Epidemiology

Question 1

What are prevalence and incidence?

Answer 1

Prevalence = proportion of diseased individuals at given time (snap shot) = number of cases at time t/total number of individuals at risk at time t
Incidence = new cases during given time per unit of population at risk

Question 2

How to calculate odds, risk, odds ratio (OR) and relative risk/risk ratio (RR)?

Answer 2

Odds = yes/no (e.g. pregnant/not pregnant, diseased/not diseased)
Risk = the probability of disease = diseased/total
Relative risk/risk ratio = risk of disease in group a/risk of disease in group b
Odds ratio = odds of group a/odds of group b

Interpreting RR and OR:
<1: protective
= 1: no association
>1: positive association with disease

Question 3

What is cumulative incidence (incidence risk)

Answer 3

Number of new cases during a specific period/population at risk during same period

Question 4

What types of diseases would have a high incidence but low prevalence and vice versa?

Answer 4

High incidence but short duration: may have low prevalence

Low incidence but long duration: may have high prevalence

Question 5

Which studies use relative risk (RR) and odds ratio (OR)?

Answer 5

RR:
- used by cohort and cross sectional
- not case-control studies
OR:
- can be measured in cohort, cross sectional and case-control studies

Question 6

What is risk difference (attributable risk)?

Answer 6

= How much extra disease occurs due to an exposure (over and above what would have happened anyway)

= risk among exposed - risk among unexposed

Question 7

What is attributable fraction?

Answer 7

What proportion of disease in exposed animals is due to exposure

= (risk among exposed - risk among unexposed)/risk among exposed
= (RR-1)/RR
~ (OR-1)/OR

Question 8

What is bias?

Answer 8

Systematic error (as apposed to random error)

Question 9

Types of selection bias?

Answer 9

Healthcare access bias: Valuable, serious or unusual (e.g. out of season) cases may be more likely to be referred or sampled

Spectrum bias: cases with certain signs may be more likely to be detected (and risk factors for showing signs may be different)

Incidence-prevalence bias: When studying survivors, the risk of survival may be related to the exposure

“Volunteer” bias: where people agreeing to be in study are different to those refusing

Non-response bias: Particularly in questionnaires – where those that respond are different to those that do not

Loss-to-follow-up: where people lost to follow are different to those that remain in the study

Missing information in analyses: only individuals with all relevant information are included in analyses. If certain individuals more likely to have missing data, can lead to bias

Question 10

What is information bias? What is it affected by? Types?

Answer 10

Occurs when there is misclassification of animals as diseased and non-diseased or inaccurate measurements of study factor

Affected by test sensitivity and specificity

Applies to all types of “tests”

Measurement bias:
- Classification of cows as mastitic, based on visual inspection of milk; mild cases less likely to be classified as mastitic, compared to serious cases

Recall/responder bias:
Farmers which have had a particular problem may be more inclined to remember previous exposures (or deny them)

Question 11

What is interpretive bias? Types?

Answer 11

Occurs when different emphasis is given to different evidence when results are evaluated

Confirmation bias: selectively reporting information that agrees with your prior convictions

Rescue bias: discounting (uncomfortable) data by finding faults with study (but applying less stringent criteria to studies which agree with prior convictions)

Question 12

Hierarchy of information in order? Features of each?

Answer 12

Systematic reviews and meta-analyses - review of all published research in a field, critical appraisal, stats to integrate results if meta analysis

RCT - control can be placebo or another treatment, always prospective, randomised, often blinded

Cohort studies - follow cases to see what happens to them, can assess different treatments, not blinded or randomised

Case-control studies - compare cases with a group of controls, usually retrospective, not blinded or randomised

Individual cases or case series - retrospective, uncontrolled, descriptive only

Anecdote or expert opinion

Question 13

What is performance bias in clinical trials?

Answer 13

Systematic differences in other factors between groups (groups handled differently during the trial)

Question 14

What is detection bias in clinical trials?

Answer 14

Systematic differences in how the outcomes are scored between groups
E.g. visual scoring (subjective)

Question 15

What is attrition bias in clinical trials?

Answer 15

Systematic differences between groups in withdrawals from a study (so less reliable if more drop outs)
Best to do intention to treat analysis (ITT) and not exclude drop outs

Question 16

What are cross sectional studies?

Answer 16

Provide information on:
The prevalence of disease(s) and/or exposure(s)
Associations between the disease and study factors

Important considerations:
Examine population at one point in time (could be short period )
Sampling fraction same for exposure and disease status
Does not give information about the temporal relationship between the disease and study factor?

Question 17

What is accuracy?

Answer 17

= proportion of correct diagnoses provided by a test

Question 18

Analytical sensitivity and specificity?

