Module 6 Flashcards
Confounding
the mixing of the effect of the exposure with the effect of another factor (the confounder) which distorts the estimated measure of association
a confounder is a factor that can be
a risk factor for the disease of interest independent of the exposure
associated with the exposure of interest
not a consequence of the exposure
Failure to account for a confounder
wipe-out an association
create an invalid association
reverse the direction of association
change the degree of association
Causal pathway
X leads to Y which leads to Z
X = exposure variable
Y = variable in causal pathway
Z = disease variable
Strategies for handling confounding
randomization
restriction = limit entrance to subjects who fall in one category of the confounder
matching = only include controls with comparable characteristics to cases
multivariable regression analysis = add confounding variables into regression models
stratification = separate ORs or RRs by different levels of a confounder
Identifying confounders
literature search = suggests potential confounders for which data should be collected biologic plausibility this changes not all confounders known results unknowingly biased
Effect modification
when the magnitude of the effect of the primary exposure of an outcome differs depending on the level of a third variable
effect modification vs confounding
effect modification is a biological phenomenon in which the exposure has a different impact in different circumstances
confounding is a form of bias
Confounding graphs
parallel lines indicate confounding effect of exposure on disease is the same at different levels of the confounder stratum-specific ORs or RRs are equal assuming the association is linear ORs/RRs = 1 = lines are horizontal ORs/RRs > 1 = lines have an upward slope ORs/RRs < 1 = lines have downward slope
Effect modification graphs
non-parallel lines indicate effect modification
exposure-disease association differs at different levels of the effect modifier
negative association = downward sloping
positive association = upward sloping
Mantel-haenszel estimator
crude OR/RR falls outside range of stratum-specific ORs/RRs
Mantel-Haenszel estimator measures the average effect of exposure across all strata
advantage = easy to compute with a calculator
disadvantage = computation difficulty increases as tables get bigger
Confounder and effect modifier possibilities
a confounder = compare crude and Mantel-Haenszel
if the values are different, then it is a confounder
an effect modifier = compare stratum 1 and stratum 2
if the values are different, then it is an effect modifier
a confounder and an effect modifier
neither a confounder nor an effect modifier