Bias and Confounding

Question 1

What does validity mean?

Answer 1

It relates to the absence of systematic error in a study result.

Question 2

What is bias?

Answer 2

The extent to which a measure of association from a study differs from the true measure of association in the source population.

Question 3

If the error in a study is repeatable, it is usually due to what?

Answer 3

Systematic error

Question 4

What is internal validity?

Answer 4

The study result is valid with respect to the population under study.

Question 5

What is external validity?

Answer 5

the study result is valid to a wider population, beyond the study.

Question 6

What does systematic error do to epidemiological studies?

Answer 6

It causes the measure of association estimated from the study population to be different from the true measure of association in the source population.

Question 7

What is a non-differential bias?

Answer 7

A bias that equally affects both groups, the exposed and non exposed.

Question 8

What is a differential bias?

Answer 8

a bias that affects one group more than another (the diseased, OR the non-diseased, not both)

Question 9

What are the 2 general sources for bias?

Answer 9

selection bias (error in selection of subjects) and information bias (error in measurement)

Question 10

What is a misclassification bias?

Answer 10

the same thing as information bias.

Question 11

What is a confounder?

Answer 11

an unknown factor that distorts the relationship between the exposure and the outcome.

Question 12

What is a problem with selecting convenience samples?

Answer 12

They lead to selection bias, as they are not representative of the entire population.

Question 13

How can you reduce selection bias?

Answer 13

ensure that controls/exposed/unexposed cases have similar demographics and collect additional info to ensure these similar demographics

Question 14

Aside from convenience sample, what is another way to get selection bias?

Answer 14

Self-selection bias, when studies are based upon volunteers, studies of working people or performance animals, who can be healthier than the population as a whole, or diagnostic bias.

Question 15

How can you reduce diagnostic bias?

Answer 15

By having a well-defined case definition and using as many objective parameters as possible. It is helpful to blind the exposure status of the animal as well.

Question 16

When can you get an information bias?

Answer 16

When you use a test with poor sensitivity, or when recall bias is involved (owners of sick pets are more likely to remember exposures than owners of healthy pets)

Question 17

What is another way to get information bias?

Answer 17

Observer variation. Different observers may interpret info differently.

Question 18

How can you minimize observer variation?

Answer 18

By using standard protocols and training the study personnel. You can also blind the observers to exposure and outcome status.

Question 19

What 3 conditions must a third factor meet to be a confounder?

Answer 19

1. Must be associated with the exposure
2. Must be associated with the outcome
3. Must not be in the causal pathway between the exposure and the outcome.

Question 20

In a study done where 508 vets were asked whether or not they wore lab coats, and were then tested for C. burnetii, in order to see if there was an association between wearing a lab coat and having Abs to C. burnetii, what is the confounding factor?

Answer 20

Whether you are a LA or SA vet. LA vets are less likely to wear lab coats overall, when compared to SA vets.

Question 21

What can you do before the study starts, to reduce confounding variables?

Answer 21

1. Match the study by selecting cases and controls so that the confounding factor is equally represented in both.
2. Restriction - do not enroll animals that have the confounding factor
3. Randomization - reduces confounding.

Question 22

What can you do after the study has been completed, if you notice a confounding variable?

Answer 22

Stratify. This partitions the results based on the confounding factor.