Answer 18

Analytical sensitivity = the smallest amount of substance in a sample that can accurately be measured by an assay
Analytical specificity = the ability of an assay to measure one particular organism or substance, rather than others, in a sample.

Question 19

Difference between se/sp and predictive values?

Answer 19

Se and Sp used to evaluate tests and compare tests
Predictive values:
- cannot be used to compare different tests
- used to interpret an individual result

Question 20

How to calculate likelihood ratios?

Answer 20

LR(+) = probability of positive test result in diseased animal/probability of positive restyle in healthy animals = Se/(1-Sp)
LR(-) = probability of negative test result in diseased animals/probability of negative test result in healthy animals = (1-Se)/Sp

Question 21

How to calculate the odds of disease after a test using likelihood ratios? Then how to convert odds to probability?

Answer 21

Odds of disease after test = odds of disease before test x LR

Probability = odds/(1+odds)

Question 22

How to convert between odds and probability?

Answer 22

Odds = probability/(1-probability)

Probability = odds/(1+ odds)

Question 23

If have pre-test probability (prevalence), how to calculate post test probability?

Answer 23

Convert probability (prevalence) to odds
Multiply odds by LR = post test odds
Convert back to probability
Can use a nomogram to work this out

Question 24

What are Receiver Operating Characteristic (ROC) curves? What used for?

Answer 24

Plot of Se against the false positive rate (1-Sp) calculated at various cut off values
Optimal cut off point occurs where curve gets closest to top left corner
But adjusted depending on the potential impact off false positives or negatives

Question 25

What are the false positive and negative rates?

Answer 25

False positive: 1-Sp

False negative: 1-Se

Question 26

What are parallel and series tests? Why used?

Answer 26

Parallel:
- animals considered positive if any of the tests are positive
- increases Se but tends to decrease Sp of the combined tests
Series:
- animals considered to be positive if all tests are positive
- increases Sp of the combined tests at the expense of Se
- can miss diseased individuals

Question 27

Koch’s postulates?

Answer 27

Koch theorized that a pathogen must be:
found in all cases of the disease examined
prepared and maintained in a pure culture
capable of producing the original infection, even after several generations in culture
[retrievable from an inoculated animal and cultured again]

Question 28

What are passive and active surveillance?

Answer 28

Passive = reporting of suspect cases
Active = regular, periodic collection of case reports, involving sampling of clinically normal animals

Question 29

Types of active surveillance?

Answer 29

Sentinel surveillance:
- farms, or species, that provide ‘early warning’
Serological surveillance (serosurveys):
- Antibody as evidence of current/past infection e.g. liver fluke surveys
Targeted surveillance:
- to collect specific information about a disease in a defined
population e.g. testing cattle >36 months for BSE
Scanning surveillance:
- Continuous watch over endemic disease (e.g. regular questionnaires to vets, TB skin tests; meat inspection)
Syndromic surveillance:
- Seeks high frequency of certain pattern, as possible sign of disease outbreak or emergence, e.g. neurological signs as indicator of WN in horses in France

Question 30

What is R0? Rt?

Answer 30

Basic reproduction rate
R0 = average number of cases (secondary infections) generated by a single infectious individual (primary infection) introduced into a totally susceptible population
R0 <1 - extinction is likely
R0 = 1 - endemic
R0 > 1: epidemic can occur
The greater R0, the more infectious the disease
Proportion needed to be vaccinated: = 1 - (1/R0)
Rt = average number of secondary cases that each infectious individual would infect if conditions remained as they were at time t (under control if <1) = R0 x S (proportion of population still susceptible)

Question 31

Define infectious disease, communicable disease, contagious disease and transmissible disease?

Answer 31

Infectious disease: a disease caused by a microorganism and therefore potentially transferable to new individuals.
Communicable disease: an infectious disease that can be transmitted from one individual to another.
Contagious disease: a very communicable disease capable of spreading readily between individuals (somewhat obsolete).
Transmissible disease: a disease that can only be transmitted between individuals by ‘unnatural’ routes

Question 32

Disease modelling

Answer 32

D = duration of infectiousness (e.g. number of days)
κ = number of animals the infected animal contacts per unit time (e.g. per day)
β = probability of transmission if contact occurs
S = proportion of the population that is susceptible

R0 = D x k x β
R = D x k x β x S

Question 33

How to calculate positive predictive value, negative predictive value, and predictive value of a positive result?

Answer 33

PPV = true positives/test positives
NPV = true negatives/test negatives
Predictive value of a positive result = (Se x prevalence)/((Se x prevalence) + (1 - Sp) x (1 - prevalence))