# Validating Owner-Reported Outcome Measures in Veterinary Research


## Key Takeaways

- Owner-Reported Outcome Measures (OROMs) are essential for capturing animal health data unavailable during clinical visits, such as subtle lameness or behavioral changes indicative of pain, by formalizing owner observations into quantifiable data.
- Rigorous OROM development necessitates a clearly defined construct (e.g., pain severity, quality of life) derived from literature and expert opinion, followed by item generation from multiple sources including owner interviews to ensure content validity and species-specific relevance.
- Linguistic validation, encompassing forward translation, reconciliation, back translation, and cognitive debriefing with target-language owners, is critical for conceptual equivalence and must precede psychometric evaluation to prevent measurement error.
- Psychometric evaluation establishes an OROM's reliability (internal consistency via Cronbach's alpha, test-retest stability) and validity (construct, convergent, discriminant, criterion), with structural validity often assessed using exploratory and confirmatory factor analysis with specified fit indices.
- Reporting of OROM validation studies must adhere to established guidelines like ARRIVE and EQUATOR, transparently detailing study design, sample characteristics, statistical methods, and the intended population for which the measure is validated.
- Common failure modes for OROMs include construct drift, respondent fatigue, and cultural/linguistic nonequivalence, which can be mitigated through periodic item analysis, monitoring completion patterns, and rigorous cognitive debriefing.

---

Owner-reported outcome measures (OROMs) are structured questionnaires that capture a caregiver's observations of an animal's clinical signs, functional status, and quality of life. In veterinary research, these instruments translate subjective owner perception into quantifiable data suitable for hypothesis testing. This article provides a procedural framework for developing, translating, and psychometrically validating OROMs across species, with emphasis on study design decisions that determine whether a measure can support valid inference. It serves veterinary researchers designing primary studies, clinicians appraising published instruments, and graduate students preparing research protocols. The scope excludes clinician-reported measures such as physical examination scoring systems and laboratory-based endpoints.

## At a Glance

| Parameter | Decision or Standard |
|---|---|
| Construct definition | Specify the latent trait (pain, quality of life, adverse events) before item generation |
| Item generation sources | Literature review, existing instruments, expert opinion, owner interviews |
| Translation method | Forward translation, reconciliation, back translation, cognitive debriefing |
| Cognitive debriefing sample | Target-language owners representing intended demographic range |
| Structural validity | Exploratory then confirmatory factor analysis with named fit indices |
| Reliability | Internal consistency (Cronbach's alpha) and test-retest stability |
| Validity types | Content, construct, convergent, discriminant, criterion where feasible |
| Reporting standard | EQUATOR Network guidelines appropriate to study design |
| Species adaptation | Items and response scales must match species-specific behavioral repertoires |

## The Rationale for Owner-Reported Measures

Animals cannot self-report. Clinical signs that manifest primarily in the home environment, such as lameness on rising, appetite changes, or behavioral indicators of pain, may be absent during a clinic consultation. Owners observe these phenomena repeatedly across contexts and time, giving them access to information that a clinician cannot obtain in a single examination. The [MSD Veterinary Manual professional reference](https://www.msdvetmanual.com/) notes that behavioral assessment in animals depends heavily on history provided by caregivers, and structured instruments formalise this history into a reproducible format.

OROMs also capture outcomes that matter to owners, including quality of life and the burden of care. In oncology, for example, owners' real-time observations of chemotherapy side effects can supplement clinic-based toxicity grading and support treatment decisions between visits. A prototype digital measure developed for dogs undergoing chemotherapy drew items from the Veterinary Comparative Oncology Group-Common Terminology Criteria for Adverse Events, existing client-reported measures, and expert opinion, then tested daily owner assessments over a 21-day period to evaluate feasibility and item performance. This example illustrates the dual function of OROMs: they generate research data and simultaneously support clinical monitoring.

