5. Interaction

Question 1

Interaction (Definition)

Answer 1

Interaction refers to a situation in which the difference in risk/rate between those with and without a risk factor is not homogeneous among strata formed by a third factor.

When the observed joint effect of some risk factor and a third variable is different from the joint effect expected on the basis of putting together their independent effects.

Question 2

Types of Interaction

Answer 2

• Synergistic or Positive Interaction: If the presence of effect modifier potentiates or accentuates the effect of the exposure on the outcome. (Observed > Expected)
• Antagonistic or Negative Interaction: If the presence of effect modifier diminishes or eliminates the effect of the exposure on the outcome. (Observed < Expected)

1. Statistical Interaction
2. Biologic Interaction: Direct chemical or physical reactions in individuals. Departure from additivity implies some type of biologic interaction
3. Public Health Interaction: Focused on excess or reduction in the number of cases

Question 3

Mathematical Models of associations

Answer 3

Additive model: The effect can be measured by attributable risk

Multiplicative model: The effect can be measured by relative difference (ratio; relative risk)

Modification effects depend on the measure we choose ! sometimes no additive interaction but has multiplicative.

Question 4

Interpretation of interaction

Answer 4

“The effect of asbestos exposure on lung cancer is modified by cigarette smoking because the effect of asbestos on lung cancer is much stronger in smokers than in non-smokers”

Question 5

Additive Interaction

Answer 5

When the attributable risk in those exposed to some factor A, varies as across strata of a third variable Z

Question 6

Multiplicative Interaction

Answer 6

When the relative difference (ratio) in the risk of an outcome Y between exposed and unexposed to some factor A, differ (is heterogeneous) as a function of a third variable Z

Question 7

Compare confounding and interaction

Answer 7

Definition:

confounding: Confounding is a distortion (inaccuracy) in the estimated measure of association that occurs when the primary exposure of interest is mixed up with some other factor that is associated with the outcome.
interaction: Interaction refers to a situation in which the difference in risk/rate between those with and without a risk factor is not homogeneous among strata formed by a third factor.

1. Identification is different
2. Confounding factors are a “nuisance” and can account for all or part of an apparent association between an exposure and a disease. Confounding factors simply need to be eliminated to prevent distortion of results.

Effect Modification is not a “nuisance”, it in fact provides important information. The magnitude of the effect of an exposure on an outcome will vary according to the presence of a third factor.

3. Deal with it differently: try to eliminate confounding by using… ; report stratum-specific measurement for interaction
4. confounder needs to associated with exposure and outcome, while interaction only needs to be associated with the outcome.