Lecture 26: Information Bias Flashcards
Define bias
any systematic error in an epidemiological study that results in an incorrect estimate of the association between exposure and risk of disease
What are the two ways that data is collected in a study? What is the problems with both of these two methods?
- by participants
- collected or measured by someone else
errors in measurement can occur in either method
How can measurement error occur? Give an example
- participants provide inaccurate responses eg. they may forget past exposures
- data is collected incorrectly/inaccurately eg. the person collecting the data may not follow the same procedure for all participants
What is the difference between subjective and objective reporting and what are the problems with these?
self reporting is subjective and it is susceptible to error because it is influenced by a person’s opinions
when data is collected by someone else, it is objective which is good because it is based on a measurement value
What is measurement error and what does it include?
measurement error is the difference between a measured quantity and its true value and it includes both random and systematic error
In random error, there is a lack of ______ whereas in systematic error, there is a lack of _______
random: lack of precision
systematic: lack of accuracy
What is the difference between lack of accuracy and lack of precision?
Precision relates to how similar the estimates are to one another, regardless of how similar to the real value whereas accuracy relates to how close the estimates are to the true value, regardless of how close they are to each other
What affect might measurement error have in a descriptive study?
it could over/underestimate the prevalence
When considering information bias in cross-sectional studies, what do we might we need to ask ourselves?
- could the prevalence of the exposure be over/under-estimated?
- could the prevalence of the outcome be over/under-estimated?
What effect might measurement error have in an analytic study?
it can lead to misclassification (people without the exposure may be classed as having the exposure, and vice versa or people without the outcome may be classed as having the outcome, and vice versa).
What are the two sorts of misclassification? What do these mean?
- non-differential
- differential
Define non-differential misclassification
when measurement error and any resulting misclassification occur equally in all groups being compared
Define differential miclassification
when the measurement error and any misclassification do not occur equally in all groups being compared
Give an example of how differential misclassification can occur in a cross sectional study where participants are supplying the data?
they could report the exposure differently to those without the outcome
Give an example of how differential misclassification can occur in a case control study where participants are supplying the data?
the cases may more accurately recall past exposures compared to controls
Give an example of how differential misclassification can occur in a case control study where data is being collected by a third party?
an interviewer who is aware they are interviewing a case might ask more probing questions about the exposure of interest
Give an example of how differential misclassification can occur in a cohort study where data is being collected by a third party?
an interviewer who is aware of the exposure status may ask more probing questions about the outcome among those exposed compared with those in the comparison group
In what way does information bias affect case-control studies?
Case-control studies involve people being with the outcome being recruited and people without the outcome being recruited and then tracing back in time to see if they had the exposure. This involves asking questions about possible exposures and therefore this study type is susceptible to recall bias, which is a type of information bias
Define recall bias
systematic error due to any differences in accuracy or completeness of recall to memory of past events or experiences
When considering recall bias within a study, what three things do we need to ask?
- who does it affect?
- how does it affect them?
- what effect does this have on the measure of association?
What are three ways to minimise recall bias?
- objective measures
- validate self-reported measures with other information
- memory aids
In what way does information bias affect control studies?
Cohort studies involve recruiting people that you know have had an exposure and then following up over time to see if they develop the outcome. This means that there is potential for misclassification of exposure/outcomes
When looking at information bias in cohort studies, what do we need to consider?
- have the participants been correctly classified?
- has the outcome status been correctly classified
How can you minimise interviewer bias?
- have a clearly defined study protocol and measures
- structured questionnaire and standard prompts
- train the interviewers
- blinding
You can get differential misclassification in cohort studies if the misclassification depends on other axes of interest. For example, if (2)
- if classification of exposure depends on outcome
- if classification of outcome depends on exposure
How can bias occur in an RCT?
people involved in the study may know what group the participants are in
- bias can occur in knowledge from the treatment/exposure category influences the assessment of the outcome
- bias could occur if measurements are undertaken differently for different treatment groups
What are was to minimise information bias?
when collecting information from participants:
- validate the survey instruments
- validate using the objective measure
when using measurement instruments
- use standardised equipment
- use calibrated equipment
when collecting information via interviewers
- blinding
- using objective measurements
- use structured interviews and standardised prompts
- train the interviews
- have a clearly defined study pool
- have well defined exposures, outcomes and other factors collected in the study
Define publication bias
the result of the tendency of authors to submit, organisations to encourage, reviewers to approve, and editor to publish articles containing “positive” findings in contrast to findings or reports that so not report statistically significant or “positive” results
Define information bias
observation of information bias which results from systematic difference in the way data on exposure or outcome are obtained from the various study groups