## Defining the Construct

The construct is the theoretical concept the instrument intends to measure. Chronic pain, health-related quality of life, and adverse event burden are all constructs, but each is multidimensional and context-dependent. A researcher must specify the construct's boundaries before writing items. For osteoarthritis pain, the construct may include pain severity, pain interference with daily activities, and gait impairment. For chemotherapy monitoring, the construct may span gastrointestinal signs, lethargy, and owner-perceived quality of life.

Construct definition should draw on existing literature and, where available, established conceptual frameworks. The [EQUATOR Network reporting guideline library](https://www.equator-network.org/) includes resources for reporting instrument development and validation studies, and consulting these before protocol finalisation helps ensure that the construct definition and subsequent validation plan align with field expectations. A poorly defined construct produces items that measure different things, and factor analysis will later expose this heterogeneity as poor model fit.

## Item Generation and Content Validity

Items should arise from multiple sources. A rapid literature review identifies existing measures and their methodological limitations, which informs both item content and design choices. Expert veterinary opinion contributes clinical relevance and ensures that items reflect signs clinicians recognize as meaningful. Owner input, obtained through interviews or focus groups, captures language owners actually use and signs clinicians may underestimate.

Content validity is the degree to which items represent all facets of the construct. It is established during development, not after data collection. The research team should document how each item maps to the construct definition and should justify exclusion of candidate items. For cross-species application, item content must be adapted to species-specific behavioral repertoires. A pain item that references "limping" may suit dogs but not cats, and a quality-of-life item about "playfulness" requires species-appropriate operational definitions.

## Translation and Linguistic Validation

When an instrument developed in one language is used in another, translation alone is insufficient. Linguistic validation ensures conceptual equivalence, also literal correspondence. The process recommended by the World Health Organization and the International Society for Pharmacoeconomics and Outcomes Research involves forward translation by two independent native speakers of the target language, reconciliation of discrepancies by a third linguist, back translation by an independent translator fluent in both languages, and comparison of the original and back-translated versions to resolve residual differences.

A Spanish translation of the Liverpool Osteoarthritis in Dogs followed this sequence and added cognitive debriefing with target-language owners to assess comprehension and identify problematic items. Similarly, a Spanish version of the Helsinki Chronic Pain Index was produced through the same structured process, with cognitive debriefing conducted in 62 dog owners from diverse demographics to evaluate clarity. These examples demonstrate that linguistic validation is a distinct phase preceding psychometric testing. A translated instrument that has not undergone cognitive debriefing may contain items that are grammatically correct but conceptually opaque to owners, producing measurement error that no statistical adjustment can fully correct.

## Psychometric Evaluation

Psychometric validation establishes that the instrument measures what it claims to measure with acceptable precision. The Korean version of the Canine Brief Pain Inventory provides a model of initial psychometric evaluation. In a cross-sectional study of 56 dogs with osteoarthritis and 22 healthy controls, the investigators assessed structural validity through exploratory and confirmatory factor analyzes, convergent validity through correlations with an owner-reported quality-of-life item, discriminant validity through patient-control group comparisons, and internal consistency through Cronbach's alpha. The exploratory analysis identified a dominant general pain factor, while confirmatory analysis showed superior fit for a two-factor model using named fit indices including the comparative fit index and root mean square error of approximation.

Reliability and validity are properties of the instrument in a specific population and context, not fixed attributes. An instrument validated in dogs with osteoarthritis may perform differently in dogs with cancer pain, and a version validated in one language requires fresh psychometric evaluation in another. The [ARRIVE guidelines for reporting animal research](https://arriveguidelines.org/) specify that outcome measures and their validation status should be reported transparently, allowing readers to judge whether the instrument supports the study's conclusions.

