Quiz 1 – Lectures 1 & 2

- Clinicians use epidemiology all the time, because:

o It’s impossible to interpret signs and symptoms seen in an individual animal except in
the context of population data
o It’s impossible to evaluate the efficacy of treatments except by using population data
(ideally in form of controlled experiments)
o Population medicine, i.e. disease control and prevention is applied epidemiology
o Epidemiology underlies diagnosis, prognosis, and selection of treatment!
- The epidemiological approach
o Show association, and then show the causal relationship through experimentation
 E.g., add and remove the agent that you believe to be causing the disease
 Showing causation involves a hypothesis and an experiment
- Epidemiology is the study of disease in populations
o Measuring disease in populations
o Identifying risk factors
o Evaluating treatment options
- Patterns of Disease
o Endemic – used in two ways:
 Habitual presence of a disease in a given area
 The usual level of occurrence in an area
o Epidemic – clearly in excess of a normal expectation
 Not a spatial relationship

o
o Pandemic – a global epidemic
 How? Very quick spread OR synchronized environmental change?
- Spatial patterns of disease
o Random vs. clustered
o Factors that shape the distribution of disease
 Epidemiologic Triad – think variation in EXPOSURE and SUSCEPTIBILITY
 
- Temporal patterns of disease
o Short-term outbreaks


o Long-term (“secular”) trends


o Cyclical trends
 Think variation in EXPOSURE and SUSCEPTIBILITY!


- Incidence
 o
o Gold standard of disease measurement – defines the risk of getting a disease
o Always specify per year, per month, etc.
o Measuring incidence


- Incidence Rate

o
o Incidence rate accounts for the fact that you may not follow-up on the same exact
population measured in the first study
o Example:
 Observe 18 dogs for testicular cancer – some drop out
  10 for 1 year
 4 for 9 months = 0.75 year
 4 for 6 months = 0.5 year
 3 dogs develop testicular cancer within the testing period
 Incidence = 3 new cases / 18 dogs at risk = 0.16 = 16% incidence
 Incidence Rate = 3 new cases / [(10 dogs x 1 year) + (4 x 0.75 year) + (4 x 0.5
year)] = 3/15 = 0.20 = 20% incidence rate
 Higher risk than if you’d just measured incidence
- Prevalence

o
o Most common measurement of disease in a population because it’s easy, but it’s less
informative than incidence
o Prevalence = Infected / (Susceptible + Infected + Resistant)
o The number of infected can change based on changes in transmission rate, changes in
death rate, etc. – prevalence doesn’t tell you anything about risk or survival time with
the disease or if a control program is working

o
 Solid lines = animals with disease
 Dotted lines = animals without disease
 Prevalence in January = 3/5
 Prevalence in July = 5/5
 Prevalence in December = 2/4
 Incidence of animals susceptible to disease becoming infected = 2/2
 Two non-infected animals at beginning of study became infected
- Relationship between Prevalence and Incidence
o Prevalence is the baseline – it tells you the number of diseased animals in the
population, and it can decrease through deaths and cures. It is the balance between
new cases coming in and deaths/cures.
o Incidence tells you the rate at which diseased animals are being added to the population
o A lot of medical interventions prolong the life of animals with a disease. This will cause
prevalence to increase, because more animals are in the diseased category for longer.
 Disease prevalence may just be increasing due to good supportive care
- Measuring Incidence and Prevalence
 o Simplified: Prevalence = Incidence x Duration of Disease

o
 If you didn’t know the difference between incidence and prevalence, you would
think that Hitown had a higher risk of disease
 In fact, Hitown has a longer duration of disease than Lotown, increasing
prevalence
 Prevalence is approximately incidence x duration
 100 cases / 25 years = 4 occurrences/year
 60 cases / 3 years = 20 occurrences/year
 Therefore, per capita risk is higher in Lotown
- Mortality Rates
o Cause-specific Mortality Rate


 Doesn’t differentiate between diseases that are really common, and diseases
that have a high risk of death
 If disease treatment improves, but a disease becomes more common, it
will appear to have a flat rate on a graph
 Allows you to rank the most important causes of death and what you target
o Case Fatality Rate


