Veterinary Epidemiology: Course Overview

Introduction to Epidemiology

• Epidemiology studies for a Master's Degree in Veterinary Medicine take place at Egas Moniz School of Health & Science.
• The course happened on March 17th, 2025.
• Professor Dr. Rita Santos (PhD) instructs the course

Program & Evaluations

• Causality is part of the program
• Frequency measures are assessed using a mini test
• Field Epidemiology will have a mini test
• Building Tools Survey includes a practical assessment
• Association Measures, particular Epid Studies will have a practical assessment
• Preparing Research Proposal is part of the program
• Sampling is part of the program
• Diagnostic Tests which include Epid. Surveillance will have a practical assessment
• Research Proposal Presentation includes an integrative test

Topics to be discussed

• Definition of Epidemiology (p22-28 in "Veterinary Epidemiology")
• Concepts and criteria of causality (p34-38 in "Veterinary Epidemiology")
• Descriptive epidemiology (p28-32 in "Veterinary Epidemiology")
• Determinants of disease and health in populations (p75-95 in "Veterinary Epidemiology")
• Ecological concepts of disease (p116-120 in "Veterinary Epidemiology")
• Agent-host interdependence and the environment
• Introduction to confounding factors, types of bias (p 40-44 in "Veterinary Epidemiology")
• Temporal and spatial patterns of disease (p137-145 in "Veterinary Epidemiology")
• The recommended reading is "Veterinary Epidemiology 3rd edition" by Thrusfield, Michael

Association measures

• Association measure involves the Definition, calculation and interpretation of measures of association such as Relative risk, incidence rate ratio, odds ratio
• Association measure also involves communication and reporting of risk

Types of epidemiological studies

• Descriptive studies include case reports, case series, cross-sectional studies and surveys
• Observational studies include ecological, cross-sectional, cohort, case-control, hybrid
• Experimental studies included randomized clinical trials, laboratory studies

Calculating the Magnitude of the Relationship Between a "Cause" and an "Effect"

• Risk Ratio/Relative Risk compares the frequency of a disease in individuals exposed and unexposed to a specific variable
• Odds Ratio compares the likelihood (odds) of exposure between cases and controls in a study

Counterfactual – individual effect

• It is possible to show the counterfactual individual effect using visual tools to show the outcomes of vaccination

Counterfactual - average effect in the population

• Average population effect can be recorded in a table to display different outcomes in a clear manner

Risk (Cumulative Incidence)

• Risk(R) is defined as all the new cases in a defined time period divided by the population at risk during the same period.
• It is the proportion of individuals who develop a disease over a specified period.
• Data about disease occurrence and exposure can be organised into a 2x2 table, with a representing exposed individuals who have disease present, b representing exposed individuals who do not have the disease, c representing non-exposed individuals with the disease and d representing those not exposed and without the disease
• Consequently, a is the number of individuals exposed with disease
• b is the number of individuals exposed without disease
• c is the number of individuals not exposed with disease
• d is the number of individuals not exposed without disease

Risk Ratio or Relative Risk (RR)

• RR can be calculated as RR = P(D+|E+)/P(D+|E-) = (a/a+b)/(c/c+d)
• RR is the ratio of risk between exposed and unexposed groups
• In a table the risk in the exposed group is a divided by a+b.
• The risk in the unexposed group is c divided by c+d.

Risk Ratio Example

• if RR = (38/5000)/(2/1000) = 3.8
• Animals with the study characteristic have 3.8 more risk to develop the disease than those without the characteristic

Another Risk Ratio Example

• RR = (??/??)/(??/??) = ??
• It is possible to calculate the RR if enough data is available

Incidence rate (I)

• It can be derived as the "New cases in a defined time period", divided by the "Number of animal – time units at risk during the defined time period"
• a is the number of individuals exposed with disease
• b is the number of individuals exposed without disease
• t₁ = Total time at risk on exposed individuals
• to = Total time at risk on not exposed individuals

Incidence Rate Ratio (IRR)