## Selecting the Measurement Instrument

The choice of an owner-reported outcome measure (OROM) begins with a specification of the intended use. A measure developed for clinical trials of analgesic drugs may perform poorly in a longitudinal study of disease progression, and a measure designed for tertiary referral oncology may be unsuitable for first-opinion practice. The researcher must define the target population, the expected range of severity, the recall period, and the mode of administration before comparing candidate instruments.

Existing instruments should be identified through a systematic search of the veterinary literature and human patient-reported outcome registries. The [EQUATOR Network reporting guidelines library](https://www.equator-network.org/) provides a structured approach to locating validated measures and their psychometric properties. For each candidate instrument, the researcher should extract information on the construct measured, the number of items, the response format, the recall period, the target species and condition, and the evidence for reliability and validity in populations similar to the intended study sample.

Instrument selection criteria should include content coverage, feasibility, and prior validation evidence. A measure with strong psychometric properties in one language or culture cannot be assumed to perform identically in another. The Spanish translation of the Liverpool Osteoarthritis in Dogs followed formal translation and linguistic validation procedures precisely because cross-cultural equivalence requires empirical testing instead of assumption [translation and linguistic validation of the LOAD into Spanish](https://pubmed.ncbi.nlm.nih.gov/38444778/). Similarly, the Korean version of the Canine Brief Pain Inventory required de novo psychometric evaluation, including factor analysis and internal consistency testing, before it could be recommended for use in Korean-speaking populations [initial psychometric evaluation of the Korean CBPI](https://pubmed.ncbi.nlm.nih.gov/42551991/).

Where no suitable instrument exists, the researcher must either adapt an existing measure or develop a new one. Adaptation is generally preferable to de novo development because it preserves prior validation evidence, but adaptation requires the same translation and validation steps as original development. De novo development is justified only when the construct is genuinely novel or when existing measures lack essential content for the target population.

## Study Design Considerations

The validation study design must be specified before data collection begins. Cross-sectional designs are appropriate for evaluating internal consistency, construct validity, and discriminant validity. Longitudinal designs are required for evaluating test-retest reliability, responsiveness, and minimally important change. The choice of design affects sample size, statistical power, and the interpretability of results.

Sample size estimation for psychometric studies depends on the planned analyzes. Factor analysis generally requires at least 5 to 10 respondents per item, with a minimum of 100 to 200 respondents for stable solutions. Studies with smaller samples, such as the 56 dogs with osteoarthritis and 22 healthy controls in the Korean CBPI evaluation, can support initial psychometric evaluation but should be interpreted with caution [initial psychometric evaluation of the Korean CBPI](https://pubmed.ncbi.nlm.nih.gov/42551991/). Confirmatory factor analysis with small samples produces unstable parameter estimates and inflated standard errors.

The reference standard against which the new measure is compared must be specified in advance. For pain measures, the reference standard may be a clinician-assessed pain score, a functional outcome such as gait analysis, or an established OROM. For quality-of-life measures, the reference standard may be a global health rating or a condition-specific measure. Convergent validity is supported when the new measure correlates moderately to strongly with measures of the same or related constructs. Discriminant validity is supported when the new measure does not correlate with measures of unrelated constructs.

## Administering the Measure

The mode of administration affects data quality and must be standardized across study sites. Paper questionnaires, telephone interviews, and digital platforms each have distinct advantages and limitations. Digital administration enables real-time data capture, automated scoring, and reduced data entry errors, but requires owners to have access to compatible devices and reliable internet connections. The prototype digital measure for monitoring chemotherapy side effects in dogs was designed specifically to facilitate owner participation in home monitoring, illustrating the potential of digital administration in longitudinal studies [development and testing of a digital owner-reported outcome measure in canine oncology](https://pubmed.ncbi.nlm.nih.gov/41202824/).

Owner training and instructions must be standardized. Written instructions should accompany every questionnaire, and study personnel should be prepared to answer owner questions without influencing responses. The recall period must be clearly stated and consistently applied. Owners should complete the measure independently, without consultation with other household members or veterinary staff, unless the study design specifically requires proxy or consensus reporting.

