Measuring Disease Frequency: Incidence and Prevalence in Animal Populations

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

Key Takeaways

• Prevalence quantifies the proportion of animals with a condition at a specific time (point prevalence) or over an interval (period prevalence), answering "how much disease exists now?" and is crucial for resource allocation and assessing chronic disease burden. For instance, a brucellosis survey in dairy herds reported animal-level prevalence ranging from 34.9% to 51.4%, indicating a significant existing disease load.
• Incidence measures the rate of new disease occurrences, with cumulative incidence (risk) representing the proportion of at-risk animals developing disease over a period, and incidence rate using animal-time as the denominator for dynamic populations. For example, yearly herd-level incidence risk for Porcine Epidemic Diarrhea Virus (PEDV) in Ontario swine herds fell from 13.5% in 2014 to 1.4% in 2016, reflecting declining new infections.
• The relationship between prevalence, incidence, and disease duration is fundamental: prevalence is approximately incidence multiplied by mean duration of disease in a steady state, explaining how chronic conditions with low incidence but long duration (e.g., brucellosis) can accumulate high prevalence. Conversely, acute, rapidly fatal diseases may have high incidence but low prevalence.
• Accurate frequency measurement necessitates explicit definitions for numerators (new cases for incidence, existing cases for prevalence) and denominators (population at risk for incidence, total population for prevalence), alongside precise case definitions and specified time periods for incidence. For example, incidence calculations require excluding already affected animals from the at-risk denominator.
• Diagnostic test performance (sensitivity and specificity) critically impacts frequency estimates; apparent prevalence must be corrected to real prevalence to account for false positives and negatives, as demonstrated by the milk ELISA for bovine brucellosis where apparent prevalence required adjustment using known test characteristics.
• Population dynamics (births, deaths, movement) and surveillance intensity can introduce significant errors, necessitating careful consideration of denominator changes and potential selection bias in voluntary surveillance programs. For instance, a herd expanding through purchases may show rising incidence due to the introduction of new susceptible animals.

This article provides the veterinary researcher with the quantitative foundation for measuring disease frequency in animal populations. It defines incidence and prevalence, distinguishes their appropriate uses, and explains the calculations that underpin population-level health assessment. The content serves clinicians designing surveillance programs, epidemiologists interpreting published literature, and graduate students preparing for research involving disease burden estimation.

Accurate measurement of disease frequency is the precondition for every subsequent epidemiological step, including risk factor analysis, outbreak investigation, and evaluation of control interventions. Confusion between incidence and prevalence produces misinterpretable results and unsound management decisions. The distinction matters clinically: prevalence describes the disease burden present in a population at a given time, while incidence describes the rate at which new cases arise. A chronic endemic disease may show high prevalence with low incidence, whereas an acute outbreak may show low prevalence with high incidence. The WOAH animal health surveillance standards require member countries to report disease frequency using defined measures, and the choice of measure determines whether the reported figures are comparable across regions and over time.

At a Glance

Defining the Measures

Prevalence

Prevalence is the proportion of animals in a defined population that have the condition of interest at a specified moment or period. Point prevalence counts affected animals at a single instant, such as the proportion of dogs with cataract among those examined during a defined study window. Period prevalence counts animals affected at any time during an interval, which captures cases that resolve or die before the interval ends. The cross-sectional study of canine cataract prevalence illustrates the point prevalence approach: 2000 dogs were examined ophthalmoscopically, and the age at which 50% of dogs showed cataract was derived from the fitted prevalence curve. The study reported a mean C50 of 9.4 years across all dogs, with all dogs over 13.5 years affected by some degree of lens opacity. Prevalence answers the question "how much disease exists here now?" It does not answer "how fast is the disease spreading?"

