Exam 2: Confounding & Effect Modification

Question 1

Confounding Variable (aka Confounder)

Answer 1

A 3rd variable that distorts an observed relationship (association; RR/OR/HR) between the exposure and the outcome (disease)

Question 2

To be a confounder, a 3rd variable must be:

Answer 2

Associated (related/correlated) with the exposure and the outcome of interest, yet independent of both. ***

But not directly in the hypothesized causal-pathway between the exposure and the outcome.

Question 3

Confounding can also be explained by a mixing of effects

Answer 3

an association (between exposure and outcome) is DISTORTED due to them being mixed with another (3rd) factor which is associated with the outcome of interest…. at the same time.

Question 4

How do you explain confounding as a confusion of effects?

Answer 4

The effect of the exposure is DISTORTED b/c the effect of an extraneous (3rd) factor is mistaken for the effect of the expousre.

Question 5

T/F, a confounder can only over estimate an association while not effecting direction of an effect?

Answer 5

False: A confounder can OVER or UNDER-estimate an association (RR/OR/HR) and can even change the apparent direction of an effect.

Question 6

What are the two main impacts of confounders?

Answer 6

1. Intensity/Magnitude/Strength
   - Produces an estiamte that is more or less extreme than the true association
2. Direction
   - Produces an estimate that moves the true association in a positive or negative direction
   - -Towards or away from a null association (RR/OR/HR=1.0)

Question 7

Step 1 of testing for confounding

Answer 7

Calculate crude outcome measure of association (OR/RR/HR)

- commonly called “unadjusted” association

Question 8

Step 2 of testing for confounding

Answer 8

Re-calculate outcome measure of association (OR/RR/HR) while mathematically (statistically) controlling for confounder

• removing effects of confounder from association measurement between exposure and outcome of interest
• commonly called “adjusted” association

Question 9

Step 3 of testing for confounding

Answer 9

Compare the two measures of association (steps 1 &2)

- the point estimate (RR/OR/HR) of the association will be different by >15% if there is confounding present

Question 10

What is the purpose of controlling for confounders?

Answer 10

To get a more precise (accurate) estimate of the true association between the exposure and the disease/outcome.

Question 11

What are the two ways to control confounding

Answer 11

1. Study design stage
   - randomization, restriction, matching
2. Analysis of data stage
   - stratification, multivariable or matched statistical analysis

Question 12

Controlling for confounding with randomization:

Answer 12

technique hopefully allocates an equal number of subjects with the known (and unknown) confounders into each intervention group

Question 13

What are the strengths of randomization in controlling confounding?

Answer 13

with sufficient sample size (N), randomization will likely be successful in serving this purpose (making groups “equal”)

Question 14

What are weaknesses with randomization in controlling confounding?

Answer 14

• Sample size (N) may not be large enough to control for all known and unknown confounders
• randomization process doesn’t guarantee successful, equal allocation between all intervention groups for all known and unknown confounders
• Practical on for interventional studies

Question 15

What is restriction in controlling for confounding?

Answer 15

study participation is restricted to only subjects who do not fall within pre-specified category(-ies) of confounder.

Question 16

What are the strengths of restriction in controlling confounders?

Answer 16

• straight forward, convenient and inexpensive

- does not negatively impact internal validity.

Question 17

What are weaknesses of restriction on controlling for confounding?

Answer 17

• sufficiently narrow restriction criteria may negatively impact ability to enroll subjects (reduced sample size N)
• if restriction criteria is not sufficiently narrow it will allow the introduction of residual confounding effects
• eliminated researchers ability to evaluate varying levels of the factor being excluded.
• Can negatively impact external validity (generalizability)

Question 18

What is matching in controlling for confounding?

Answer 18

Study subjects selected in matched-pairs related to the confounding variable to equally distribute confounder among each of the study groups.

Question 19

What is a strength of matching in controlling for confounding?

Answer 19

intuitive, some feel it gives greater analytic efficiency.

Question 20

What is a strength of matching in controlling for confounding?

Answer 20

intuitive, some feel it gives greater analytic efficiency

Question 21

What are weaknesses of matching in controlling for confounding?

Answer 21

• difficult to accomplish, very time consuming and expensive.
• doesn’t control for any confounders other than those matched on
• • over-matching possible; this will mask findings.

Question 22

What is stratification in controlling for confounding?

Answer 22

statistical analysis of the data by evaluating the association between the exposure and disease within the various strata (layers) within the confounding variable(s)

Question 23

What are the strength of stratification in controlling for confounding?

Answer 23

intuitive (to some), straight-forward and enhances understanding of the data.

Question 24

What is the weakness in stratification in controlling for confounding?

Answer 24

Impractical for simultaneous control of multiple confounders, especially those with multiple strata within each variable being controlled.

Question 25

What is multivariate analysis in controlling for confounding?

Answer 25

statistical analysis of the data by mathematically factoring out the effects of the confounding variable(s)

Question 26

What are strengths of multivariate analysis in controlling for confounding?

Answer 26

• can simultaneously control for multiple confounding variables
• in logistic regression, beta coefficients can be directly converted to OR’s

Question 27

What are weaknesses with multivariate analysis in controlling for confounding?

Answer 27

• process can cause some individuals to not clearly understand(interpret) the data (results)
• examples:
• -linear & logistic Regressions
• • Cox proportional hazards modeling.

Question 28

What is effect modification (interaction)?

Answer 28

a 3rd variable, that when present, modifies the magnitude of effect of an association by varying it within different levels of a 3rd variable (effect modifier)

Question 29

t/f unlike confounding, an effect modifying variable should be described and reported at each level of the variable, rather than controlled for

Answer 29

true

Question 30

How is testing for effect modification different from confounding?

Answer 30

The stratum-specific estimates are compared directly to see if they are different
- The point estimate (RR/OR/HR) for the association will be different by 15% between the lowest and highest strata (layers) if there is effect modification (interaction) present.