Equivalence and Non-Inferiority Trials in Veterinary Medicine

By Dr. Zubair Khalid, DVM, MS, PhD · 2026-08-11

Key Takeaways

• Equivalence and non-inferiority trials are designed to demonstrate that a new veterinary treatment is not unacceptably worse than an established one, addressing practical questions about generic drugs, reformulated vaccines, or alternative administration routes.
• The critical element is the pre-specified non-inferiority margin (delta), representing the largest clinically unimportant difference, which must be justified by historical comparator efficacy against placebo and clinical judgment, not statistical convenience.
• Statistical analysis relies on confidence intervals; non-inferiority is concluded if the lower bound of the confidence interval for the treatment difference lies above the pre-specified margin, while equivalence requires both bounds to be within the defined symmetric margin.
• Common errors include selecting the margin post-hoc, confusing it with the expected treatment difference, and misinterpreting a confidence interval crossing the margin as proof of equivalence rather than an indication of uncertainty or underpowering.
• Assay sensitivity is paramount; a non-inferiority trial is only interpretable if the active comparator has demonstrated efficacy against placebo in the same setting, otherwise, both treatments could be ineffective.
• Reporting standards, such as CONSORT extensions and ARRIVE guidelines, are crucial for transparency, requiring justification of the margin, definition of analysis populations (intention-to-treat and per-protocol), and reporting of confidence intervals.

Veterinary researchers increasingly need to show that a new treatment is not worse than an established one, instead of proving it is better. Equivalence and non-inferiority trials answer a different question from conventional superiority trials: they test whether two interventions have effects that are acceptably close. This article explains the design logic, statistical foundations, and interpretation of these trial formats for veterinary researchers working across species. It covers when each design is appropriate, how to set margins, how to analyze results, and how to avoid the common errors that invalidate conclusions.

The clinical question that motivates these designs is practical. A generic drug, a reformulated vaccine, a new route of administration, or a less expensive alternative must demonstrate that it preserves the clinically important benefits of the reference product. The reader is assumed to be familiar with randomised controlled trial structure, hypothesis testing, and confidence intervals, but not necessarily with the specific logic of non-inferiority testing.

At a Glance

The Logical Distinction from Superiority Trials

A superiority trial tests whether one treatment is better than another. Its null hypothesis states that the treatments are equal, and the trial seeks evidence to reject that hypothesis. A non-inferiority trial reverses this logic. The null hypothesis states that the experimental treatment is worse than the active comparator by more than a pre-specified margin, and the trial seeks evidence to reject that hypothesis. An equivalence trial extends this to both directions, testing whether the experimental treatment is neither meaningfully worse nor meaningfully better than the comparator.

The practical consequence is that a non-inferiority trial can conclude that two treatments are effectively the same even when the confidence interval includes zero difference. This is impossible in a superiority trial, where a non-significant p-value only means that the data do not exclude equality. The distinction matters for regulatory submissions and for clinical decision-making. A veterinarian choosing between two antimicrobials for bovine respiratory disease needs to know whether the cheaper or more convenient option preserves the essential efficacy of the reference product, not whether it is superior Randomized, non-inferiority trial comparing a nitric oxide releasing solution.

The Non-Inferiority Margin

The margin, often denoted delta, is the largest difference between treatments that the investigator is willing to accept as clinically unimportant. It must be specified before the trial begins and justified on clinical grounds. The margin is not a statistical quantity. It is a clinical judgment about how much efficacy can be sacrificed in exchange for other advantages such as lower cost, fewer adverse effects, or easier administration.

The margin should be smaller than the effect size that the active comparator has demonstrated against placebo in historical trials. If the comparator's own superiority over placebo is only 5 percentage points, a non-inferiority margin of 8 percentage points would allow the experimental treatment to be worse than placebo. The margin must preserve a meaningful fraction of the comparator's established benefit. For the bovine respiratory disease trial cited above, the margin was set at 8.5 percentage points for the difference in disease prevalence between a nitric oxide releasing solution and tilmicosin Randomized, non-inferiority trial comparing a nitric oxide releasing solution. The observed difference was 17 percentage points with a lower confidence bound of 11 percentage points, which exceeded the margin and led to a conclusion of inferiority.