Incidence

Incidence measures the occurrence of new cases. Cumulative incidence, also called incidence risk, is the proportion of at-risk animals that develop the condition during a specified period. It requires a defined cohort, a defined observation interval, and explicit handling of losses to follow-up. Incidence rate, also called incidence density, uses animal-time as the denominator, counting each animal for the time it actually contributes to the at-risk population. The herd-level surveillance of porcine epidemic diarrhea virus in Ontario swine herds demonstrates the distinction in practice: yearly herd-level incidence risk for PEDV was 13.5% in 2014, falling to 3.0% in 2015 and 1.4% in 2016, while herd-level prevalence in the last week of each year was 4.4%, 2.3%, and 1.4% respectively. The gap between incidence risk and prevalence reflects the duration of herd-level infection and the timing of surveillance relative to outbreak waves.

The Relationship Between Incidence and Prevalence

Prevalence is a function of incidence and disease duration. For a condition in steady state, prevalence approximately equals incidence multiplied by mean duration of disease. This relationship explains why chronic conditions accumulate in populations: a disease with low incidence but long duration produces substantial prevalence. Conversely, a rapidly fatal disease may have high incidence but low point prevalence because affected animals die quickly. The brucellosis survey in dairy herds in Hubei Province illustrates the practical consequence: animal-level prevalence ranged from 34.9% to 51.4% across 15 herds, yet the 3-month incidence risk in the 10 farms meeting inclusion criteria was only 0.4%. The high prevalence with low incidence indicates chronic infection with prolonged infectious periods, a pattern typical of brucellosis in endemic herds.

Choosing the Correct Measure

When Prevalence Is Appropriate

Prevalence suits cross-sectional surveys, resource allocation decisions, and assessment of chronic disease burden. It is the measure of choice when the question concerns how many animals need treatment or how much disease exists in a population at one time. Prevalence is also the practical measure when incidence is difficult to ascertain because disease onset is insidious or case detection depends on a single survey. The milk ELISA evaluation for bovine brucellosis used prevalence as the primary survey output because the cross-sectional sampling design could not follow animals over time.

When Incidence Is Required

Incidence is required for studying disease aetiology, evaluating prevention measures, and understanding transmission dynamics. Incidence rate is the preferred measure when the at-risk population changes over time, when follow-up duration varies among animals, or when the researcher needs to compare disease occurrence between populations with different age structures. The longitudinal study of Neospora caninum in Swiss dogs used monthly coproscopic sampling over one year to calculate yearly incidence of Hammondia/Neospora-like oocysts at 9.2%, a figure that prevalence alone could not provide because it captured the dynamics of shedding over time.

Numerators and Denominators

The numerator for prevalence is the count of affected animals at the survey time. The numerator for incidence is the count of new cases arising during the observation period. Both require an explicit case definition. The denominator for prevalence is the total population examined, including unaffected animals. The denominator for incidence is the population at risk, which excludes animals that cannot develop the disease, such as vaccinated animals in a vaccine efficacy study or animals already affected at the start of follow-up.

The systematic review of incretin-based therapies and pancreatitis risk illustrates the importance of denominator specification in observational research. The review pooled six studies comparing pancreatitis incidence between patients receiving incretins and non-users, reporting an odds ratio of 1.08 with a 95% confidence interval of 0.84 to 1.40. The authors noted that a risk increase lower than 35% could not be excluded based on the power calculation. This example, drawn from human medicine but methodologically identical to veterinary pharmacovigilance, demonstrates how incidence comparisons require explicit definition of exposure groups, at-risk periods, and outcome ascertainment.

Time and Its Complications

Incidence is meaningless without a time specification. Cumulative incidence requires a fixed observation period, and the reported value applies only to that period. Incidence rate uses animal-time, which accommodates varying follow-up durations but requires decisions about how to count time after an animal develops the disease, dies, or is lost. Standard practice removes affected animals from the denominator after disease onset because they are no longer at risk. The CDC principles of epidemiology in public health practice provide the conventional framework for these calculations, including the distinction between fixed cohorts and dynamic populations.

