Chapters 10-14

Question 1

What is bias?

Answer 1

a systematic error in the design or conduct of study that leads to an erroneous association between the exposure and disease

Question 2

What is confounding?

Answer 2

The mixing of effects between the exposure, the disease, and a third variable that is termed a confounder. Like bias, confounding distorts the relationship between an exposure and disease

Question 3

What is random error?

Answer 3

The probability that the observed result is due to “chance,” an uncontrollable force that seems to have no assignable cause.

Question 4

What are key factors of bias?

Answer 4

an alternative explanation for an association

Usually does not result from a “prejudiced” investigator

Can pull an association either toward or away from the null

Amount of bias can be small. moderate, or large

Is avoided when the study is carefully designed and conducted

Question 5

What is selection bias?

Answer 5

An error resulting from subject selection and participation. Association among study subjects differs from association among eligible source population.

Question 6

What are some examples of selection bias in a cohort study?

Answer 6

loss to follow up
the “healthy worker” effect
self selection bias
differential surveillance, diagnosis, or referral of subjects

Question 7

What are key factors of selection bias?

Answer 7

It is more likely to occur in case control and retrospective cohort studies (both exposure and disease have occurred by the time of subject selection)

It can also occur in experimental studies and prospective studies due to loss to follow up

Little can be done to fix this bias once it occurs

It can pull an association either towards or away from the null

Question 8

What is Information Bias?

Answer 8

a flaw in measuring exposure, covariate, or outcome variables that results in different quality (accuracy) of information between comparison groups.”

Question 9

What is recall bias?

Answer 9

This occurs when there is a differential level of accuracy in the information provided by compared groups.

Question 10

What is interviewer bias?

Answer 10

a systematic difference in soliciting, recording, or interpreting information that occurs in studies using in person or telephone interviews.

Question 11

What is misclassification?

Answer 11

A measurement error

There is an error in the classification of the exposure or the disease (essentially a clerical error)

Question 12

What are characteristics of information bias?

Answer 12

Results in participants who are incorrectly classified as either exposed or unexposed or as diseased or not diseased

Occurs after subjects have entered a study

Occurs in both retrospective and prospective studies

Specific types include recall bias, interviewer bias, and differential / nondifferential misclassification

Methods for prevention include masking / double masking (like blind and double blind), using multiple measurements, specific criteria, and detailed questionnaires / surveys=

Question 13

What is differential / non-differential misclassification?

Answer 13

Non-differential misclassification occurs when the probability of individuals being misclassified is equal across all groups in the study. Differential misclassification occurs when the probability of being misclassified differs between groups in a study (Porta et al. 2014).

Question 14

What is an example of a confounding variable?

Answer 14

The association between Diabetes and Dementia; those in a certain study were an older population. The RR of Diabetes leading to dementia was determined to be about 3.5

When another study was conducted with “control” of the confounding aspect (age,) risk of diabetes leading to dementia was revealed as closer to 2.

Question 15

What criteria does a variable need to meet in order to be confounding?

Answer 15

It must be associated with the exposure in the population that produced cases (it most be more or less common in the exposed group than the control group). The variable must be an independent cause or predictor of the disease (association between confounder and disease is present among both exposed and unexposed individuals).

It also cannot be an intermediary step in the causal pathway between exposure and disease.

Question 16

How do we calculate the magnitude of a confounding variable?

Answer 16

RR crude - RR adjusted / RR adjusted

Question 17

What are methods for controlling confounding variables in design and analysis within a study?

Answer 17

Design stage: randomization, restriction, and matching

Analysis Stage: standardization, stratified analysis, matched analysis (needed for case control only), multivariate analysis

Question 18

What is residual confounding?

Answer 18

a term for the confounding that remains even after many confounding variables have been controlled. Residual confounding can come from several sources (inability to collect needed data, categories too large (e.g. 10 year age groups with many differences within this category), misclassification, various uncontrollable factors

Question 19

What is precision?

Answer 19

Lack of random error
Most epidemiological methods are not exact, leading to the need for the term “measurement error” (or standard error?)

Question 20

What is statistical inference?

Answer 20

the process of making generalizations from a sample to the source or parent population (based on the mathematical theory of probability)

Question 21

What is the null hypothesis versus the alternative hypothesis?

Answer 21

The null hypothesis is often stated as the assumption that there is no change, no difference between two groups, or no relationship between two variables. The alternative hypothesis, on the other hand, is the statement that there is a change, difference, or relationship

Question 22

What are characteristics of p values?

Answer 22

Can be one or two sided (e.g. upper value is probability of observing RR greater than the expected value, the lower value is the probability of observing RR smaller than the expected value).

A continuous statistic ranging from 0 to 1; a small p value indicates a low degree of compatibility between the observed data and the null hypothesis (there is a very small chance this result would have generated, indicating a difference within the population )

Question 23

How do the terms “(not) statistically significant” relate to the null hypothesis?

Answer 23

If results are not statistically significant, this means the data failed to provide sufficient evidence to doubt the null hypothesis.

Question 24

What are alpha and beta errors?

Answer 24

Alpha errors refer to errors which would “prove” null hypothesis is not correct,

whereas Beta errors refer to errors which would “prove” alternative hypothesis is not correct

Question 25

What is the strict statistical definition of a 95% confidence interval?

Answer 25

if a study were repeated 100 times and 100 point estimates and 100 confidence intervals were calculated, 95 out of 100 confidence intervals would contain the true measure of association

Question 26

What are the 3 probability distributions most typically used in epidemiological research?

Answer 26

normal distribution
Poisson distribution
binomial distribution (including Bernoulli trials)

Question 27

When is the Poisson Distribution used?

Answer 27

this is a discrete distribution commonly used to describe rare events

Question 28

What are epidemiological methods used for hypothesis testing?

Answer 28

the T test
Chi Square
Z test

Question 29

What are absolute versus relative measures of comparison?

Answer 29

Risk Difference (and its Confidence Intervals) is absolute

Relative Risk / Odds Ratio (and their Confidence Intervals) are relative

Question 30

What are the two main sources of random error?

Answer 30

Measurement error
Sampling variability

Question 31

What is internal validity?

Answer 31

The extent to which the results of a study accurately reflect the true situation in the study population

Question 32

What is external validity?

Answer 32

The extent to which the results of the study are applicable to other population

Question 33

True or False, If a study lacks internal validity, it can still have external validity.

Answer 33

FALSE (if a study lacks internal validity, external validity is not even examined)

Question 34

What is effect measure modification?

Answer 34

the strength of the association between an exposure and disease differs according to the level of another variable (effect modifier)

Question 35

What does homogeneity effect mean?

Answer 35

effect modification is absent

Question 36

What does heterogeneity / effect modification mean?

Answer 36

change in association according to the 3rd variable

(dependent upon ratio measure or difference measure)

Question 37

How do confounding variables and effect measure modification differ?

Answer 37

confounding variables are considered a nuisance

effect measure modification is an interesting phenomenon that epidemiologists aim to describe. Its presence depends on a particular measure of association

It is measured in detail via stratified analyses, through visual inspection, statistical tests, or evaluation of excess risks

Question 38

What are synergy and antagonism in terms of effect modification?

Answer 38

synergy occurs when two factors work together to produce more disease than one would expect based on the action of either factor alone

antagonism occurs when one factor reduces the impact or even cancels out the effect of another factor

Question 39

Describe criteria for effectively evaluating epidemiological studies

Answer 39

Question 40

Describe the decision making process for evaluating epidemiological data (especially for confounding variables and effect modifiers)

Answer 40