For continuous outcomes, the margin is expressed in the units of the outcome. A non-inferiority trial of milk fat globule membrane enriched infant formulas used a margin of 3.0 g per day for mean weight gain, a threshold chosen to represent a clinically acceptable difference in growth Safety and tolerance evaluation of milk fat globule membrane-enriched. A periodontal trial used a margin of 1 mm for clinical attachment level gain, with a stricter 0.5 mm margin for clinical relevance Efficacy of inorganic bovine bone combined with leukocyte and. The choice of margin requires input from clinicians who understand the natural history of the disease and the magnitude of change that matters to outcomes.

Statistical Analysis and Confidence Intervals

The analysis of non-inferiority trials rests on confidence intervals, not p-values. The decision rule is straightforward: non-inferiority is concluded if the lower bound of the confidence interval for the treatment difference lies above the pre-specified margin. For equivalence, both the lower and upper bounds must lie within the two-sided margin. The confidence level is conventionally 95% for the two-sided interval, which corresponds to a one-sided alpha of 0.025 for the non-inferiority comparison.

The feline infectious peritonitis trial comparing oral remdesivir with GS-441524 illustrates the reporting standard. The difference in survival was 22% with a 90% confidence interval from -13.5% to 57.5%, and the authors concluded that remdesivir met the criteria for non-inferiority Efficacy of Oral Remdesivir Compared to GS-441524 for Treatment. The choice of a 90% interval in that trial reflects the one-sided nature of the comparison. The trial was small, with only 18 cats enrolled, and the wide confidence interval shows the uncertainty that small samples produce.

A common analytical error is to conclude non-inferiority from a non-significant superiority test. This is invalid because a wide confidence interval that includes both zero and the margin is compatible with either conclusion. The confidence interval must be examined directly against the margin. Another error is to use a per-protocol analysis exclusively. The intention-to-treat population tends to dilute differences between treatments and can bias toward non-inferiority, while the per-protocol population can bias in either direction. Most authorities recommend reporting both analyzes and requiring consistent conclusions.

Reporting Standards and Transparency

Veterinary trials should follow the reporting standards catalogued by the EQUATOR Network, which includes the CONSORT statement extension for non-inferiority and equivalence trials EQUATOR Network Reporting Guidelines. The ARRIVE guidelines provide the minimum information required for transparent reporting of animal research, including randomisation procedures, blinding, sample size calculations, and outcome definitions ARRIVE Guidelines 2.0 for Reporting Animal Research. These standards require that the margin be justified, that the analysis populations be defined, and that confidence intervals be reported instead of p-values alone.

The reporting standards exist because non-inferiority trials are particularly vulnerable to bias. Poor compliance, dropout, and measurement error all tend to make treatments look more similar than they are, which is exactly the direction that favours a non-inferiority conclusion. A trial with sloppy conduct can produce a spurious finding of equivalence. The reader of a veterinary non-inferiority trial should therefore scrutinise the methods for blinding, allocation concealment, and completeness of follow-up before accepting the conclusion.

Selecting the Comparator and Control Arm

The choice of comparator determines what a non-inferiority trial can claim. An active comparator establishes that the test intervention preserves a clinically acceptable fraction of the standard treatment's effect. A placebo or no-treatment control cannot support non-inferiority claims, because the assay sensitivity of the trial is unknown. If the active comparator would not have separated from placebo in the same setting, then a finding of "no important difference" between test and active comparator is uninterpretable.

For veterinary products, the active comparator should be the current first-line intervention for the target indication in the relevant species and production system. This may differ by region. A product that is standard in one country may be unavailable, differently labelled, or subject to different withdrawal periods elsewhere. The comparator should also be administered at its registered dose and by its registered route. Deviations from the label compromise the claim that the test product preserves the effect of real-world best practice.

In some settings the comparator is not a drug but a procedure or management strategy. The periodontal trial comparing inorganic bovine bone with leukocyte and platelet-rich fibrin against the same graft with a collagen membrane used an active surgical comparator and set the non-inferiority margin on clinical attachment level gain clinical attachment level and radiographic outcomes in a periodontal non-inferiority trial. The choice of comparator here reflects a clinical question about which adjunct is acceptable, not whether grafting itself works.

Blinding reduces differential co-intervention and outcome ascertainment bias. Blinding is feasible when the test and comparator have identical presentation, as in the blinded comparison of oral remdesivir and GS-441524 in cats with effusive feline infectious peritonitis blinded non-inferiority comparison of oral antivirals for feline infectious peritonitis. When blinding is impossible, the protocol should specify who remains masked, how outcome assessment is protected, and what steps prevent contamination of the comparison.