Sources of Error in Frequency Measurement

Diagnostic Test Error

Prevalence and incidence estimates inherit the sensitivity and specificity of the diagnostic test used. The milk ELISA study for bovine brucellosis reported field sensitivity of 87.2% and specificity of 92.0% based on Bayesian latent class analysis, with a Cohen's kappa of 0.747 against serum ELISA. Apparent prevalence must be corrected for test error to obtain real prevalence, and the correction requires reliable estimates of test performance in the target population. Test performance varies with disease stage, sample quality, and laboratory conditions, so published sensitivity and specificity values may not transfer across settings.

Population Dynamics

Births, deaths, culling, and movement alter both numerators and denominators. A herd that expands through purchases may show rising incidence simply because new susceptible animals enter the population. A herd that culls test-positive animals will show falling prevalence that reflects management intervention instead of reduced transmission. The Ontario swine coronavirus surveillance relied on voluntary participation in an industry-led disease control program, which introduces selection bias because participating herds may differ systematically from non-participants. The authors reported incidence risk with 95% confidence intervals, acknowledging the uncertainty inherent in surveillance data.

Surveillance Intensity

More intensive surveillance detects more cases. Passive surveillance systems underreport disease frequency, and the degree of underreporting varies with disease awareness, diagnostic capacity, and reporting incentives. The WOAH terrestrial animal health code specifies reporting requirements for listed diseases, but the completeness of national reporting depends on the strength of each country's surveillance infrastructure. Researchers comparing disease frequency across regions must assess whether differences reflect true epidemiological variation or differential surveillance intensity.

Worked Example: Estimating Incidence Risk in a Defined Herd

Consider a 500 sow farrow-to-finish herd that experiences an outbreak of porcine epidemic diarrhea. The herd manager wants to know how quickly the virus spread after introduction. The surveillance window runs from 1 March to 31 May. At the start of the period, all 500 sows are susceptible. During March, 40 sows develop clinical disease. In April, a further 25 sows are affected. In May, 10 more sows become clinical. No sow is affected twice, and none are removed from the population during the study period.

The incidence risk for the full three month period is calculated as the number of new cases divided by the number of animals at risk at the start. That is 75 divided by 500, or 15%. The 95% confidence interval, calculated using the Wilson method, is approximately 12.1% to 18.4%. This interval width reflects the uncertainty inherent in a single herd observation. Reporting the interval matters because a herd of this size cannot yield a precise estimate of the true risk.

The same data can be expressed as an incidence rate. Each sow contributes one sow-month at risk for each month she remains unaffected. The total sow-months at risk is the sum of monthly contributions: 500 for March, 460 for April (the 40 affected sows no longer contribute), and 435 for May. The total is 1395 sow-months. The incidence rate is 75 divided by 1395, or 0.054 cases per sow-month. This converts to 54 cases per 1000 sow-months, or approximately 0.65 cases per sow-year. The rate is lower than the risk because it accounts for the shrinking population of susceptible animals over time.

The choice between risk and rate here depends on the question. If the manager asks what proportion of the herd was affected during the outbreak, the risk of 15% is the correct answer. If the question concerns the speed of transmission, the rate is more informative because it standardizes for the time each sow spent at risk. For a rapidly spreading pathogen in a closed population, the rate will decline as susceptible animals are depleted, even if transmission efficiency remains constant.

Worked Example: Prevalence in a Cross-Sectional Survey

A practice in central China is asked to estimate the burden of bovine brucellosis in dairy herds. The diagnostic laboratory has validated a milk ELISA against serum ELISA, with a Cohen's kappa of 0.747 and field sensitivity and specificity of 87.2% and 92.0% respectively, as reported in a study of dairy herds in Hubei Province evaluation of a milk ELISA for brucellosis prevalence and incidence in dairy herds. The practice samples 3091 cows from 15 herds. The apparent prevalence, the proportion of test-positive animals, is calculated directly from the test results. The real prevalence, the proportion truly infected, must be estimated by adjusting for test error.