Missing data are inevitable in longitudinal studies. The researcher must specify a priori how missing items will be handled, whether by imputation, exclusion, or sensitivity analysis. The proportion of missing data should be reported for each time point, and patterns of missingness should be examined to determine whether data are missing at random or in association with owner or patient characteriztics.

## Monitoring Data Quality

Data quality monitoring should occur throughout the study, also at the analysis stage. Response patterns that suggest careless or inconsistent reporting, such as identical responses across all items or implausible combinations of responses, should be flagged for review. The frequency of such patterns should be reported, and sensitivity analyzes should examine whether excluding these responses changes the study conclusions.

Floor and ceiling effects should be examined at each time point. A measure that produces scores clustered at the maximum or minimum of the scale has limited ability to detect change in the affected direction. Floor effects are common in pain measures administered to populations with mild disease, while ceiling effects are common in quality-of-life measures administered to populations with severe disease. The presence of floor or ceiling effects in more than 15% of respondents indicates a measurement limitation that should be reported.

## Reporting the Validation Study

Reporting standards for OROM validation studies follow the general principles of the [ARRIVE guidelines for reporting animal research](https://arriveguidelines.org/), which specify the minimum information required for transparent and reproducible publications. The report should describe the study design, the sample size and recruitment strategy, the inclusion and exclusion criteria, the demographic and clinical characteriztics of the study population, and the statistical methods used for each psychometric property evaluated.

The report should include the full instrument, the scoring instructions, and the interpretation guidance. This enables replication and independent evaluation by other research groups. The report should also state the intended population and the conditions under which the measure is valid, and should explicitly identify populations for which validation evidence is lacking.

## Practical Decision Framework

| Decision Point | Consideration | Recommended Action |
|---|---|---|
| Instrument selection | Construct match, prior validation, feasibility | Use existing validated measure where available, adapt only with full validation |
| Language and culture | Target population language, cultural relevance | Follow formal translation and linguistic validation protocols |
| Study design | Cross-sectional vs longitudinal, sample size | Match design to psychometric properties being evaluated |
| Administration mode | Paper, telephone, digital | Standardize across sites, verify owner access and competence |
| Data quality | Missing data, response patterns, floor and ceiling effects | Monitor throughout, report and analyze as specified a priori |
| Reporting | Transparency, reproducibility | Follow ARRIVE and EQUATOR reporting standards |

The correct choice at each decision point depends on the study context. A validation study conducted in a single referral hospital with a small sample may be appropriate for initial psychometric evaluation but cannot support claims of generalizability to primary care populations. A measure validated in dogs may require separate validation in cats, and a measure validated in one production species cannot be assumed to transfer to another. The researcher must state these limitations explicitly and identify the populations for which further validation is required.

## Recognized Complications and Failure Modes

Owner-reported outcome measures fail in predictable ways. The most common failure is construct drift, where the instrument gradually measures something other than the intended clinical state. This occurs when items are ambiguous, when owners interpret response scales differently across visits, or when the measure is applied to a population outside its validated range. Early detection requires periodic item-level analysis. Track response distributions for each item across time. Ceiling or floor effects exceeding 30% of responses at either extreme suggest the item lacks discriminative capacity for the study population. Compare item-total correlations at interim analysis, values below 0.3 indicate items that may be measuring a different construct.

A second failure mode is respondent fatigue, which produces patterned responding. Owners begin selecting the same response category for every item, often the middle option or the first option presented. Detection relies on examining response patterns instead of total scores. Calculate the proportion of identical consecutive responses and the time to completion if the measure is digital. The prototype digital measure developed for dogs receiving chemotherapy used daily assessments over 21 days, and the investigators specifically examined completion patterns to identify burden-related attrition. If completion rates decline after the first week or if response variance collapses, shorten the instrument or extend the recall period.