Setting the Margin in Practice

The non-inferiority margin has two components. The first is the smallest effect the active comparator can be expected to preserve relative to placebo, estimated from historical trials. The second is the largest loss of effect the field will tolerate, a clinical judgment. The margin cannot exceed the historical effect of the comparator, because a margin larger than the comparator's own effect would permit a test product with zero efficacy to be declared non-inferior.

Regulatory authorities and specialty groups publish margin guidance for some indications. Where no published margin exists, the investigator must justify the chosen value in the protocol with reference to historical data and clinical consequence. A common approach is to set the margin at a fraction of the historical comparator effect, often one-half, but this convention is not universal and should be defended explicitly.

The margin must be fixed before unblinding or interim analysis. Post hoc widening of the margin to secure a non-inferiority conclusion is a form of p-hacking and invalidates the inference. The margin should also be clinically meaningful in the units of the primary outcome. In the bovine respiratory disease trial, the margin was set at 8.5 percentage points for the difference in BRD prevalence over 40 days, a value the authors judged to represent an acceptable loss of disease control relative to tilmicosin nitric oxide releasing solution versus macrolide for bovine respiratory disease control.

Margins differ by outcome type. For continuous outcomes such as daily weight gain, the margin is expressed in the measurement unit, as in the infant formula trial where the margin was set at -3.0 g/day for mean weight gain safety and tolerance of milk fat globule membrane enriched infant formulas. For binary outcomes such as survival or seroconversion, the margin is expressed as a risk difference or risk ratio. For time-to-event outcomes, the margin is usually a hazard ratio bound.

Interpretation When the Confidence Interval Crosses the Margin

The decision rule is mechanical. Compute the two-sided confidence interval for the difference between treatments at the pre-specified level, usually 90% for a one-sided alpha of 0.05. If the entire interval lies on the favourable side of the margin, non-inferiority is concluded. If the interval crosses the margin, non-inferiority is not established, and the result is indeterminate. This is not the same as demonstrating inferiority. Inferiority is demonstrated only when the entire interval lies beyond the margin on the unfavourable side.

The bovine respiratory disease trial illustrates the distinction. The observed difference in BRD prevalence was 17 percentage points in favour of tilmicosin, with a 95% confidence interval of 11 to 23 percentage points. Because the lower bound exceeded the 8.5 percentage point margin, the authors concluded inferiority of the nitric oxide releasing solution nitric oxide releasing solution versus macrolide for bovine respiratory disease control. This is a stronger conclusion than failure to establish non-inferiority, and it changed the practical recommendation: the product was not acceptable for high-risk cattle.

A confidence interval that crosses the margin can arise from a truly inferior product, from a trial that is underpowered, or from poor assay sensitivity. The interpretation depends on the width of the interval. A narrow interval that just touches the margin suggests a product near the boundary of acceptability. A wide interval that spans the margin and the null suggests the trial was too small to answer the question. The report should present the confidence interval, also a p-value, and should discuss which of these explanations is most plausible.

Species and Production System Effects on Margin Choice

The same clinical endpoint may require different margins in different species or production contexts. Growth rate in food animals is economically meaningful, and a margin that preserves a specified fraction of expected gain may be acceptable in one production system but not another. In companion animals, owner-reported outcomes such as pain scores or quality of life may have margins anchored to the minimum detectable difference on the instrument, which can vary by instrument and by rater.

Disease prevalence and baseline risk modify the practical consequence of a given margin. In the bovine respiratory disease trial, the margin of 8.5 percentage points was chosen for high-risk cattle, the population in which BRD control programs are targeted nitric oxide releasing solution versus macrolide for bovine respiratory disease control. A margin that is acceptable in a low-risk population may be unacceptable in a high-risk population because the absolute number of affected animals is larger.

Regulatory expectations differ by jurisdiction and by product class. Generic drug approval pathways in many regions rely on equivalence or non-inferiority frameworks with margins specified in guidance documents. Vaccine licensure may require non-inferiority of serological response, as in the canine rabies vaccine trial that compared cold-chain stored vaccine with vaccine stored in a passive cooling device canine rabies vaccine storage outside the cold chain. Investigators should consult the relevant national authority and the World Organization for Animal Health terrestrial standards where international movement of animals or products is involved WOAH terrestrial animal health standards.

