Confounding, Effect Modification, and Stratification Yu-Fen

1
Confounding, Effect Modification, and
Stratification

• Yu-Fen Li

2
Adding a Third Dimension to the RxC contingency
tables
3
Confounding

• A confounding variable is associated with the
  exposure and it affects the outcome, but it is
  not an intermediate link in the chain of
  causation between exposure and outcome.

4
Confounding example
5
Confounding example
Smokers
Non-smokers
6
Confounding example
Smoking
X
Drinking
Lung cancer

• Drinking is not associated with lung caner
• Smoking is a confounder

7
Controlling for confounders

• Confounders can be controlled for in the design
  phase of a study (randomization/
  restriction/matching).
• Confounders can be controlled for in the analysis
  phase of a study (stratification/ multivariate
  model).

8
Analytical identification of confounders

• Stratification
• Mantel-Haenszel Procedure
• Take a 2x2 table and break it into a series of
  smaller 2x2 tables (one table at each of G levels
  of the confounder yields G tables)
• Homogeneity test/ Mantel-Haenszel Summary Odds
  Ratio
• Example in testing for an association between
  lung cancer and alcohol drinking (yes/no),
  separate smokers and non-smokers.

9
Mantel-Haenszel Procedure Pros and Cons

• Advantages
• Mantel-Haenszel summary statistic is easy to
  interpret and calculate
• Gives you a hands-on feel for the data
• Disadvantages
• Requires categorical confounders or continuous
  confounders that have been divided into intervals
  
• Cumbersome if more than a single confounder

10
Effect modification

• Effect modification occurs when the effect of an
  exposure is different among different subgroups.
  
• aka interaction

11
Interaction example
12
Interaction example
Smokers
Non-smokers
13
Interaction example
Smoking
Drinking
Lung cancer

• Drinking is associated with lung caner
• Smoking is an effect modifier

14
Investigating interaction

• Effect modifiers can NOT be controlled
• Effect modification is not a bias, but a
  phenomena
• Effect modifiers can be evaluated in the analysis
  phase of a study (stratification/ multivariate
  model with an interaction term).

15
HOMEWORK Q1
16
HOMEWORK Q2
17
HOMEWORK Q2