• IRR equation is: IRR = (Incidence Rate in Exposed Group) / (Incidence rate in Unexposed Group)
• The incidence rates of disease or events are compared between two groups.
• To calculate, use IR = (a/t₁)/(b/to)

Example IRR

• if IR = (18/250)/(8/236) = 2.12
• The disease rate on exposed animals is 2.12 times higher than on not exposed animals

Another IRR Example

• if IR = (??/??)/(??/??) = ??
• It is possible to calculate the IRR if enough data is available

Odds ratio (OR)

• To calculate the odds ratio, a 2x2 table is used, as with prior measures
• a = nº de indivíduos expostos que têm doença (number of exposed individuals who have the disease)
• b = nº de indivíduos expostos que não têm doença (number of exposed individuals who do not have the disease)
• c = nº de indivíduos não expostos que têm doença (number of non-exposed individuals who have the disease)
• d = nº de indivíduos não expostos que não têm doença (number of non-exposed individuals who do not have the disease)

Odds ratio (OR) equations

• Animals that are exposed have 7.9 times higher odds of developing the disease compared to animals that are not exposed.
• ORdisease = odds(D+|E+)/P(D+|E-) = (a/b)/(c/d)
• ORexposition = odds(E+|D+)/P(E+|D-) = (a/c)/(b/d)

Odds ratio (OR) numeric example

• If the oddsratio can be calculated this way: ORdisease = odds(D+|E+)/P(D+ | E-) = (100/30)/(50/120)=3.33/0.417=8

Odds ratio (OR) context example

• The oddsratio can be calculated this way: ORexposition = odds(E+|D+)/P(E+ | D-) = (100/50)/(30/120)=2/0.25=8
• Animals with the disease have 8 times higher odds of having been exposed compared to animals without the disease

Another Odds ratio (OR) example

• It is possible to get the following result using ORdisease = odds(D+|E+)/P(D+ | E-) = (??/??)/(??/??)=??
• ORexposition = odds(E+|D+)/P(E+|D-) = (??/??)/(??/??)=??

Symmetry of the Odds Ratio (OR)

• OR disease = (a/b)/(c/d)
• Can be expanded as (a * d)/(b * c)
• (a/c)/(b/d) = ORexposure

Another Odds ratio (OR) numeric example

• If a scenario delivers ORdisease = (38/4962)/(2/998) = 3.82
• Animals with the characteristic in study have 3.8 times higher “chance" of developing the disease than those without the characteristic

Interpreting the results

• There is a scale to interpret the results of data analysis, with OR, IR, and RR involved
• The scale shows that 1 is the null value
• Below 1 is the reduced risk and above 1 is the increased risk

Effect measures

• Effect measures are absolute measures that quantify the impact of an exposure on disease occurrence
• They can be calculated for the exposed group or the population

Effect measures – Risk Difference

• Risk Difference involves the absolute difference in risk between an exposed group and an unexposed group.
• P(D+|E+) = Probability of disease in the exposed group (a / (a + b))
• P(D+|E-) = Probability of disease in the unexposed group (c / (c + d))

Calculating Risk Difference

• RD = P(D+|E+)-P(D+|E-) = (a/a+b) - (c/c+d)

Risk Difference example results

• If RD = P(D+|E+)-P(D+|E-) = (60/100) - (20/120)=0.6-0.17=0.43

Yet another Risk Difference example

• RD = P(D+E+)-P(D+|E-) = (??) - (??)=???
• If enough data is collected, it is always possible to calculate it

Effect measures – Incidence rate

• This is defined as the "absolute difference in incidence rates" (new cases per unit of time) between an exposed group and an unexposed group.
• ID = (a/t₁) - (b/to)

Hypothesis tests and confidence intervals

• A series of equations are used to do these calculations including:
• Odd Ratio (OR): Var(lnOR) =(1/a)+(1/b)+(1/c)+(1/d)
• Risk Ratio (RR): Var(InRR) = b/a(a + b) + d/c(c+d)
• Risk Difference (RD): Var(RD) = a over a+b (1-a+b) + c over c+d(1-c+a); a+b c+d