 Measure of the disease severity – How likely is it that infected animals will die?
o Age-adjustment
 Problem: age structure can skew death rates. Populations with different age
compositions can affect how the risk looks. Could also apply to sex, breed, etc.
 Comparison of total death rates in a population at two different times:


 
o Increase in death rate could be misleading because all that
changed was the age composition
 Define age-specific death rates and compare those!
- Summary:
o Incidence represents the likelihood of becoming a case = RISK.
o Prevalence represents the likelihood of being a case = DISEASE BURDEN in the
population.
o In the simplest case, prevalence = incidence x duration of disease.
o Mortality rates allow comparing importance of different diseases in a population.
o Case fatality rates inform about severity of disease.
o When comparing populations, rates may need to be adjusted according to factors such
as age, sex or breed.
- Challenge of working up wildlife mortality/disease issues:
o May be more difficult to follow up individuals
o Relies heavily on outbreak investigation
o Opportunistic sampling/study design common
o More challenging to acquire good history on patients
- Steps in an outbreak investigation
o Verify the diagnosis via necropsy, isolation of pathogen/toxin, etc.
 Obtain more samples if necessary/possible
 Continue to document any morbidity/mortality
 Use specialists – toxicologists, pathologists
o Confirm the outbreak
 Outbreak = morbidity/mortality above expected background levels
 Look for trends
o Case definition – like a detailed description of the situation
o Descriptive epidemiology
 Who, what, where: describe environmental factors, species affected, etc.
o Develop a hypothesis
 Why did this happen?
 Test the hypothesis – go back to site and investigate if necessary
o Implement control and preventative measurements
o Communication of findings
 Press release, discuss results
Quiz 2 – Lectures 3 & 4
- Clinical data mean nothing until interpreted in the context of expected values for the
population!!
- Two basic properties of distributions can be used to summarize data: central tendency and
dispersion.
o Measures of central tendency: mean, median, mode
o Measures of dispersion: standard deviation, range, percentile
- Validity of Screening Tests
o Validity (accuracy) = ability of the test to distinguish between which individuals have a
disease and which do not.
o Sensitivity = ability to identify correctly those that DO have the disease.
 Sensitivity = TP / (TP + FN)
o Specificity = ability to identify correctly those that DO NOT have the disease.
 Specificity = TN / (TN + FP)
o Tests of continuous variables – where to draw the line?
 Low cut-off value = low specificity
 High cut-off value = low sensitivity
- Predictive Value of a Test
o Positive predictive value:
 If the test result is positive, how likely is it that the patient actually has the
disease?
 PPV = TP / (TP+FP)
o Negative predictive value:
 If the test result is negative, how likely is it that the patient really does not have
the disease?
 NPV = TN / (TN+FN)

o
- Using Multiple Tests Together
o Sequential (Two-stage testing):
 “Positive” = test positive on BOTH tests.
 HIGH specificity, LOW sensitivity
o Simultaneous testing
 “Positive” = test positive on EITHER test
 LOW specificity, HIGH sensitivity
Quiz 3 – Lectures 5 & 6
- Exposure: having come into contact with a (suspected) cause of, or possessing a characteristic
that is a suspected determinant of, a particular health problem.
o E.g., location of residence (village vs. farm dogs); demographic characteristics (sex, age,
breed, indoor/outdoor, diet); health characteristics (vaccination status, preventive
medication regimes, medical history); etc.
- Cross-sectional study – a study looking at disease and exposures during a certain time period
o Subjects chosen based on convenience
o Purpose: Collect data on exposure and disease at the same time from each animal (i.e.
compile data on group of subjects, record exposures and outcomes at one point in
time).
 Compare prevalence of disease in exposed vs. unexposed
 Compare prevalence of exposure in diseased vs. non-diseased

o
 Prevalence of disease compared in exposed and non-exposed:
a c
 vs .
a+b c +d
o Advantages:
 Relatively “easy” and cheap to carry out (e.g. with surveys or through disease
surveillance)
 Observed associations (among many factors and disease) can generate
hypotheses to guide further research
o Disadvantages:
 Prevalent cases are measured, not incident cases – measures only current
disease cases
 Cannot determine incidence of disease
 Cannot determine risk for developing disease
  Association may be between factor and survival with disease, not developing
the disease!
 Temporal sequence unclear: did disease develop first, and associated factor is a
result of disease, or vice versa?
- Case-control study – a study which compares proportions of exposure retrospectively
o Purpose: Collect retrospective data on exposure in animals with disease (cases) and
without disease (controls), and compare the association between disease and
exposure(s) (i.e. histories of diseased and healthy subjects compiled to look for past
exposures)
o Subjects chosen based on outcome/disease

