Confounding Factors (Cut off Exam 2)

Question 1

Confounding

Answer 1

• Extraneous factor that wholly or partially accounts for the observed effect
• Distortion of the association that occurs due to the effect of an extraneous factor

Question 2

Confounding

Answer 2

Distortion of association due to a third variable

Question 3

Confounder

Answer 3

• Variable causing the confounding

- Associated with BOTH exposure and outcome

Question 4

Criteria for being a Confounder

Answer 4

1. Risk factor for the outcome
2. Associated with the exposure of interest
3. Not an intermediate step between exposure and outcome

Question 5

Covariate

Answer 5

• Potential confounder

- Select a list of these in your DB and determine whether they are confounders or not

Question 6

Risk Factor

Answer 6

Variable positively associated with the outcome

Question 7

Confounding by Indication

Answer 7

• Reason of prescription
• Underlying disease severity
• Worst prognosis are allocated preferentially to a particular treatment

Question 8

Confounding by Co-medication

Answer 8

• Difficulty to isolate the effect of a specific drug

- Compliancy increasing their likelihood to outside interventions and improve their outcomes

Question 9

Confounding by Dose/Drug Potency

Answer 9

• Analysis for a class or medications vs individual medications
• Dose-response (quantity/frequency)
• Potency

Question 10

Controlling for Confounding: Study Design

Answer 10

1. Restriction
2. Matching
3. Randomization: intervention studies only

Question 11

Controlling for Confounding: Analysis

Answer 11

1. Stratification

2. Multivariate Analysis

Question 12

Restriction

Answer 12

• Restrict the study population to only one level of the confounding factor
• Subjects have the same level of a confounding factor
• Including only women or people in a certain age range for example

Question 13

Restriction Advantages

Answer 13

• Simple

- Effective

Question 14

Restriction Disadvantages

Answer 14

• Generalizability of results is limited
• Residual confounding if do not restrict too narrow
• Cannot evaluate the effects of factors that have been restricted for
• Reduces the number of eligible subjects (sample size problem)

Question 15

Matching

Answer 15

• Ensure that study groups do not vary with respect to possible confounders
• Make groups artificially similar for potential confounders
• EX: match someone of the same age in each group

Question 16

Matching Advantages

Answer 16

• Simple - when few confounders

- Efficient

Question 17

Matching Disadvantages

Answer 17

• Very difficult if there are several confounders
• Limits sample size
• Cannot evaluate the effect of matched factors
• Required a special type of statistical analysis: McNemar, conditional logistic regression

Question 18

Randomization

Answer 18

• Random distribution into study groups

- Confounders given “equal chance” to be in either treatment group

Question 19

Randomization Advantage

Answer 19

• Very efficient
• Both known and unknown confounders are distributed equally
• Cannot be achieved with other techniques

Question 20

Randomization Disadvantages

Answer 20

-Only feasible with interventional studies

Question 21

Stratification

Answer 21

• Technique to control for confounding in the analysis of a study
• Evaluation of the association within homogeneous categories (strata) fo the confounding variable
• Form of post hoc restriction (during analysis)
• First derive statum-specific estimates and then calculate their weighted average

Question 22

Stratification Advantages

Answer 22

• Easy to carry out
• Produces a single summary estimate that is unconfounded can be calculated
• Gives a fairly good picture of what’s going on

Question 23

Stratification Disadvantages

Answer 23

• Inability to control simultaneously for multiple confounders
• Stratification may require a larger sample size

Question 24

Multivariable Analysis

Answer 24

• Analytical technique that adjust for several variables simultaneously
• Efficient estimation of measures of association while controlling for a number of confounders
• Involves the construction of a mathematical model to describe most efficiently the association between exposure and disease and other factors
• Most common techniques: multiple linear regression, logistic regression, ANCOVA, Cox regression

Question 25

Multivariate Analysis Advantages

Answer 25

• Very powerful technique

- Allows to control for multiple confounders in the same model (even when stratification would fail)

Question 26

Multivariate Analysis Disadvantages

Answer 26

• Requires sophisticated skills in biostatistics and epidemiology
• Potential to over-adjust