Documentation and Reporting Checklist

The following table summarizes the items that should appear in a non-inferiority trial report. The checklist follows the spirit of the CONSORT extension for non-inferiority trials, available through the EQUATOR Network library of reporting guidelines EQUATOR Network reporting guidelines.

The intention-to-treat and per-protocol analyzes often disagree in non-inferiority trials. Intention-to-treat tends to dilute differences between groups and can bias toward non-inferiority. Per-protocol analysis excludes major protocol violators and can be more conservative. Both should be reported, and the conclusions should be consistent across them. If they diverge, the report should explain which analysis is primary and why.

Reporting standards for animal research also apply. The ARRIVE guidelines specify the minimum information needed for transparent and reproducible animal studies, including randomisation, blinding, sample size, and outcome definitions ARRIVE guidelines for reporting animal research. Veterinary-specific extensions of CONSORT, such as REFLECT for livestock trials, are indexed in the EQUATOR library and should be consulted where applicable EQUATOR Network reporting guidelines.

Recognized Complications and Failure Modes

Non-inferiority trials fail in characteriztic ways that are detectable before, during, or after data collection. The most damaging failure is a margin that was chosen after the results were known. This practice, sometimes called margin drift, converts a confirmatory design into a descriptive exercise. The margin must be fixed in the protocol before enrollment begins, and any post hoc revision should be reported as a protocol deviation.

A second failure mode is poor assay sensitivity. A non-inferiority trial can only demonstrate equivalence if the active comparator would have beaten placebo in the same setting. If the comparator is ineffective, underdosed, or administered incorrectly, the trial may conclude non-inferiority when both treatments are worthless. This risk is highest when the comparator has never been tested against placebo in the target species or production system. Reviewers should ask whether historical placebo-controlled data for the comparator exist and whether those data were generated under comparable conditions.

Differential withdrawal between arms distorts non-inferiority conclusions. If animals receiving the test treatment are removed because of poor response, the remaining animals may look artificially similar to the comparator group. The analysis set matters more here than in superiority trials. The per-protocol population is often the primary analysis in non-inferiority trials because it excludes protocol violators, but it can introduce bias if withdrawals are treatment-related. The intention-to-treat population, which includes all randomised animals, tends to bias toward non-inferiority and is therefore a useful sensitivity analysis. Both should be reported and should agree.

A third failure mode is a confidence interval that is wide enough to cross the margin but narrow enough to exclude clinically meaningful superiority. This outcome is sometimes misread as proof of equivalence. It is not. The correct interpretation is that the trial was underpowered or the between-animal variability was higher than anticipated. The confidence interval width, not the point estimate, determines whether the question has been answered.

Common Errors in Design and Interpretation

Less experienced investigators frequently confuse the margin with the expected treatment difference. The margin is the largest difference that would still be clinically acceptable, not the difference the investigator expects to observe. Setting the margin equal to the anticipated effect size produces a trial that cannot reject inferiority even when the test treatment is meaningfully worse.

A related error is selecting a margin smaller than the variability of the outcome can support. A margin of 1 mm for periodontal attachment gain may be clinically sensible, but if the standard deviation of the measurement is 2 mm, the required sample size becomes prohibitive. The margin must be reconciled with the measurement error of the outcome instrument. Where the outcome is subjective, such as lameness scoring or lesion grading, the margin should exceed the expected inter-observer variability.

Students and clinicians also misread the confidence interval. A result in which the entire confidence interval lies on the favourable side of zero does not demonstrate superiority unless the interval also excludes the margin. Non-inferiority and superiority are separate claims. A trial designed for non-inferiority can sometimes demonstrate superiority as a secondary finding, but the protocol should state in advance whether superiority will be tested and how multiplicity will be handled.

Another common error is applying a one-sided test without justification. Non-inferiority is inherently one-sided in its question, but the analysis is usually reported with a two-sided confidence interval. The convention is a 95% two-sided interval, equivalent to a 2.5% one-sided test. Using a 90% interval is acceptable if stated in advance, but it changes the type I error rate and should not be chosen after seeing the data.

Limitations of the Current Evidence

The veterinary non-inferiority literature is sparse relative to human medicine, and the available studies cluster in a few therapeutic areas. The bovine respiratory disease trial comparing a nitric oxide releasing solution with tilmicosin illustrates both the utility and the fragility of the design. The margin was set at 8.5% for the difference in disease prevalence, and the observed difference of 17% with a confidence interval of 11% to 23% led to a clear conclusion of inferiority. That clarity is the exception. Many veterinary non-inferiority studies report confidence intervals that cross the margin, leaving the question unresolved.