o
o Advantages:
 Relatively fast and cheap to carry out
 Relatively few animals required
 Good for rare diseases
 Can examine the associations of many risk factors with the disease → test
hypotheses about associations
o Disadvantages:
 Finding accurate historical data can be problematic (e.g. recall bias)
 Finding appropriate controls can be difficult
 Without temporal information → cannot determine incidence of disease
 Cannot determine risk for developing disease (may only see surviving animals)
→ cannot infer causation
 Scope of inference limited to population represented by cases and controls
- Cohort study – a study which compares rates of disease
o Purpose: Collect data on development of disease in animals exposed to factor(s) of
interest or concern and compare incidence of disease in relation to exposure(s)
o Subjects chosen based on exposure
o
o Advantages:
 Good control of data quality
 Good for rare exposures
 Can calculate incidence of disease in exposed and unexposed → calculate risk
for developing disease
 Can examine multiple diseases or outcomes in exposed vs. unexposed groups
 Known temporal relationship of exposure and disease (especially in prospective)
o Disadvantages:
 Can assess effects of exposure to relatively few factors (i.e. only those recorded
at outset)
 Results not available for a long time (if study concurrent) → slow
 Slow, inefficient, and expensive to carry out because many animals are needed
over a long time period and not all of them experience the event of interest
 Impractical for rare diseases → not enough cases
 Higher potential for biased assessment of disease of interest (e.g.
overestimation)
 If beginning with a specific exposure, one can only study this factor that
differentiates the groups
- Analyzing case-control study data:
o Not a direct measure of risk (incidence) – use the odds ratio
 “Odds” – likelihood that an event occurs / likelihood that it doesn’t occur
 = P/(1-P)
o Odds Ratio (OR) = Odds Case Exposed / Odds Control Exposed
 Odds Ratio > 1 → supports a positive association between exposure and disease
(i.e. exposure is a risk factor)
 Odds ratio = 1 → no association
 Odds ratio < 1 → supports a negative (protective) association between exposure
and disease (i.e. those exposed are at lesser/lower risk)
 
- Example: A case control study with unmatched controls
o You want to see whether regular oral health checks have any effect on the occurrence of
severe dental decay in elderly cats.
 In your clinic database, you locate all the cases of elderly cats which had to have
more than one tooth removed in the last two years.
 You compile a group of cats of similar age that have had no dental problems as
controls.
 You then look at the two groups’ histories to see whether they experienced
different levels of oral care during the preceding years.
o Results:
 You find records for 44 cats with severe dental decay. Of these, 8 had dental
check-ups for at least three years prior to teeth being removed.
 You find records for 123 cats of similar age with no major dental problems. 27 of
these had regular dental checks in the three years preceding your study.

Dental Disease + Dental Disease -

Regular Oral Health Checks (+) 8 27
No Regular Oral Health Checks (-) 36 96
Odds: 8/36 27/96
8 ∙ 96
 Odds Ratio = =0.79
36 ∙ 27
 0.79 < 1 – an odds ratio less than one supports a negative (protective)
association between exposure and disease
 In this case, disease is associated with non-exposure (i.e., not having
regular oral health checks)
- Example: A case control study with matched controls

Cases Controls
Exposed Not Exposed
Exposed Exposed
 Not Exposed Not Exposed
Exposed Not Exposed
Not Exposed Exposed
Not Exposed Not Exposed
Exposed Not Exposed
Exposed Exposed
Exposed Not Exposed
Not Exposed Not Exposed
 Concordant pairs – either:
 Both exposed
 Neither exposed
 Discordant pairs – either:
 Case exposed, control not exposed
 Case not exposed, control exposed

Controls
Exposure + Exposure -
Cases Exposure + 2 4
Exposure - 1 3
 Exclude cells A and D (orange and grey) because they do not contribute to the
hypothesis. Only count the other two cells, B and C, in the odds ratio.
b 4
 Odds Ratio= = =4
c 1
 Since 4 > 1, there is a positive association between exposure and
disease (i.e. exposure is a risk factor)
o The matched study will always be the strongest study because it cuts out extraneous
information!

