Evidence for population health

Question 1

Define: Epidemiology

Answer 1

Determinants + distribution of disease + death

Question 2

What are the pros of anecdote + case series?

Answer 2

• Quick
• Easy to perform in clinic
• Provides new unobserved conditions
• Provides new potential risk factors

Question 3

What are the cons of anecdote + case series?

Answer 3

• Not scientific (can’t test a hypothesis)
• Seriously affected by observer bias
• Difficult to make inference about disease cause

Question 4

What is a cross-sectional survey?

Answer 4

• A snapshot of people with an outcome

- Count the number of people with a disease in a short time period in a pre-defined population

Question 5

What are the pros of a cross-sectional survey?

Answer 5

• Quick

- Good at estimating prevalence of the disease and the burden

Question 6

What are the cons of a cross-sectional survey?

Answer 6

• Only represents that point in time
• Cannot estimate incidence of disease
• Sampling frame may lead to bias

Question 7

Define: Prevalence

Answer 7

Cases of disease that are present in the group

Question 8

Define: Incidence

Answer 8

New cases of disease

Question 9

How is incidence measured?

Answer 9

• A register is used

- Record new cases in a pre-specified population

Question 10

What is a counterfactual method?

Answer 10

Would the disease have happened at the same time in the same person if the factor was not present?

This is not a practical method for answering the cause of the disease

Question 11

What are the pros of ecological studies?

Answer 11

• Cheaper
• Less prone to bias due to participation
• Easy to perform using routinely collected data
• Provides new hypotheses about the causes of a disease or condition
• Provides new potential risk factors

Question 12

What are the cons of ecological studies?

Answer 12

• Ecological fallacy - do population-level measures hold for the individual?
• Assume average value of risk factor applies to all individuals
• Assume average incidence applies to all individuals
• Data collection may vary

Question 13

Describe: case-control study

Answer 13

Case = someone who has the disease
Control = someone who does not have the disease

They are compared for an exposure

It is retrospective (i.e. after the disease has been diagnosed)

Question 14

What are the pros of a case-control study?

Answer 14

• Good for rare disease/exposures

- Fairly quick since no need to follow-up

Question 15

What are the cons of a case-control study?

Answer 15

• Prone to selection bias
• Prone to participation bias
• Finding a suitable control group can be difficult
• Difference in recall leading to bias

Question 16

Describe: Cohort study

Answer 16

• Start with a population with no people diagnosed with the disease
• Look at who is exposed/not
• Look at rate of outcome for disease

Question 17

Define: cohort

Answer 17

People who share a common experience or condition e.g. smokers, birth cohort

Question 18

What are the pros of a cohort study?

Answer 18

• Good for rare exposures
• Can look at multiple outcomes
• Reduces information bias
• Direct measure of incidence

Question 19

What are the cons of a cohort study?

Answer 19

• Inefficient for rare diseases
• Expensive
• Retrospective is quicker
• Loss to follow-up

Question 20

Describe: randomised control trial

Answer 20

GOLD-STANDARD

Study population = health people for preventative trial OR sick people for therapeutic trial

Study population is assigned randomly to treatment/control group

Control group does not receive treatment

Follow the groups through time + look at rates of outcome

Question 21

What are the pros of a randomised control trial?

Answer 21

• Strongest evidence for causality
• Selection bias removed
• Observer bias removed

Question 22

What are the cons of a randomised control trial?

Answer 22

• Not real life
• High cost
• Unethical for many research questions

Question 23

What are the different types of blinding in a randomised control trial?

Answer 23

1. Patient doesn’t know if they are having the treatment or not
2. Clinical doesn’t know what the patient is having
3. Analyst doesn’t know what drug A or drug B are

Question 24

Describe: intention to treat analysis

Answer 24

Assuming that every person follows their part of the randomised control trial correctly + analyses the data as if the patient did the therapy for the whole trial

Makes it the best representative of real life

Question 25

Define: Error

Answer 25

The difference between an estimated/measured value and the true value

Question 26

Give potential sources of error

Answer 26

1. Study design
2. Sample collection
3. Lab analysis
4. Data analysis
5. Data management
6. Data collection

Question 27

Define: diagnostic bias

Answer 27

When diagnosis is made based on exposure

Question 28

Give examples of: self-selection bias

Answer 28

• Participants contacting the study through adverts

- Some people are more likely to participate due to family history

Question 29

Define: information bias

Answer 29

Bias arising from measurement error

Question 30

What types of bias are included in information bias?

Answer 30

1. Recall bias
2. Interviewer bias
3. Surrogate bias

Question 31

When does misclassification bias occur?

Answer 31

When data is placed in categories