Interpretation of Confidence Intervals

• If the confidence intervals includes the null value, the effect is not statistically significant
• RR/IR/OR -1
• RD - 0

Attributable Fraction in the Exposed Group

• The proportion of disease in exposed individuals that is due to the exposure, assuming a causal relationship
• AFe = RD/P(D + |E +) = {(a / a+b)-(c/c+a)}/ (a / a + b)
• AFe = (RR-1)/RR
• AFe = (OR - 1)/OR

Example: vaccination

• If P(E+) = vaccinated animals
• P(D+|E+) = 0.20
• P(D+|E-) = 0.05
• RD = 0.20 – 0.05 = 0.15
• AFe = 0.15/0.20 = 0.75 = 75%
• The vaccine prevented 75% on the vaccinated group.

Association and Effect measures

• Cumulative Incidence Ratio (Relative Risk)
• Odds Ratio
• Incidence Rate Ratio
• Risk Difference (Attributable Risk)
• Attributable Fraction (in Exposed Individuals)

Types of epidemiological studies

• Descriptive: case reports, case series, cross-sectional studies and surveys
• Observational: ecological, cross-sectional, cohort, case-control, hybrid
• Experimental: randomized clinical trials, laboratory studies

Epidemiological studies

• Descriptive Studies describe a disease and its evolution within a population over time and space
• Descriptive studies help to form hypotheses regarding risk factors
• Key parameters in descriptive studies are epidemiological indicators, i.e. Prevalence, Incidence, Mortality, confidence intervals for epidemiological indicators
• Analytical Studies are for testing hypotheses about the relationship between a variable and a specific event
• Key Parameters for analytical studies are: Measures of association: Relative Risk (RR), Incidence Rate Ratio (IRR), Odds Ratio (OR), Attributable Fraction (AF)
• Statistical Analysis uses Confidence intervals for association measures and p-value

Clinical Trials

• Clinical trials are planned studies in individuals in their natural environment
• Treatment groups are usually assigned randomly
• Are the best epidemiological study design to estimate a cause-effect relationship.
• Clinical Trials assess the impact of an intervention on a specific clinical outcome
• Clinical Trials are useful for evaluating the effectiveness of prophylactic or therapeutic drugs, diagnostic methods, and public health programs.

Developing a Study Protocol is Essential for Clinical Trials

• Study objectives
• Target population
• Method of selecting individuals
• Intervention under study
• Blinding procedures
• Study duration
• Specification of the outcome and how it will be measured
• Statistical analysis plan
• Ethical considerations

More Considerations for Clinical Trials

• Study objectives – Avoid an excessive number of objectives
• Primary and secondary outcomes
• Target population
• Unit of analysis – Individuals, groups, etc.
• Eligibility criteria
• Sample size determination
• Method of selecting individuals
• Random selection / Randomisation
• Systematic selection

Intervention Under Study in Clinical Trials

• Clear instructions on how to administer the study drug is important
• Methods to apply blinding (masking) in studies:
• Single-blind – The treatment group is unknown to the study participants.
• Double-blind – The treatment group is unknown to both the participants and the researchers.
• Triple-blind – The treatment group is unknown to the participants, researchers, and data analysts
• Comparison Strategies:
• New drug vs. existing drug
• New drug vs. placebo (inert substance)
• Comparison between already approved drugs

Defining elements in Clinical Trial

• Follow-up duration – Ensuring equal observation time across groups
• Compliance – Participants adhering to the study protocol
• Define and Measure the Study Event
• Objective events are data-measurable outcomes such as Mortality or Quantifiable clinical signs (e.g., fever) or Productive parameters (e.g., milk yield)
• Subjective events relay in the perception of events, such as Fatigue or Behavioral changes

Clinical Trial final parts

• Statistical Analysis should define the appropriate statistical method to be used in the
• Ethical Considerations always apply, such as weighing the side effects of the intervention or comparing medication vs placebo
• When considering Sample size, take ethical considerations seriously: Too small size fails to establish a cause-effect relationship, Too large size is Unnecessarily sacrifices excessive animals
• Informed consent should include agreement from study participants or animal owners for inclusion in the study.

