Lectures 11-12

Question 1

Effect Modification

Answer 1

A 3rd variable that modifies the magnitude of effect of an association by varying it within different levels of that 3rd variable

Question 2

What is difference in how you handle

1. Confounder
2. Effect modifier

Answer 2

1. Control/adjust for variable

2. Evaluate and describe strata

Question 3

(T/F) Confounder is associated with both the exposure and the outcome?

Answer 3

True

Question 4

(T/F) Confounder can be in causal pathway between exposure and outcome.

Answer 4

False

Question 5

Two effects confounder has on association between exposure and outcome?

Answer 5

1. Change in Magnitude

2. Change in Direction

Question 6

Residual Confounding

Answer 6

When confounder measured imperfectly so adjustment using this imperfect value does not completely remove effect

Question 7

Testing for Confounding Steps

Answer 7

1. Crude RR/OR
2. Adjusted RR/OR
3. If difference is greater than or equal to 20%, confounder is present

Question 8

Testing for Effect Modifier Steps

Answer 8

1. Crude RR/OR
2. Crude RR/OR for each strata
3. Difference between the highest and lowest strata is greater than or equal to 20%, variable is effect modifier

Question 9

Ways to control confounding:

Answer 9

1. Randomization
2. Restriction
3. Matching
4. Stratification
5. Multivariate Analysis

Question 10

Randomization

Answer 10

Allocates equal number of subjects with known and unknown confounders in each intervention group

Question 11

Randomization weaknesses

Answer 11

• Sample size may not be large enough to control ALL known and unknown confounders
• Doesn’t guarantee equal allocation for ALL confounders
• Interventional studied only

Question 12

Restriction

Answer 12

Restrict specific subjects

Question 13

Randomization Strengths

Answer 13

With large enough sample size, will likely serve purpose of making groups equal

Question 14

Restriction strengths

Answer 14

Straight forward, convenient, inexpensive

Question 15

Restriction weaknesses

Answer 15

• Difficult to enroll subjects
• If not narrow restriction may have residual confounders
• Eliminates evaluating levels of variable
• Generalizing

Question 16

Matching

Answer 16

Study subjects selected in matched pairs related to confounder to equally distribute confounder among study groups

Question 17

Matching Strengths

Answer 17

Intuitive; gives greater analytical efficiency

Question 18

Matching weaknesses

Answer 18

Difficult to accomplish
Can only control confounders in matching
Over-matching masks findings

Question 19

Stratification

Answer 19

Statistical analysis by evaluating association between the exposure disease within the various strata of confounding variable

Question 20

Stratification strengths

Answer 20

Straight forward and enhances understanding of data

Question 21

Stratification weaknesses

Answer 21

Impractical for simultaneous control of multiple confounders

Question 22

Multivariate Analysis

Answer 22

Mathematically factor out effects of confounding variable(s)

Question 23

Multivariate strengths

Answer 23

Simultaneous control of multiple confounders

Question 24

Multivariate wekanesses

Answer 24

Some people may not understand data analysis

Question 25

Confounding Variable

Answer 25

A 3rd variable that distorts an observed relationship between the exposure and the outcome