The adjustment uses the formula: real prevalence equals apparent prevalence plus specificity minus one, divided by sensitivity plus specificity minus one. With a sensitivity of 87.2% and a specificity of 92.0%, the denominator is 0.792. If the apparent prevalence is 40%, the real prevalence is (0.40 plus 0.92 minus 1) divided by 0.792, which equals 0.404, or 40.4%. The adjustment is small at this prevalence level because the test errors partially cancel. At low apparent prevalence, the correction becomes larger and can produce negative estimates when the apparent prevalence falls below the false positive rate. A negative adjusted estimate indicates that the test is not performing adequately in that population, and the survey should be interpreted with caution.

The study from which these test parameters are drawn reported animal level real prevalence ranging from 34.9% to 51.4% across 15 herds, with 93.3% of herds containing at least one test-positive animal evaluation of a milk ELISA for brucellosis prevalence and incidence in dairy herds. The incidence risk in 10 herds meeting inclusion criteria was 0.4% per three months. The contrast between high prevalence and low incidence indicates chronic infection with slow transmission, a pattern typical of brucellosis in endemic herds.

Comparing Incidence and Prevalence in Practice

The table below summarizes the decision framework for selecting between incidence and prevalence measures in common veterinary scenarios.

The choice of measure changes the interpretation of the same underlying data. A disease with high incidence and rapid recovery will show low prevalence at any single time point. A disease with low incidence and lifelong persistence will accumulate high prevalence over years. The relationship between the two measures is governed by the average duration of disease, and this relationship can be used to estimate duration when both measures are known.

Species and Production System Modifications

The correct measure and the method of calculation vary with the production system. In companion animal practice, the population at risk is often poorly defined. A practice catchment area does not equal the population of dogs or cats that might attend. Prevalence estimates from hospital populations are subject to selection bias, as demonstrated in a study of canine cataract where the examined dogs were predominantly from veterinary hospital populations, rehoming charities, and breeding kennels prevalence of canine cataract in a cross-sectional study. The study reported that the age at which 50% of dogs had cataract was 9.4 years, but this figure applies to the sampled population, not to all dogs in the region.

In production animal medicine, the population at risk is usually enumerable. Herd records provide the denominator, and individual animal identification enables accurate follow-up. The Ontario swine surveillance program provides a model for this approach, calculating yearly herd-level incidence risk for porcine epidemic diarrhea virus as 13.5%, 3.0%, and 1.4% for 2014, 2015, and 2016 respectively herd-level prevalence and incidence of PEDV and PDCoV in Ontario swine herds. These estimates required a voluntary disease control program with diagnostic and epidemiological data linked to individual herds.

Wildlife and free-ranging populations present the greatest challenge. The population at risk is rarely known with precision, and follow-up of individuals is difficult. Capture-recapture methods and repeated cross-sectional surveys are often the only feasible approaches. Incidence measures require longitudinal follow-up and are frequently replaced by prevalence measures with the caveat that duration of infection cannot be estimated. In such populations, the reporting standard should state explicitly which measure was used and what assumptions were made about population stability.

Documenting Frequency Measures

Every reported frequency measure should include the numerator, the denominator, the time period, and the diagnostic criteria. A statement that prevalence was 10% is insufficient. The report should specify whether this is apparent or real prevalence, which test was used, and the population from which the sample was drawn. The World Organization for Animal Health surveillance standards require that disease reporting includes case definitions and population descriptions to enable international comparison WOAH animal health surveillance standards.

Confidence intervals should accompany all estimates. The Wilson method is preferred for proportions because it performs well at extreme values. For incidence rates, the exact Poisson interval is appropriate when the number of cases is small. When reporting to regulatory authorities, the format prescribed by the relevant code should be followed. The WOAH Terrestrial Animal Health Code provides the international framework for notification of listed diseases WOAH terrestrial animal health code. Regional requirements may differ, and the reporting veterinarian should confirm the applicable standard before submission.