A third failure mode is cultural or linguistic nonequivalence. Translation alone does not preserve construct validity. The Spanish translation of the Liverpool Osteoarthritis in Dogs followed structured guidelines including reconciliation, back translation, and cognitive debriefing precisely because direct translation introduces subtle shifts in meaning. Similarly, the Spanish version of the Helsinki Chronic Pain Index required cognitive assessment with 62 owners from diverse demographics to confirm clarity before psychometric testing could proceed. Detection requires cognitive interviewing with a sample of the target population, also bilingual review.

| Observation | Likely cause | Discriminating check |
|---|---|---|
| Response variance collapses after first week | Respondent fatigue or burden | Compare item-level variance week 1 versus week 3, review completion times |
| Total scores do not correlate with clinical measures | Construct drift or poor convergent validity | Re-examine item-total correlations, verify the clinical anchor is appropriate |
| Owners select the same response for all items | Patterned responding | Calculate longest run of identical responses, inspect for straight-lining |
| Scores differ between language versions | Translation nonequivalence | Conduct cognitive debriefing with target-language owners, compare factor structures |
| Ceiling or floor effects exceed 30% | Poor item discrimination | Review item response distributions, consider item removal or scale expansion |

## Common Errors and Corrective Action

Less experienced researchers often skip the cognitive debriefing step, treating translation and back translation as sufficient. This is a substantive error. Cognitive debriefing identifies items that owners misinterpret despite linguistically accurate translation. The Korean version of the Canine Brief Pain Inventory underwent exploratory and confirmatory factor analysis precisely because translation accuracy does not guarantee structural equivalence. Corrective action: always include cognitive interviews with at least 10 to 15 owners representing the target demographic before proceeding to psychometric evaluation.

A second common error is validating a measure in a narrow sample and then generalizing. The Korean CBPI study included 56 dogs with osteoarthritis and 22 healthy controls, an appropriate initial sample but insufficient to establish measurement invariance across breeds, ages, or disease severities. Corrective action: state the intended population explicitly and recruit a sample that reflects its diversity. If the measure will be used in multiple languages, plan for cross-cultural validation instead of assuming equivalence.

A third error is conflating internal consistency with validity. Cronbach's alpha values above 0.9 are sometimes interpreted as evidence of a superior instrument, but very high alpha can indicate item redundancy. Corrective action: examine the inter-item correlation matrix. Mean inter-item correlations between 0.2 and 0.4 suggest a coherent scale without redundancy. Values above 0.7 warrant item reduction.

## Limitations of Current Evidence

The evidence base for owner-reported outcome measures in veterinary medicine remains uneven. Most published instruments were developed in English and validated in North American or Western European populations. Cross-cultural validation is sparse. The Spanish translations of the LOAD and HCPI represent linguistic validation only, psychometric validation has not yet been completed for either. This distinction matters. A linguistically validated instrument is not necessarily psychometrically valid in the target population.

Expert opinion differs on the optimal number of items and the appropriate recall period. Some argue for brief instruments with 24-hour recall to minimize recall bias, while others favour longer instruments with weekly recall to capture fluctuating conditions. The evidence does not currently resolve this debate. The digital chemotherapy monitoring prototype used daily assessments for clinical signs and weekly surveys for quality of life, reflecting a pragmatic compromise. Researchers should justify their choice based on the condition's temporal profile and the owner's reporting burden.

Reporting standards for veterinary instrument validation studies are less developed than for clinical trials. The ARRIVE guidelines specify minimum information for animal research reporting, and the EQUATOR Network provides a library of reporting guidelines including CONSORT and STROBE, but no veterinary-specific guideline exists for instrument validation studies. Researchers should follow the closest applicable human guideline and state this explicitly.

