Veterinary Epidemiology: An Introduction

• The presentation covers an introduction to epidemiology within the context of a Master's Degree in Veterinary Medicine at Egas Moniz School of Health & Science.
• The date of the presentation is March 15th, 2025.
• The course unit (CU) for this presentation is Epidemiology.
• The presenter is Professor Dr. Mauro Braganca, who holds MRCVS, DVM, MSc, and MBA credentials, and is an Epid Specialist.

Program & Evaluations

• Causality
• Frequency Measures
• Field Epidemiology
• Building Tools Survey
• Association Measures/Epidemiological Studies
• Preparing Research Proposal
• Sampling
• Diagnostic Tests/Epidemiological Surveillance
• Research Proposal Presentation
• Integrative Test
• Practical assessments are part of the Building Tools Survey, Association Measures/Epidemiological Studies and Diagnostic Tests/Epidemiological Surveillance

Sampling Techniques

• Includes understanding representativeness and inference.
• Sample size calculations are important in determining prevalence and differences between groups.
• The central limit theorem is important.
• Inclusion and exclusion criteria must be considered.
• Errors of precision, validity, and study type, including random and systematic errors, are important.
• Sampling methods include random (simple, stratified), convenience, multi-stage, and cluster sampling.
• Informed consent
• Ethics in research involves informing the ethics and welfare committee.
• Recommended reading covers pages 228-242 from support material, specifying "Veterinary Epidemiology 3rd edition: Thrusfield, Michael".

Concepts to Understand Different Sample Objectives

• The lecture offers concepts required to understand different sample objectives
• An understanding of Prevalence and Incidence is critical.
• Sample size calculations help to determine prevalence, detect disease, and estimate means or differences.
• Knowing study designs in Epidemiology helps decide sample sizes for cohort studies and case-control studies.

Prevalence and Incidence

• Prevalence (P) is the proportion of animals in a population with a particular disease at one time
• Used to measure the total disease burden in a population.
• Incidence refers to the occurrence of new cases of a disease over time, which measures the risk of developing the disease.
• Cumulative Incidence (CI) represents the proportion of a population at risk that develops a disease over time
• Indicates the risk of an individual developing the disease.
• Incidence Rate (I) measures the rate at which new diseases occur in a population over time
• Considering the total time at risk contributed by all individuals.

Types of Studies

• Descriptive Studies describe population characteristics (who, where, what, when)
• Used to describe diseases and can provide a basis for hypotheses.
• Epidemiological indicators include prevalence, incidence, and mortality.
• Analytic studies assess the magnitude of association between variables and outcomes, related to exposure and disease.
• Cohort studies track exposed groups for disease development
• Case-control studies compare diseased and non-diseased groups to identify risk factors.
• Studies can be Prospective, Retrospective, or Cross-Sectional based on objectives, urgency, disease prevalence, and financial constraints.
• In prospective studies, people are tracked forward in time.
• Retrospective studies look backward from already affected individuals.
• Cross-sectional studies assess a population at one specific time.

Study Design Implications

• Cohort studies estimate disease incidence and assess rare exposures. However, they are long in duration and have a risk of loss to follow-up.
• Case-control studies enable the simultaneous study of multiple risk factors and are comparatively fast and low-cost.
• Cross-Sectional studies allow the determination of disease prevalence, which describes the situation at the time of data collection.
• Ethics committees are needed in research proposal examples.
• Informed consent from animal owners is needed

Census versus Sampling

• A census evaluates all individuals, or all individuals in a population are evaluated.
• Sampling assess only a subgroup of a population

Populations

• Target population is the group to which study conclusions apply.
• Source population is from where a study sample originates.
• Population sample refers to the selected individuals for the study.

Accuracy and Precision

• Accuracy refers to how close sample measurements are to the actual population value.
• Precision refers to how consistent and repeatable the measurements are.

Validity

• Internal validity: The findings are trustworthy for the people in the study.
• External validity: The findings can be applied to a larger group or whole population.

Sampling Basis

• Requires a list of sampling units.
• For example, a list of veterinary practices, pig farms, dogs in a kennel, or cows on a livestock farm.

Types of Errors

• Alfa error means the incorrect detection of an effect when it is not present in reality (false positive).
• Beta error -> a failure to detect a real, existing effect (false negative).

Hypothesis testing

• Type I error (α) is a false positive and it represents the probability of rejecting the null hypothesis (H0), when it is actually true.
• The confidence level (1-α) is the probability with which the result of the study reflects the truth.
• Type II error (β) is a false negative and it expresses the probability of not rejecting the null hypothesis when it is false.
• Power (1-β) is the probability of rejecting the null hypothesis when it is false.
• Power and confidence levels are parameters used in sampling tools

Types of Sampling

• Non-probability sampling involves the choice of an individual is not based on a probabilistic method.
• In probabilistic sampling, all individuals share a probability greater than 0 of being sampled.