The diagnostic test used to identify cases must be validated for the species and sample type. A test validated in serum may perform differently on milk, as demonstrated by the kappa agreement of 0.747 between milk and serum ELISA for brucellosis evaluation of a milk ELISA for brucellosis prevalence and incidence in dairy herds. The report should state the test's sensitivity and specificity in the population studied, and the adjustment method used to convert apparent to real prevalence. Where test performance is unknown, the report should acknowledge this limitation instead of presenting unadjusted estimates as definitive.

Recognized Complications and Failure Modes

Incidence and prevalence measures fail in predictable ways. The most common failure is denominator drift, where the population at risk changes during the observation period through births, deaths, sales, or movement. A herd that loses 20% of its animals to an outbreak will show a lower incidence risk than the true risk among animals that remained, because the denominator shrinks as cases accumulate. Detect this early by recording population size at each visit or interval, also at the start. The CDC principles of epidemiology in public health practice recommend using person-time denominators when the population is unstable, because this approach accounts for varying periods of contribution.

Left truncation is a second failure mode. Animals that enter the study already infected are excluded from incidence numerators but remain in the denominator unless tested on entry. This biases incidence downward. The corrective action is baseline testing of all entrants with a test of known sensitivity and specificity. Where baseline testing is impossible, restrict the analysis to a closed cohort and state the restriction explicitly.

A third failure is the conflation of period prevalence with incidence. A survey that detects all current cases at two time points and labels the difference as incidence misses cases that occurred and resolved between surveys. This error is common in cross-sectional designs repeated at long intervals. The discriminating check is whether the interval between surveys exceeds the typical duration of the condition. For acute, self-limiting diseases such as porcine epidemic diarrhea, the interval must be short, often weekly, as demonstrated in the surveillance of PEDV and PDCoV in Ontario swine herds, where weekly counts were used to track incidence during an active outbreak.

Common Errors and Corrective Action

Less experienced analysts frequently misclassify the unit of analysis. Herd-level and animal-level measures answer different questions. A herd-level incidence risk of 13.5% for PEDV in Ontario in 2014 means 13.5% of herds experienced at least one case, not that 13.5% of animals were infected. The Ontario swine coronavirus study reports both levels separately, and the distinction matters for control programs that operate at the herd level.

A second recurring error is the use of prevalent cases in the numerator of an incidence calculation. This happens when a surveillance database does not distinguish new from existing diagnoses. The corrective action is to require a disease-free interval or a negative test before study entry, and to verify that each case has a recorded onset date. The milk ELISA study for bovine brucellosis in Hubei Province illustrates the consequence: the authors reported animal-level prevalence ranging from 34.9% to 51.4% across herds, but the incidence risk over three months was only 0.4%. Using the prevalence figure as an incidence estimate would have overstated transmission by two orders of magnitude.

A third error is ignoring diagnostic test performance when interpreting frequency measures. A test with 87.2% sensitivity and 92.0% specificity, as reported for the milk ELISA in the brucellosis study, will produce false positives in low-prevalence populations. The apparent prevalence must be corrected using the test's operating characteriztics, or the reported measure must be labelled as apparent prevalence. The brucellosis milk ELISA evaluation used Bayesian latent class analysis to estimate true prevalence, an approach that should be considered whenever a reference standard is unavailable.

Limitations of the Evidence and Areas of Disagreement

The veterinary literature on disease frequency is uneven. Many published estimates come from convenience samples, referral populations, or voluntary surveillance programs, and these sources introduce selection bias that is difficult to quantify. The review of limb loss epidemiology noted that studies varied in scope, quality, and methodology, making comparisons between studies difficult. The same criticism applies to veterinary frequency measures. Voluntary disease control programs, such as the Ontario swine coronavirus program, depend on producer participation, and non-participating herds may differ systematically from participating herds.

