Epid 5310-Lecture 5

Front 1

A confounder must fulfill 3 conditions

Back 1

1. A risk factor for the disease in the unexposed
2. Associated with exposure in the population from which the cases arose
3. Not an intermediate step in the causal pathway between exposure and disease

Front 2

1. Is it a risk factor for disease?

Back 2

-Confounder must be a risk factor for the disease INDEPENDENT of any association with exposure
-Prior knowledge of disease risk factors should guide identification of confounders
-When prior knowledge is inadequate, study data may serve as a guide to the relation between a potential confounder and disease

**You find out if something is causally related because you do a literature review.

Front 3

2. Is it associated with exposure in the population from which cases arose?
In Cohort Study

Back 3

Cohort study
-a confounder must be associated with exposure among subjects at start of follow-up
-Relation between potential confounder and exposure can be evaluated from study data prior information from other populations is not relevant

Front 4

2. Is it associated with exposure in the population from which cases arose?
Case-control study

Back 4

In a case control study:
-A confounder must be associated with exposure in the population form which cases arose
-The relation between a potential confounder and exposure should be assessed ideally in the target population, but this is seldom possible.
-If the control series is large and represents the target population (no selection bias), it will provide a reasonable estimate of the association between the confounder and exposure.

Front 5

2. Is it associated with exposure in the population from which cases arose?
Randomized experiments

Back 5

-Confounding can occur in studies that randomly allocate exposure, though to a lesser extent than in nonrandomized studies.
-Randomization does not guarantee no association between exposure and extraneous factors
-It is a probabilistic procedure that can leave some association between exposure and extraneous factors, esp. if sample is small.
-BUT- confounding tends to be negligible in very LARGE well-conducted randomized studies.

Front 6

3. Is it an intermediate step in the causal pathway between exposure and disease?

Back 6

-A factor that represents an intermediate step in the exposure-disease pathway is an intervening variable or mediator
-Adjusting a relative risk estimate for an intervening variable creates bias by removing the exposure effect that operates through the intervening variable.

** if you try and adjust for the intervening variable, you will have a biased measure of association.

Front 7

3. Is it an intermediate step in the causal pathway between exposure and disease?

Back 7

- An intervening variable cannot be distinguished statistically form a confounder
-Knowledge of biological mechanisms must be used to distinguish intervening variables from confounders

Front 8

DAGs.
The process of looking at a confounder

Back 8

Is the confounding variable causally and non-causally related to the exposure?
Causally related to the outcome?
and Not on the causal pathway?

Front 9

Checking to see if a factor is a confounder.
PROCESS

Back 9

Step1: Find the crude measure of association and verify that the confounder is associated with the exposure and the outcome
Step2: Stratify the confounder. If the stratum are similar to each other and different than the crude then we know it is a confounder.
Step 3: Again when there is a confounding effect, the association seen across the strata potential confounder are of similar magnitude to each other and are different than the crude value.

Front 10

types of confounding

Back 10

Positive
Negative
Qualitative

Front 11

Positive

Back 11

Confounder overestimates true strength of association

Front 12

Negative

Back 12

Confounder underestimates true association

Front 13

Qualitative

Back 13

Confounder results in an inversion of the direction of the association

Front 14

Methods to control for confounding in the design

Back 14

Randomization
restriction
Matching (so that the distribution is the same)

**Methods at design phase are preferable.

Front 15

Methods to control for confounding in the analysis

Back 15

Stratified Analysis
Multivariate analysis

Front 16

Adjusting for confounding

Back 16

Idea is to use some statistical model to estimate what the association between exposure and outcome would be, given a constant value or level of the suspected confounding variable.

Front 17

Prevention and confounded associations

Back 17

Primary prevention: Prevention or cessation of risk factor exposure
-requires causal association (i.e., No confounding) in order to affect disease outcome.

Secondary prevention: screening high risk individuals
-sufficient that there be a statistical association between risk factor and disease (i.e., association can be confounded) in order to identify these high risk individuals.

Front 18

Primary prevention

Back 18

prevention or cessation of risk factor exposure
e.g., saturated fat intake and atherosclerosis.

Causal association must be present: otherwise, intervention on risk factor will not affect disease outcome.

Front 19

Secondary prevention

Back 19

Early diagnosis via selective screening of "high-risk" subjects

Association may be either causal or statistical: that is the association may be confounded. For example, African American have higher hypertension than white because of confounding variables but nevertheless they have higher hypertension and are at high risk.