Reading...
Play button
Play button
Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
76 Cards in this Set
- Front
- Back
|
Four types of selection bias
|
Self-Selection Bias, Healthy Worker Effect/Healthy volunteer effect, Loss to follow up, compensating bias/nondiferential bias
|
|
|
Definition of Selection Bias (self, healthy)
|
A systematic error resulting in differential probability of being selected into the study sample between study groups.
|
|
|
Why does selection bias occur?
|
Differences in characteristics between those who are selected and those who are not.
|
|
|
When does selection bias occur in case-control study?
|
Likelihood of selection is based on exposure status.
|
|
|
When does selection bias occur in a cohort study?
|
If likelihood of selection is based on outcome status.
|
|
|
Direction of selection bias
|
May be towards or away from null depending on example.
|
|
|
How to avoid selection bias
|
Use identical selection criteria to define the same/similar source population for cases and controls.
|
|
|
Why does self selection bias occur?
|
refusal, non response or agreement to participate that is related to exposure and disease
|
|
|
When does self-selection bias occur?
|
subjects in a particular exposure or disease category are more or less likely to participate than subjects in another category.
|
|
|
Direction of self-selection bias
|
direction of bias may be towards or away from null depending on example
|
|
|
How to avoid self-selection bias:
|
obtain high participation rates, recruit participants, use experimental design (randomly assigning screening)
|
|
|
Why does healthy worker effect/healthy volunteer effect occur?
|
sampling individuals who remain in the workforce, who tend to be healthier thus underestimating the mortality or morbidity related to occupational exposures; healthy volunteers may also affect cohort studies.
|
|
|
When does healthy worker effect occur?
|
When the general population, which consists of healthy and ill people, is selected for comparison to a relatively healthy working population which tend to be healthier therefore the rates of death and disease are lower than the general population
|
|
|
When does healthy volunteer effect occur?
|
When participants who volunteer for studies are also likely to be healthier than counterparts not participating.
|
|
|
How to avoid healthy worker effect?
|
Compare the workers in a specific job with workers in other jobs that differ in occupational exposure.
|
|
|
Definition of Loss to Follow-Up
|
A systematic error resulting in differential probability of being followed up during the study period.
|
|
|
Why does loss-to-follow up bias occur?
|
probability of having complete follow up information differs
|
|
|
Differential loss to follow up:
|
dependent on exposure or outcome
|
|
|
Non-differential loss to follow up:
|
independent of exposure or outcome
|
|
|
When does differential loss to follow up bias occur?
|
losses are unequally likely (more or less likely) to occur among exposed individuals who develop the disease (sicker) than among unexposed individuals who develop the disease (healthier)
|
|
|
When does non-differential loss to follow up bias occur?
|
losses of diseased individuals are equally likely among exposed and unexposed
|
|
|
Direction of Differential Loss to follow up
|
towards or away from the null dependent on example
|
|
|
Direction of non-differential loss to follow-up
|
absolute measures biased towards the null; relative measures unchanged
|
|
|
How to avoid loss to follow-up bias:
|
maximize retention, trace subjects, sensitivity analyses for a given analytic question
|
|
|
Definition of compensating bias (non-differential bias)
|
Bias equally affects the selection of comparison groups (cases vs. controls or exposed vs. non-exposed)
|
|
|
Why does compensating bias/nondifferential bias occur in case-control or cohort studies?
|
The magnitude of the bias equally affects selection of cases and controls in case-control study or of exposed and unexposed participants in a cohort study.
|
|
|
When does compensating bias/nondifferential bias occur?
|
Case-control studies often attempt to use this bias to balance bias in selecting cases and controls.
|
|
|
Example of using compensating bias in case-control study
|
Selecting both cases and controls from screening studies since both would be subject to self-selection bias and since both cases and controls would be subject to that self-selection bias, it would compensate for each other
|
|
|
Direction of compensating/nondifferential bias
|
biased absolute measures but unbiased relative measures
|
|
|
How to avoid compensating/non-differential bias
|
avoid selection bias
|
|
|
When to plan for compensating/non-differential bias
|
Use as in nested case-control study where cases and controls are selected from a cohort of participants who agreed to participate (healthy participant/volunteer bias)
|
|
|
7 types of Information/Measurement Bias
|
Recall Bias, Respondent Bias, Interviewer Bias, Observer Bias, Differential Misclassification, Non-differential Misclassification, Medical Surveillance/Surveillance/Detection Bias
|
|
|
Definition of Recall bias
|
Systematic error resulting in differential probability of individuals' placement into exposure or outcome categories.
|
|
|
What type of study is recall bias usually seen?
|
Case-control
|
|
|
Why does recall bias occur
|
differential accuracy in the information provided by comparison groups related to exposure or outcome
|
|
|
When does recall bias occur?
|
in a case-control study if cases are more or less likely than controls to recall and report prior exposures
|
|
|
Direction of recall bias
|
away or towards the null ; determined by which group has less accurate recall
|
|
|
How to avoid recall bias
|
validate exposure or disease data, use a control group with different disease, design more complete questionnaire, use objective measurements (biological), change study design to obtain exposure information from the outset ( change to cohort, nested case-control or case-cohort study)
|
|
|
What type of study is respondent bias usually seen?
|
Cohort
|
|
|
Definition of respondent bias:
|
systematic error resulting in differential probability of individuals' placement into exposure or outcome categories
|
|
|
Why does respondent bias occur?
|
differential accuracy in the information provided by participants related to the exposure or outcome (over-reporting or under-reporting)
|
|
|
When does respondent bias occur?
|
In a cohort study if exposed subjects are more or less likely tun unexposed subjects to recall and report subsequent disease or exposures.
