Chapter 8

Question 1

What is confounding

Answer 1

A mixing or muddling of effects that occurs when the relationship of interest is confused by the effect of something else.

Question 2

Which types of studies have the largest probelms for confounfing

Answer 2

Observational study desgins. Cohort, case-control, cross-sectional. It is important for the researchers to control for confounding in these study designs.

Question 3

3 criteria that msut be met for something to be confounding

Answer 3

1. It must be a risk factor for the outcome of interrest
2. It must be associated with the exposure of interest
3. It must NOT be intermediary between the exposure and the outcome (it must not lie on the CAUSAL PATHWAY)

Question 4

What are the effects of confounding

Answer 4

When confounding is present, the observed relationship will not reflect the true, causal relationship.
1. The researcher may observed type 1 error (exposure related to outcome but not actually)
2. Type 2 error (they DON’T see exposure related to outcome when actually is)
3. The size of effect estimate is over or underestiamte of true effect of exposure on outcome.

Question 5

Control of confounding through study design. 1,2,3,

Answer 5

Randomization - only possible for intervention studies
Restriction - Restrict the study sample to people either with or without the confounding characteristic.
Matching - Match the particpants on the presence or absence of the confounding factors

Question 6

Control of confounding during statistical analysis

Answer 6

1) Stratified analysis - Can only be done for single confounder. Divided total sample into smaller groups based on specific characteristics.
2) Statistical modeling - Can be used for one or multiple confounders. Outcome is simultaneously predicted by the exposure and confounder(s). Mathematically control for the effect.

Question 7

What is residual confounding

Answer 7

When there is still some leftover or “residual” confounding becuase a study did not fully accoutn for the influence of the confounders.

Question 8

Why might Residual confounding occur?

Answer 8

• The confounder was not controlled during the study desgin or statistical analysis phase
• The confounder was poorly measured

Question 9

Increasing sample sizes and counding. What happens with observational studies and for intervention studies?

Answer 9

-Increasing sample size will NOT address for confounding in observational studies.
-Intervention studies that use randomization, the bigger the sample, the more likely that possible confounders will be balanced across the treatment