Non-probability Sampling

• Subjective or targeted sampling involves implicit criteria from the researcher.
• Convenient sampling comes from easy samples during research.
• Quota sampling builds on specific characteristics of the elements of a population.

Probabilitistic sampling:

• Simple random sample is when selecting objects/elements for a sampling with equal probability to be selected.
• Systematic is selecting with regular intervals, like every nth person
• Stratified selection bases on a stratum or sub-group of the population
• Cluster's primary unit is larger than the experimental

Sample Determination

• Requires specification of the expected result, expected prevalence, power, and confidence interval.
• Definition of the recruitment objective.
• Leads to the development of timetables and budgets.
• Aims to avoid conducting small, inconclusive studies.
• Reduces the unnecessary use of animals.

Qualitative versus Quantitative Approach to Sample Determination

• The qualitative approach determines whether the infection is present.
• The quantitative approach quantifies the presence of infection.

Sample Determination

• Precision relates to how close measurements are to each other
• Variability signifies how different the results are from one another.
• Confidence level reflects the certainty that results represent the true population
• Power of analysis reveals a study's ability to detect an effect if it exists.

Z-Score

• Most formulas incorporate the concept of Z, which represents a probability area under a normal distribution curve.
• The Z-score helps determine the probability of an event occurring given a normal distribution.

Estimating a Proportion/Prevalence

• As an example, a calf farm owner estimates the prevalence of salmonellosis caused by Salmonella dublin.
• Prevalence is assumed to be 50% if unknown for sample size determination.
• Using a lower margin of error (i.e. 5%) will improve the precision.
• Requires a 95% confidence level (Z = 1.96).

Prevalence (Proportion) Estimation

• Z is the value from the standard normal distribution matching the wanted confidence level (z=1.96 equates to 95% confidence interval)
• P equals the expected true proportion, while Q equals 1-P.
• L is the actual precision used in the experiments (half the desired CI width).
• Sample Size Calculation: The formula details how to calculate sample size, factoring in the Z-value, P, and L.

Sample Size for Estimating Proportions or Prevalence

• Basic Description The utility is described with the equation needed
• The tool requires inputs of Estimated proportion, the "Desired precision ±" and the "confidence level" to estimate different prevalence values

Sample Size: Difference Between Two Proportions

• Basic equal-sample size formula can be used for same size groups
• Unequal allocation ratio formula is required if one group is much larger than the other.
• Fleiss' formula is used for large studies.
• Cochran's formula is for rare diseases w/ small incidence rates.
• Cluster-adjusted formula is needed when data is aggregated (large animal epid unit such as herds).

Selecting a Method

• Selecting the adequate tool or formula depends on the problem
• All cited formulas should use evidence based medicine to justify
• Quick hack:
• Look at similar studies at their METHODS section related to sample calculation If all hope is lost, collect n > 30 because of the CENTRAL LIMIT THEOREM
• Small adjustment to these parameters can heavily effect the data
• These parameters are here to familiarize not memorize

Sample Calculation for Finite Populations

• A formula is given to calculate sample size for finite populations.
• Example: a population of Bovine respiratory disease
• Shows how to determine the sample size (n) by plugging in 0.05(Alfa), expected affected animals (30) and the total affected animals

Calculating the Sample Size to Estimate the Mean

• Z 1-a/2 corresponds to desired confidence level (1.96 for 95% confidence)
• Population standard deviation is (or an estimate from past studies)
• Margin or error corresponds to difference between same mean and true mean

Example with Milk Farm

• Includes weighing calves to confirm the daily weight gain on the farm between 950 and 1050 g/day, shows how a sample size to estimate
• Includes calculation of daily rates
• Shows how to to include standard deviation, margin of error, and confidence levels

Sample Size to Estimate Difference Between Two Proportions

• Bovine population with expected high population
• Provides calculations with current and desired incidence, significance and % power (α and β respectively)

Frequent Sample Calculations

• Difference between two means
• The equation is given with corresponding sample sizes needed to determine the differences

Cohort Study,

• Sample sizes are similar to the two previous examples
• Requires statistical power and power of the test, to determine the minimum effect

Case With Student Group

• Student group calculating needed needed animals while considering if environment impacted Leishmaniosis while assuming 95%
• Provided estimations with running and different equations
• Provided that using AI might present different result

Case-Control Study

• Sample size or a case control of rare canine degenerative myelopathy (CDM)
• Provided an equation with case-control setting, along with associated calculations