Bias,Confounding. Flashcards
What is loss to follow-up bias
- Individuals who are lost to follow-up may be different than those who remain under observation.
- Important consideration in prospective cohort studies and randomized trials.
- Even a small percentage of losses, if related to the exposure and outcome, can threaten validity of observed associations
What is reporting bias?
Study participants may be reluctant to truthfully report an exposure that carries socio-cultural meaning. (Social desirability bias)
T/F: Recall bias is the major bias in Case-Control study.
True
What are the two major types of errors and their definitions?
Random Error: positive and negative fluctuations around the true value.
Systematic Error: a deviation from the true value (Bias)
T/F: Identification and control of error is critical in epidemiologic studies
True
Validity: Goal, Definition, Categories.
- *Goal:** to ascribe any observed association as truly the effect of exposure
- *Definition:** the degree to which an observation is free of error
Categories: Internal/External
What is information bias
systematic differences in the accuracy of data on exposures and/or outcomes between study groups
Potential for bias selecting into study groups
T/F: Selection bias is one major source of bias for cohort studies.
False. It’s not because in cohort studies, exposure comes before the disease.
Bias: two major types
selection bias
information bias
What are the 3 types of information bias
- Recall bias
- Observer bias
- Reporting bias
What is the primary concern of information bias
misclassification on exposure or disease status
- nondifferential
- differential
Presence and direction of bias depends on the pattern of misclassification; requires exploratory data analyses
Bias: definition, where it results from?
Def: systematic deviation of an observed association from truth
Bias results from systematically higher or lower counts of exposure (case-control) or incident disease (prospective study) in the numerator and/or denominator because of:
- the process used to select study groups
- differences in measurement accuracy between groups
- confounding*
What is recall bias?
It results from a difference in the ability to recall past exposure between cases (who are ill) and controls
What is internal validity and what can influence it?
Accuracy of observations made within a study.
influenced by:
(1) selection of study groups
(2) measurement error
What is observer bias?
When exposure or outcome status is known to the observer, assignment to outcome or exposure (yes/no) may be biased
What is the response (participation) bias
If differences in characteristics between responders (participants) and nonresponders (nonparticipants) are correlated with exposure (or outcome), a biased measure of association could result
What is Healthy Worker Effect and where does it apply
- Def: Bias manifesting from selection of an external comparison (nonexposed) group that is less healthy than the exposed group
- Traditionally seen as an issue in occupational cohort studies ( In general, people in the workforce tend to exhibit lower death rates than the general population)
- Also applies to non-occupational cohort study designs that include an external comparison group
What are the possible solutions to recall bias (2)
Validate case and control responses (sub-studies)
Utilize cohort study or nested case-control study design
T/F: Measurement error and misclassification are bias.
False. They are NOT bias, although they can contribute to association measures (OR or RR).
Selection of controls may introduce bias
What is external validity and what can influence it?
Accuracy of study observations in other groups. (generalizability)
influenced by:
(1) representativeness of study participants
(2) measurement error
What’s the influence, source(s) of systematic error?
Threatens accuracy (validity)
Sources:
- Selection of study groups
- Measuring information
- Confounding* (Mike doesn’t agree)
What are the possible solutions to interviewer bias (3)
- Data gatherers blinded to group status
- Standardize case ascertainment & interview procedures
- Monitor quality control during interviews (taping)
T/F: For bias to be present, generally the error must be correlated with either the exposure or the outcome (need to look at the data)
True.
Selection bias: Definiton, concerns in ca-co and cohort study
When the process of selecting cases or controls (case-control study) or, exposed or nonexposed (cohort study) results in an obscuring of the true association between the exposure and outcome.
Ca-Co: Is selection of cases and/or controls correlated with exposure status?
Cohort study: Is selection into exposed or nonexposed groups correlated with disease incidence?
T/F: Bias threatens the external validity of study findings
False. It threatens the internal validity.
What’s the influence, source(s) of random error?
Threatens precision (reliability) e.g. CI’s range
Source: Sampling error, measurement variability (inconsistency)
* Larger sample sizes reduce influence
4 sources of selection bias
- selecting study sample into study groups
- selecting the controls
- Cohort Study loss to follow-up
- healthy worker effect