Pharmaceutical clinical trials phases (translated)

• Phase 1: Conducted on healthy individuals to evaluate safety and tolerance of a new medication.
• Phase 2: Initial evaluation of a medication in a small number of patients from the target population.
• Phase 3 (randomisation): Large-scale clinical trials with the objective of comparing the medication under study with another already used in terms of efficacy, safety, and therapeutic benefit.
• Phase 4 (randomisation): Studies conducted after the medication is approved with the objective of improving its utilisation."

Contrafactual

• Contrafactual – individual effect can be visually represented using a cow that either gets vaccinated with a disease absent or a cow that does not get vaccinated and the disease is positive.
• Counterfactual – average population effect can be displayed using a grid and the formula P(D+|V+) = P(D+|V-)
• Counterfactual – average population effect uses Clinical Trial randomization which delivers Balance and similar groups

Cross-sectional studies

• Cross-sectional studies are done at a specific point in time using a sample or census of the target population
• Difficult to determine causal relationships
• Features lack of temporal ordering
• The estimated parameter is prevalence
• Cross-sectional studies are best when studying variables that do not change over time (e.g., sex, race)
• Repetition of cross-sectional studies may allow the determination of incidence rates
• Improve the comparability of groups and increase the study's rigor to reduce biases by incorporating proper Sampling methods (Random and Restricting to specific population groups) and implementing Statistical considerations
• It is possible to visualize the data is a graph and measure Cataracts in a group where Age and Relative Risk is a factor

Cohort Studies

• Cohort - a group of individuals who share a common characteristic Exposure variable is measured on the Cohort for the duration
• To measure the Exposure outcomes, metrics such as Date of Death, presence of certain Diseases and Recovery are used
• Cohort Studies can measure Permanent Exposure, Unique expositions,or Non-permanent exposition
• Types:
  • Permanent Exposure e.g. Age, sex, race
  • Unique expositions eg vaccinations Non-permanent exposition eg Diet, treatment, environment factors
• Cohort study "gold standard" of the observational studies
• Population Populações that give origin to cohorts are varied e.g.
  • Sample of the general population
  • Geographic area
  • Professional group
  • Risk groups for a certain disease
• Graphically one can describe cohort studies using lines to show a disease presence or absence, and then track the incidence rates (E+) and (E-)
• Relative Risk is an important factor to measure here
• Time is an important factor to measure and visualize data with
• Improve cohort comparability and increase study rigor to reduce bias by careful Cohort selection (Restrict to groups within the population to reduce variability in results) and implementing Statistical considerations.

Relative Risk (RR) calculation

• RR = P(D+E+)/P(D+|E-) = (a/a+b)/(c/c+d)
• It is presented in a table

Incidence Rate Ratio (IRR) calculation

• IR = (a/t₁)/(b/to)
• It is presented in a table

Effect measures: Risk Difference (RD) calculation

• RD = P(D+E+)/P(D+|E-) = (a/a+b) - (c/c+d)
• It is presented in a table

Association measures & type of studies factors

• Consider:
  • Measure
  • Definition
  • Equation
  • Cross-sectional
  • Cohort
  • Case control For the factors like Relative Risk. Incidence, Odds Ratio, Risk Difference

Evidence Hierarchy

• Evidence Hierarchies go in the following order:
  • Consistency of results across multiple sources.
  • Systematic review of large clinical trial results.
  • Clear results from a large clinical trial.
  • Systematic review of small clinical trials.
  • Literature review of cohort study results.
  • Clear results from a cohort study.
    • Empirical observations with expert opinion.

Takes of the Day - Association and Effect measures

• Cumulative Incidence Ratio (Relative Risk)
• Odds Ratio
• Incidence Rate Ratio
• Risk Difference (Attributable Risk)
• Attributable Fraction (in Exposed Individuals)

Takes of the Day - Epidemiological studies

• Descriptive: case reports, case series, cross-sectional studies and surveys
• Observational: ecological, cross-sectional, cohort, case-control, hybrid
• Experimental: randomized clinical trials, laboratory studies