## Referral, Consultation, and Regulatory Considerations

Most validation work does not require regulatory approval, but the context of use can change this. If the instrument will support a licensing claim, a pharmacovigilance program, or a regulatory submission, consult the relevant authority early. The World Organization for Animal Health terrestrial standards address welfare assessment and surveillance, and an owner-reported measure used in those contexts should align with applicable standards. Regional requirements differ. The American Veterinary Medical Association provides practice resources that address professional standards for client communication and informed consent, which apply when owners complete measures as part of clinical care. Consult the MSD Veterinary Manual for species-specific considerations when the measure targets a condition with recognized clinical assessment standards.

Statisticians with psychometric training should be involved before data collection, not after. Confirmatory factor analysis, measurement invariance testing, and item response theory require specialised expertise that general veterinary biostatisticians may not provide. If the instrument will be used across multiple sites or languages, involve a psychometrician during the design phase. For measures intended to support regulatory submissions, consult the relevant regulatory body before finalising the validation protocol.

## Frequently Asked Questions

### How Much Does It Cost to Validate an Owner-Reported Outcome Measure, and Can I Do It on a Limited Budget?

Costs scale with the number of owners enrolled, the number of measurement occasions, and whether paid translators or statisticians are involved. A full psychometric evaluation with factor analysis and test-retest reliability typically requires 100 or more completed responses per language version. On a limited budget, prioritize content validity through expert review and cognitive debriefing with a small owner sample, then report internal consistency using Cronbach's alpha. Convergent validity against a single related measure requires fewer animals than a full factor analytic approach. The [EQUATOR Network reporting guidelines](https://www.equator-network.org/) can help you identify which validation components are essential for your study design and which can be deferred to later work.

### What Should I Do When the Only Available Measure Is in Another Language?

Follow the translation and linguistic validation pathway described in the [Spanish validation of the Liverpool Osteoarthritis in Dogs](https://pubmed.ncbi.nlm.nih.gov/38444778/). Two independent forward translations, reconciliation, back translation, and cognitive debriefing with owners from the target population are the minimum steps. The [Helsinki Chronic Pain Index translation](https://pubmed.ncbi.nlm.nih.gov/41012737/) demonstrates that even well-established measures require this process before use in a new language. Psychometric validation must follow linguistic validation, a translated measure that has not been tested for reliability and validity in the new language should be treated as an experimental instrument.

### How Do I Validate an Owner-Reported Measure in a Species Other Than Dogs?

The psychometric principles transfer across species, but the item content must reflect species-specific behaviors and owner observations. For cats, items about grooming, litter box use, and hiding behavior replace canine items about gait and leash activity. For horses, lameness observation and behavior during handling require different item wording. Content validity therefore requires species-specific expert panels and cognitive debriefing with owners of that species. The statistical methods for construct validity and internal consistency are identical. Cross-species generalization of an existing measure is not defensible without repeating the full validation process, as shown in the [Korean Canine Brief Pain Inventory psychometric evaluation](https://pubmed.ncbi.nlm.nih.gov/42551991/).

### What Records Must I Keep During Validation and Subsequent Use?

Maintain a version-controlled master copy of the questionnaire, a log of every item change with rationale and date, and documentation of owner consent and data handling procedures. Record response rates, missing item patterns, and any owner comments about confusing wording. For clinical trials using the measure, the [ARRIVE guidelines](https://arriveguidelines.org/) specify the minimum information needed for transparent reporting, including how the measure was administered, who completed it, and how missing data were handled. Keep raw data in a format that allows re-analysis if the psychometric properties are questioned during peer review.

### How Do I Explain Validation Requirements to a Supervisor or Funding Body?

Frame validation as a risk management issue. An unvalidated measure can produce misleading treatment effects, wasted sample size, and rejected manuscripts. Present the two-stage pathway: linguistic validation where needed, then psychometric evaluation. The [AVMA practice resources](https://www.avma.org/resources-tools) provide context on professional standards for evidence-based practice that can support your argument. Emphasize that validation is a one-time investment per language and species, and that the resulting instrument can be reused across multiple studies. Offer a phased plan that delivers preliminary data quickly while the full validation proceeds.

