health Surveillance Standards](https://www.woah.org/en/what-we-do/animal-health-and-welfare/disease-data-collection/). WOAH.
- [CDC Principles of Epidemiology in Public Health Practice](https://www.cdc.gov/csels/dsepd/ss1978/index.html). CDC.
- [MSD Veterinary Manual, Professional Edition](https://www.msdvetmanual.com/). MSD Veterinary Manual.

## Related Articles

- [Cluster Sampling in Veterinary Field Studies](/knowledge/veterinary-medicine/veterinary-epidemiology/cluster-sampling-veterinary-field-studies)
- [Outbreak Investigation in Veterinary Medicine: A Step-by-Step Guide](/knowledge/veterinary-medicine/veterinary-epidemiology/outbreak-investigation-veterinary-medicine-step-by-step-guide)
- [Stratified Sampling for Disease Prevalence Estimation](/knowledge/veterinary-medicine/veterinary-epidemiology/stratified-sampling-disease-prevalence-estimation)
- [Basic Reproductive Ratio (R0) in Veterinary Epidemiology](/knowledge/veterinary-medicine/veterinary-epidemiology/basic-reproductive-ratio-r0-veterinary-epidemiology)
- [Bayesian Hierarchical Models for Veterinary Disease Mapping](/knowledge/veterinary-medicine/veterinary-epidemiology/bayesian-hierarchical-models-veterinary-disease-mapping)

> 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.