|
|
|
Direction of respondent bias
|
Towards or away from null. Determined by which group over- or under-reports disease or exposures.
|
|
|
How to avoid respondent bias:
|
confirm self-report with objective measures, use standardized and validated questionnaires, collect data on related symptoms and outcomes, self-administered or computer assisted questionnaires important for sensitive topics.
|
|
|
Definition of interviewer bias
|
occurs in an exposure assessment; a systematic error resulting in differential probability of individuals' placement into exposure or outcome categories
|
|
|
Why does interviewer/observer bias occur?
|
Differential intensity in the ascertainment of exposure or disease by the interviewers or observers.
|
|
|
When does interviewer bias occur?
|
In case-control or cohort study if an interviewer tries harder to assess exposures in a certain category of participants rather than others. (ex. more likely to ask follow-up questions if an African-American participant does not report hypertension.
|
|
|
When does observer bias occur?
|
In case-control or cohort study if disease diagnosis is more likely for certain category participants (ex. more likely to diagnose renal failure in African American participants)
|
|
|
Direction of interviewer/observer bias:
|
Towards or away from null; determined by which cell has increased ascertainment (more people inappropriately placed in group)
|
|
|
How to avoid interviewer/observer bias:
|
masking in regard to disease status (case-control) or exposure status (cohort), standardized/validated questionnaires, well-trained observers, quality control for data collection including written protocol, monitoring and reliability and validity sub studies
|
|
|
Definition of Differential Misclassification in Case Control Study
|
A systematic error resulting in: differential probability of individuals' placement into exposure categories dependent upon outcome status
|
|
|
Definition of Differential Misclassification in Cohort Study
|
A systematic error resulting in differential probability of individuals placement into outcome categories dependent upon exposure status
|
|
|
Why does Differential Misclassification occur?
|
Due to differential identification and categorization of exposure or outcome
|
|
|
Differential Misclassification in case-control study:
|
occurs during exposure classification with unequal likelihood when unequal proportions of cases and controls are assigned to inappropriate exposure classifications
|
|
|
Differential Misclassification in cohort study:
|
occurs during outcome classification with unequal likelihood when unequal proportions of each exposure group are assigned to inappropriate outcome classifications (ex. missed diagnosis)
|
|
|
Direction of differential misclassification
|
towards or away from null dependent upon example
|
|
|
How to avoid differential misclassification
|
improve accuracy of collected data by taking multiple measurements of exposure and disease, examine validity and reliability of measurement tools, consider revised exposure and disease definitions or categorizations, sensitivity analysis to attain plausible estimates
|
|
|
Definition of Non-differential Misclassification in Case Control
|
A systematic error resulting in differential probability of individuals' placement into exposure categories independent of (unassociated with) outcome status
|
|
|
Definition of Non-differential Misclassification in Cohort study
|
A systematic error resulting in differential probability of individuals' placement into outcome categories independent of (unassociated with) exposure status
|
|
|
Why does non-differential misclassification occur?
|
due to differential identification and categorization of exposure and outcome
|
|
|
Non-differential misclassification in case-control:
|
if error occurs in exposure classification, it occurs among diseased and non-diseased with equal likelihood
|
|
|
Non-differential misclassification in cohort:
|
If error in outcome classification, it occurs among exposed and non-exposed with equal likelihood
|
|
|
When does non-differential classification occur in a case-control study with 2 exposures:
|
when equal proportions of cases and controls are assigned to inappropriate exposure classifications (yes/no)
|
|
|
When does non-differential classification occur in a case control study with more than 2 exposures?
|
When the same proportion of cases and same proportion of controls are assigned to inappropriate exposure classifications (none, low, high)
|
|
|
When does non-differential classification occur in cohort study?
|
May occur in cohort study when equal proportions of each exposure group are assigned to inappropriate outcome classifications (missed diagnosis)
|
|
|
What is the direction of the bias in non-differential misclassification for 2 exposures?
|
Towards the null
|
|
|
What is the direction of the bias in non-differential misclassification for more than 2 exposures?
|
towards or away from the null
|
|
|
How to avoid non-differential misclassification bias?
|
improve accuracy of collected data by taking multiple measurements of exposure and disease, choose measurement tools with optimal validity and reliability, consider revised exposure and disease definitions or categorizations, sensitivity analysis to attain plausible estimates
|
|
|
Definition of surveillance/medical surveillance/detection bias
|
a systematic error caused by closer surveillance of exposed or unexposed for a disease outcome
|
|
|
Why does surveillance bias occur?
|
when an exposure (medical or non-medical) causes a closer surveillance for study outcomes that increases outcome detection in exposed individuals
|
|
|
When does surveillance bias occur?
|
When outcome is ascertained through regular healthcare channels (electronic health records); when the exposure is a medical condition or therapy that leads to frequent or thorough checkups and the outcome has a high proportion of subclinical cases (e.g. diabetes) which is more likely to be detected by individuals who have regular checkups
|
|
|
How does surveillance bias act in case-control study?
|
As a selection bias, if cases are more likely to be identified and therefore selected into study if they are exposed
|
|
|
How does surveillance bias act in a cohort study?
|
tends to act as information bias if exposed individuals are more thoroughly examined than unexposed individuals
|
|
|
Direction of surveillance bias
|
Could bias towards or away from null but will bias away from null if it increases the outcome detection in exposed individuals
|
|
|
How to avoid surveillance bias
|
systematically assess outcome (e.g. screening) regardless of exposure, mask exposure status in outcome ascertainment
|
|
|
How to avoid surveillance bias in case-control study
|
perform sensitivity analyses to compare exposed vs. unexposed by frequency, intensity, quality of medical care, awareness of health problems, disease severity at diagnosis, SES and education
|