### When Is It Acceptable to Use a Measure That Has Not Been Fully Validated?

Use an unvalidated measure only for hypothesis generation or pilot work, and state this limitation explicitly in any presentation or publication. If you must proceed with an unvalidated measure in a clinical study, include a concurrent validation substudy with a recognized comparator where feasible. The [MSD Veterinary Manual](https://www.msdvetmanual.com/) notes that clinical assessment tools require documented reliability before their results can guide treatment decisions. For regulatory submissions or multicentre trials, fully validated measures are expected. When no validated measure exists for your target condition and species, developing and validating a new measure is the appropriate response instead of adapting an unsuitable existing tool.

## Related Clinical & Scientific Guides

* [Conducting Systematic Reviews of Veterinary Diagnostic Test Accuracy](/knowledge/veterinary-medicine/veterinary-research-methods/conducting-systematic-reviews-veterinary-diagnostic-test-accuracy)
* [Bias in Veterinary Research: Types, Sources, and Mitigation](/knowledge/veterinary-medicine/veterinary-research-methods/bias-veterinary-research-types-sources-mitigation)
* [Cluster Randomized Trials in Veterinary Research: Design and Analysis](/knowledge/veterinary-medicine/veterinary-research-methods/cluster-randomized-trials-veterinary-research-design-analysis)


## References and Further Reading

- [Translation and linguistic validation into Spanish of the Owner-Reported Outcome Measure "Liverpool Osteoarthritis in Dogs".](https://pubmed.ncbi.nlm.nih.gov/38444778/). 2024.
- [Development and Testing of an Owner-Reported Outcome Measure of Clinical Signs and Quality of Life in Dogs Treated With Chemotherapy.](https://pubmed.ncbi.nlm.nih.gov/41202824/). 2026.
- [Translation and Linguistic Validation into Spanish of the Owner-Reported Outcome Measure "Helsinki Chronic Pain Index" (HCPI).](https://pubmed.ncbi.nlm.nih.gov/41012737/). 2025.
- [Initial psychometric evaluation of the Korean version of the Canine Brief Pain Inventory for canine osteoarthritis.](https://pubmed.ncbi.nlm.nih.gov/42551991/). 2026.
- [ARRIVE Guidelines 2.0 for Reporting Animal Research](https://arriveguidelines.org/). PLOS Biology, 2020.
- [EQUATOR Network Reporting Guidelines](https://www.equator-network.org/). EQUATOR Network.
- [MSD Veterinary Manual, Professional Edition](https://www.msdvetmanual.com/). MSD Veterinary Manual.
- [American Veterinary Medical Association Practice Resources](https://www.avma.org/resources-tools). American Veterinary Medical Association.
- [WOAH Terrestrial Animal Health Code](https://www.woah.org/en/what-we-do/standards/codes-and-manuals/terrestrial-code-online-access/). WOAH.

## Related Articles

- [Outcome Measures in Veterinary Clinical Trials: Selection and Validation](/knowledge/veterinary-medicine/veterinary-research-methods/outcome-measures-veterinary-clinical-trials-selection-validation)
- [Selecting Outcome Measures for Veterinary Pain Studies](/knowledge/veterinary-medicine/veterinary-research-methods/selecting-outcome-measures-veterinary-pain-studies)
- [Using Mixed Methods in Veterinary Research](/knowledge/veterinary-medicine/veterinary-research-methods/using-mixed-methods-veterinary-research)
- [Conducting Qualitative Research in Veterinary Settings](/knowledge/veterinary-medicine/veterinary-research-methods/conducting-qualitative-research-veterinary-settings)
- [How to Write a Research Protocol for Veterinary Studies](/knowledge/veterinary-medicine/veterinary-research-methods/write-research-protocol-veterinary-studies)

> 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.