The feline infectious peritonitis trial comparing oral remdesivir with GS-441524 enrolled only 18 cats. The confidence interval for the survival difference spanned from -13.5% to 57.5%, a range so wide that neither non-inferiority nor inferiority can be claimed with confidence. Small sample sizes are a recurring limitation in companion animal trials, where recruitment is slow and the cost per animal is high. The periodontal study in humans, while methodologically rigorous, cannot be extrapolated to veterinary patients without species-specific validation.

Expert opinion still differs on whether the margin should be derived primarily from historical placebo data or from clinical judgment. The historical approach is more objective but depends on the availability of adequate placebo-controlled trials in the target species. The clinical judgment approach is more flexible but invites criticism that the margin was chosen to make the test treatment look acceptable. Reporting standards such as the ARRIVE guidelines and the EQUATOR Network resources require that the margin be justified, but they do not prescribe which justification is correct. The REFLECT statement, available through the EQUATOR Network, provides livestock-specific reporting guidance that addresses some of these gaps.

When to Refer, Consult, or Report

Statisticians should be consulted before the margin is fixed, not after the data are analyzed. A veterinary epidemiologist or biostatistician with experience in non-inferiority designs can estimate the required sample size, advise on the analysis population, and flag assay sensitivity problems before resources are committed. For regulatory submissions, the relevant authority should be consulted early because margin requirements differ between jurisdictions and between product categories. The World Organization for Animal Health terrestrial standards address vaccine efficacy and quality expectations that may affect comparator selection and acceptable margins.

Laboratory involvement is warranted when the outcome is measured with error. Serological assays, such as the fluorescent antibody virus neutralisation test used in the rabies vaccine storage study, require standardized laboratory conditions and defined thresholds for seroconversion. The choice of threshold, for example 0.5 IU/mL, must be justified and applied consistently across arms. Where the outcome is a continuous laboratory measurement, the laboratory should be blinded to treatment allocation and should use validated quality control procedures.

Regulatory reporting is required when a non-inferiority trial reveals that a licensed product performs below its labelled claim. This situation arose in the bovine respiratory disease trial, where the test treatment was concluded to be inferior to the macrolide comparator. If the test treatment had been a licensed product, the sponsor would be obliged to report the finding to the relevant authority. Adverse events that occur during a non-inferiority trial, such as the higher eczema rate observed in the milk fat globule membrane formula study, must be reported according to the pharmacovigilance requirements of the jurisdiction. The MSD Veterinary Manual and the American Veterinary Medical Association practice resources provide guidance on adverse event reporting obligations, but the specific requirements vary by country and product type.

Frequently Asked Questions

How do I justify a non-inferiority margin to a funding body or ethics committee?

Justify the margin with clinical reasoning, not statistical convenience. State the smallest difference in the primary outcome that would change your treatment recommendation, then defend that threshold with published evidence or consensus opinion. Reference the comparator's historical effect size and explain why preserving a specific fraction of that effect matters for your target population. For regulatory submissions, consult the relevant authority's guidance documents. The WOAH terrestrial animal health standards provide context for vaccine and biologic evaluations, while the EQUATOR Network reporting guidelines list the reporting standards reviewers will expect. Document your margin selection in the protocol before enrollment begins.

What can I do when the sample size for a properly powered non-inferiority trial exceeds my budget?

Consider a crossover design if the condition is chronic and stable, because within-animal comparisons reduce variance and required sample size. Alternatively, narrow the primary endpoint to a more sensitive continuous measure instead of a binary outcome. You can also revisit the margin, but only if clinical justification supports a wider threshold. Do not simply reduce sample size and retain the original margin, because an underpowered non-inferiority trial cannot distinguish true equivalence from imprecision. If resources remain insufficient, publish the work as a pilot study with explicit confidence intervals and no claim of non-inferiority. The ARRIVE guidelines require transparent reporting of sample size calculations, so document any compromise and its rationale.

How does non-inferiority testing differ for laboratory animal studies compared with clinical veterinary trials?