Quiz 4 – Lectures 7 & 8

- Randomized Trials – evaluate the effectiveness and side effects of new and established
interventions for disease prevention and therapy
- Essential features of effective studies:
o Controlled
o Randomized
o Blind – ideally, double-blind
- Selection of Subjects and Allocation to Treatment Groups
o Define the study population IN WRITING because this is the population that you can
make inference to in terms of your experimental results!
o Randomization – assignment to a group should be made before seeing the patients or
by an outside party (NOT the investigator)
o Stratified Randomization – avoids the slight chance that most males will end up in one
group, and most females in another – helps to balance randomization
 
- Criteria for the Validity of Clinical Trials
o Internal Validity – the extent to which study results are correct for a sample of patients
being studied
 RANDOMIZED, CONTROLLED, BLIND
o External Validity – generalizability, the extent to which results of a study can be
generalized to the whole population
 Is the sample of patients in the study REPRESENTATIVE of the patient population
as a whole?
- Relative Risk
o Answers the question: how much more likely are exposed individuals to become
diseased relative to unexposed?
Risk ∈Exposed Incidence∈ Exposed
o Relative Risk = =
Risk ∈Unexposed Incidence∈Unexposed
o Interpreting relative risk:
 RR = 1: Risk in exposed is equal to risk in unexposed: no association between
exposure and disease
 RR > 1: Risk in exposed is greater than risk in unexposed: positive association,
possibly causal
 RR < 1: Risk in exposed is smaller than risk in unexposed: negative association,
possibly protective
o Example:
 You treat 29 dogs with anthelmintic drug (“A”), and 30 dogs with anthelmintic
drug (“B”).
 21 dogs on “A” come back for follow-up, and 24 dogs on “B”.
 Fecal egg counts show positive results for GI helminths in 7 of the dogs
on the old treatment, and in 4 of the dogs on the new drug.

 Old Treatment  New Treatment

 29 dogs treated  30 dogs treated
 21 dogs follow-up  24 dogs follow-up
 7 are worm positive  4 dogs worm positive

 New = Exposure
 Risk in exposed = 4/24 = 0.17
 Old = Not Exposed
 Risk in unexposed = 7/21 = 0.33
 Risk Ratio = Risk in Exposed / Risk in Unexposed
 (4/24) / (7/21) = 0.5
  RR < 1 – so it indicates a negative, possibly protective, association
o Risk in exposed is smaller than risk in unexposed
o Therefore, this study points to the new drug working better
than the old drug
- Attributable Risk
o What incidence of disease is attributable to exposure?
o Compares the level of risk in exposed vs. non-exposed group
 Everyone has some risk – known as background risk
o Attributable Risk = Risk in Exposed – Risk in Unexposed
o How many must be treated to prevent one case?
 “Number needed to treat” = NNT = 1/AR
o Example:
 A field trial evaluates a FeLV vaccine. 30 feral kittens are vaccinated, and twenty
controls are not. All cats are captured again a year later.
 9 of the vaccinated cats became viremic during the study.
 17 of the control cats became viremic.
 What was the incidence of FeLV in both groups?
 Vaccinated = Exposed
o 30 vaccinated, 9 FeLV +
o = 9/30 = 0.3/year
 No Vaccine = Unexposed
o 20 unvaccinated, 17 FeLV +
o = 17/20 = 0.85/year
 What was the relative risk of FeLV in vaccinated cats compared to controls?
 Relative Risk = 0.30 / 0.85 = 0.35
o RR < 1: Risk in those exposed to the vaccine is smaller than risk
in unexposed: negative association; possibly protective.
o Therefore, vaccinated cats are 1/3 less likely to get FeLV than
unvaccinated. Or, unvaccinated cats are 3x more likely to
contract FeLV.
 What reduction in incidence can be achieved by vaccinating kittens?
 Attributable Risk = Risk in Exposed – Risk in Unexposed
o = 0.3 – 0.85 = (-0.55)
o 55 cases of FeLV could be prevented per 100 kittens by
vaccinating
 How many kittens must be vaccinated to prevent one case of FeLV?
 AR = (-0.55)/year
 NNT = 1/AR = 1/(0.55) = 1.8
o So for (approximately) every two kittens vaccinated, you can
prevent one case.
- Comparing Relative and Attributable Risk
o Relative risk is a measure of the strength of association.
o Attributable risk is a measure of the relative importance of a risk factor and its potential
as a target for disease prevention.