View source for 27. Confounders, effect modifiers and possibilities for elimination

* Confounding
** A confounding variable is a variable which has not been considered in the study, but which correlates with the risk factor and disease
** What the study observes: risk factor exposure –> disease
** What the reality is: risk factor exposure < –> confounding variable –> disease
*** There is no direct association between risk factor exposure and disease
** Example 1:
*** A study finds that people who drink coffee have increased risk for lung cancer, and therefore concludes that there is an association between coffee and lung cancer
*** However, many coffee-drinkers also smoke, and that it is the smoking which increases the risk for lung cancer
*** Smoking is the confounding variable here
** Example 2:
*** A study finds that children born later in the birth order have higher risk of Down syndrome, and therefore the study concludes that there is an association between late birth order and Down syndrome
*** However, the children who are born later in the birth order are often born by older mothers, and it is the maternal age which increases the risk for Down syndrome
*** Maternal age is the confounding variable here
** To prevent confounding
*** Perform multiple studies with different populations
*** Select comparable groups
*** Randomize study groups
*** Matching
*** Standardization
* Effect modifiers
** An effect modifier is a third variable which has different effect between study groups
*** This influences the study outcome
** Example 1
*** Tetracycline discolours teeth in children, but not in adults
*** Tetracycline has different effects between study groups
** Example 2
*** Hypertension is more likely to cause myocardial infarction in people with hypercholesterolaemia than people without
*** Hypertension has different effects between study groups

[[Category:Public health 2 - General epidemiology and demography]]