Laboratory animal studies typically allow tighter environmental control, uniform genetics, and scheduled euthanasia for tissue collection, which reduces variability and permits smaller margins. Clinical trials in client-owned animals face heterogeneous populations, variable owner compliance, and ethical constraints on invasive monitoring. The margin must reflect the clinical setting where the intervention will actually be used. For laboratory studies, a margin based on biological relevance may be appropriate, whereas clinical work demands a margin anchored to patient-relevant outcomes. Species-specific physiology also matters, for example metabolic rate affects drug clearance and therefore the window for detecting differences. The MSD Veterinary Manual provides species-specific pharmacological context that informs realistic margins across taxa.

What records should I keep specifically for a non-inferiority trial?

Maintain the protocol with the pre-specified margin and analysis plan, all versions of the statistical analysis code, and the complete case report forms. Document every protocol deviation, including missed doses, withdrawal dates, and reasons for exclusion. Record the randomisation sequence generation method and allocation concealment mechanism. Keep logs of assay calibration, equipment maintenance, and personnel training that could affect outcome measurement. Store raw data in a format that allows independent re-analysis, also summary statistics. The EQUATOR Network reporting guidelines specify the minimum information readers need to assess trial validity, and your records should support every item on that checklist. Retain records for at least the period required by your institution or sponsor.

How do I explain non-inferiority results to a client whose animal was in the trial?

Explain that the trial asked whether the new treatment is no worse than the standard by a pre-defined, clinically acceptable amount. Use an analogy, for example two routes to the same destination where one may be slightly slower but still arrives within an acceptable time. State the outcome in absolute terms, such as the percentage of animals that recovered in each group, instead of quoting confidence intervals. Be honest about uncertainty, particularly if the confidence interval crossed the margin. Emphasize that individual responses vary and that the trial result describes the group average. The AVMA practice resources offer guidance on communicating clinical evidence to owners. Avoid implying that non-inferiority means the treatments are identical.

When should I choose an equivalence trial instead of a non-inferiority trial?

Choose equivalence when you must rule out both clinically meaningful superiority and inferiority of the test treatment. This arises for generic drug approval, where the product must match the reference within a symmetric margin, and for formulations intended to be interchangeable, such as different routes of administration for the same drug. Non-inferiority is appropriate when the test treatment offers an advantage, such as lower cost, fewer doses, or reduced withdrawal period, that justifies accepting some loss of efficacy. The distinction matters for sample size, because equivalence trials require roughly double the sample size of non-inferiority trials to achieve the same precision. Consult the WOAH terrestrial animal health standards for regulatory contexts where equivalence testing is mandated.

Related Clinical & Scientific Guides

• Conducting Systematic Reviews of Veterinary Diagnostic Test Accuracy
• Bias in Veterinary Research: Types, Sources, and Mitigation
• Cluster Randomized Trials in Veterinary Research: Design and Analysis

References and Further Reading

• Safety and tolerance evaluation of milk fat globule membrane-enriched infant formulas: a randomized controlled multicenter non-inferiority trial in healthy term infants.. 2014.
• Efficacy of inorganic bovine bone combined with leukocyte and platelet-rich fibrin or collagen membranes for treating unfavorable periodontal infrabony defects: Randomized non-inferiority trial.. 2021.
• Randomized, non-inferiority trial comparing a nitric oxide releasing solution with a macrolide antibiotic for control of bovine respiratory disease in beef feedlot calves at high-risk of developing respiratory tract disease.. 2016.
• Efficacy of Oral Remdesivir Compared to GS-441524 for Treatment of Cats with Naturally Occurring Effusive Feline Infectious Peritonitis: A Blinded, Non-Inferiority Study.. 2023.
• Randomized sibling-oocyte study using recombinant human hyaluronidase versus bovine-derived Sigma hyaluronidase in ICSI patients.. 2008.
• Investigating the Efficacy of a Canine Rabies Vaccine Following Storage Outside of the Cold-Chain in a Passive Cooling Device.. 2021.
• ARRIVE Guidelines 2.0 for Reporting Animal Research. PLOS Biology, 2020.
• EQUATOR Network Reporting Guidelines. EQUATOR Network.
• MSD Veterinary Manual, Professional Edition. MSD Veterinary Manual.

Related Articles

• Designing Equivalence Trials for Veterinary Generic Drugs
• Designing Adaptive Clinical Trials for Veterinary Medicine
• Critical Appraisal of Randomized Controlled Trials in Veterinary Medicine
• Designing Crossover Trials for Veterinary Therapeutics
• Evaluating Prognostic Factor Studies in Veterinary Medicine

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.