Expert opinion differs on the preferred denominator for incidence in production animal populations. Some authorities advocate animal-time at risk, arguing that it is the only measure that handles variable follow-up correctly. Others prefer incidence risk for its interpretability and its direct relevance to herd-level decision making. The WOAH terrestrial animal health code provides reporting standards but does not mandate a single measure, leaving the choice to the investigating epidemiologist.

A further area of uncertainty is the handling of subclinical infection. For diseases such as brucellosis, where many infections are asymptomatic, incidence estimates depend heavily on the sensitivity of the surveillance test and the frequency of testing. The WOAH animal health surveillance standards emphasize that surveillance systems must be designed with explicit case definitions and testing protocols, but the choice of testing interval remains a matter of judgment informed by the biology of the agent and the resources available.

Referral, Consultation, and Regulatory Reporting

Referral to a veterinary epidemiologist or a diagnostic laboratory is warranted when the frequency measure will inform a control program, a trade decision, or a regulatory submission. Situations that require specialist input include estimating incidence in a population with complex movement patterns, correcting prevalence for imperfect test performance, and designing surveillance for a newly introduced pathogen. The MSD Veterinary Manual provides species-specific guidance on disease presentation, but the design of frequency studies falls outside its scope and belongs to the epidemiologist.

Regulatory reporting obligations vary by jurisdiction and by disease. The WOAH terrestrial animal health code defines the list of notifiable diseases and the reporting timelines for member countries. Veterinarians who detect a notifiable disease must report it through the designated national authority, and the frequency measures they submit must be calculated according to the case definitions in the code. The WOAH animal health surveillance standards describe the expected content of surveillance reports, including the population at risk, the case definition, and the time period covered.

Laboratory involvement is required when the case definition depends on diagnostic testing. The brucellosis milk ELISA study demonstrates that the choice of specimen and test affects both prevalence and incidence estimates. A laboratory should be consulted before the study begins to confirm that the test is fit for purpose, that sample handling is appropriate, and that the test's sensitivity and specificity are known for the target population. This consultation is particularly important for diseases with legal or trade consequences, where the cost of a misclassified case is high.

Frequently Asked Questions

How Do I Calculate Incidence Risk When the Population at Risk Changes During the Observation Period?

When animals enter or leave the population, the denominator is unstable. Use incidence rate instead of incidence risk, with animal-time as the denominator. Each animal contributes time at risk until the event, death, or end of the study. For example, in the Swiss dog study, 249 dogs were sampled monthly for one year, and the yearly incidence of Hammondia or Neospora-like oocysts was calculated at 9.2%, with farm dogs showing higher incidence than urban dogs. If animals are lost to follow-up, censor them at the last known negative test. The CDC principles of epidemiology provide worked examples of person-time denominators that transfer directly to animal populations.

What Is the Minimum Herd Size Needed to Produce a Meaningful Prevalence Estimate?

There is no universal minimum, but the precision of the estimate depends on the number of animals sampled and the expected prevalence. For a herd of 50 animals, a sample of 10 gives wide confidence intervals. For a herd of 500, the same sample gives narrower intervals. Use an online sample size calculator or standard formulae with the expected prevalence and desired confidence level. In the Ontario swine study, herd-level prevalence of PEDV was reported with 95% confidence intervals, and the authors noted that small herd numbers produced wide intervals in later years. When resources are limited, pool samples or use a targeted sampling strategy, but report the sampling method and its limitations explicitly.

How Do I Report Prevalence When My Diagnostic Test Has Imperfect Sensitivity and Specificity?

