Principles of case control study (Epidemiology)

Question 1

Define a risk factor

Answer 1

Any characteristic that identifies a group of people as at increased risk of disease e.g. homelessness for TB.

DOES NOT need to be causal, independent or modifiable

Question 2

Define “cause of disease”

Answer 2

A factor which, of itself, increases the risk of a disease occurring.

An event/condition/characteristic without which the disease is less likely to occur e.g. hypertension –> haemorrhagic stroke

Question 3

Confounding factor

Answer 3

A factor associated with the “exposure” being studied and disease outcome. i.e. the association between the air travel and VTE could be confounded by age.

Question 4

What is a case control study?

Answer 4

Case of disease occurs

Was the person exposed of not

Compare with control and if control was exposed or not

Question 5

Design features in case control studies?

Answer 5

1. Hypothesis
2. Size of study and statistical power - Needs to be big enough to find an association if present. If too small there is scope for random error to occur
3. Select cases of disease and case control.
   Define definition of disease case - newly diagnosed or established cases
   Ideal case control - similar confounding factors as the cases - consider case matching age, smoking status etc. Person who were they to develop the disease would have come a case
4. Conduct study - Measure exposure
   Consider setting, method/instrument used, informant, environment
   Participants should be blind to hypothesis
   Objective measure should be used e.g. self administered questionnaire
5. Manage confounding factors
6. Approach to data analysis

Relative risk = risk of disease in exposed/risk of disease in unexposed

Interpreting positive results - question if association is true
Could the association be due to chance/bias/confounding factors/reverse causation?

Interpreting negative results - question is null association is true.
Is study to small, measurement inaccurate, bias

Question 6

How to deal with confounding?

Answer 6

Case matching - match each case and control with similar confounding factors, e.g. for every 60 year old man with exposure, a 60 year old man without

Complete whole study in people with same level of confounder e.g. all smokers

Manage at analysis stage

Question 7

How to deal with bias?

Answer 7

Information bias -