Y2 Session 7 - Interpreting Epidemiological Findings pt.2

Question 1

What is confounding?

Answer 1

It is the effect of an extraneous variable that wholly or partially accounts for the apparent effect of the study exposure, or that masks an underlying trust association.

Question 2

What are the four common ways to identify confounding?

only need one to confirm or deny confounding

Answer 2

1. Explore knowledge of the subject matter
2. Check the three conditions for confounding
3. Stratification - split the sample by the potential confounding
4. Compare crude and adjusted (adjusted for potential confounding)estimates

Question 3

What are the three conditions for confounding? (three questions that if the answer is yes for all then you have found a confounding)

Answer 3

1. Is it associated with the exposure in the source population?
2. Is it associated with the outcome in the absence of the exposure?
3. Is it not a consequence of the exposure?

Question 4

What is effect modification?

Answer 4

When the strength of the association varies over different levels of a third variable e.g. a drug may be more effective in women than men, this is effect modification.
This is a natural phenomenon and cannot be avoided.

Question 5

How do we present results with effect modification?

Answer 5

Present stratified results e.g. the results for both men and women if sex results in effect modification.

Question 6

How can we test for effect modification?

Answer 6

Mathematical tests:

1. Breslow-Day test
2. Q test
3. Interaction terms in regression models (most frequent)

Question 7

What is synergism and antagonism?

Answer 7

Synergism - effect modifier potentiates effect of exposure

Antagonism - effect modifier diminishes effect of exposure.