Report apparent prevalence, which is the proportion of test-positive animals, and then correct to true prevalence using the test's sensitivity and specificity. The milk ELISA study for bovine brucellosis used Bayesian latent class analysis to estimate field sensitivity of 87.2% and specificity of 92.0%, then calculated real prevalence ranging from 34.9% to 51.4% across herds. Apparent prevalence alone can mislead, especially in low-prevalence populations where false positives dominate. State which measure you report and the test characteriztics used. The MSD Veterinary Manual provides species-specific guidance on interpreting diagnostic test results in population contexts.

How Do I Explain the Difference Between Incidence and Prevalence to a Producer or Practice Owner?

Use a concrete analogy. Prevalence is a snapshot, the number of animals with disease on a given day. Incidence is a movie, the number of new cases over a period. For a dairy herd with brucellosis, prevalence tells you how many cows are infected now, while incidence risk tells you how quickly new infections appear, for example 0.4% per three months in the Hubei Province study. This distinction matters for control decisions. A high prevalence with low incidence suggests an old, stable infection. A low prevalence with high incidence signals an active outbreak. The AVMA practice resources offer communication frameworks that can be adapted for producer discussions.

What Records Do I Need to Maintain to Support Future Incidence Calculations?

Maintain a census of the population with entry and exit dates for every animal, including births, purchases, deaths, and culls. Record each disease event with a date and a diagnostic basis. Keep laboratory submissions linked to individual animal identifiers. Document changes in testing protocols, because changes in sensitivity affect comparability over time. For herd-level incidence, record the date the herd first becomes positive and the date it is declared negative. The Ontario PEDV surveillance program relied on a voluntary disease control database that captured diagnostic and epidemiological information, which allowed calculation of weekly incidence counts and yearly herd-level incidence risk. The WOAH animal health surveillance standards specify minimum data elements for notifiable diseases.

How Do I Handle Frequency Measures in Wildlife or Free-Ranging Populations Where I Cannot Individually Identify Animals?

Individual-based incidence is usually impossible. Use prevalence from repeated cross-sectional surveys and report it as such. For incidence, use proxy measures such as seroconversion detected through repeated sampling of marked individuals, or use capture-recapture methods to estimate population size and then estimate the number of new cases. Be explicit about the assumptions. The canine cataract study examined 2000 dogs from multiple sources and reported age-specific prevalence, not incidence, because individual follow-up was not feasible. For wildlife, the WOAH terrestrial animal health code provides guidance on surveillance design when individual identification is impractical.

Related Clinical & Scientific Guides

• Evaluating Veterinary Surveillance System Attributes
• Network Analysis for Infectious Disease Spread in Animal Populations
• Randomized Controlled Trials in Veterinary Field Settings

References and Further Reading

• Epidemiology of limb loss and congenital limb deficiency: a review of the literature.. 2003.
• Herd-level prevalence and incidence of porcine epidemic diarrhea virus (PEDV) and porcine deltacoronavirus (PDCoV) in swine herds in Ontario, Canada.. 2018.
• Evaluation of a milk ELISA as an alternative to a serum ELISA in the determination of the prevalence and incidence of brucellosis in dairy herds in Hubei Province, China.. 2020.
• Incidence of Neospora caninum and other intestinal protozoan parasites in populations of Swiss dogs.. 2006.
• Incretin-based therapies and acute pancreatitis risk: a systematic review and meta-analysis of observational studies.. 2015.
• Prevalence of canine cataract: preliminary results of a cross-sectional study.. 2004.
• WOAH Animal Health Surveillance Standards. WOAH.
• CDC Principles of Epidemiology in Public Health Practice. CDC.
• MSD Veterinary Manual, Professional Edition. MSD Veterinary Manual.

Related Articles

• Risk Factor Analysis for Disease in Animal Populations
• Stratified Sampling for Disease Prevalence Estimation
• Network Analysis for Infectious Disease Spread in Animal Populations
• Using Capture-Recapture Methods to Estimate Animal Disease Prevalence
• Compartmental Models in Veterinary Disease Dynamics

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.