7.2: Information Bias (Measurement Error) ✅ Flashcards
Systematic error
Selection bias
Information bias
Information bias
Systematic error arising from mistakes during measurement of key study variables
-from wrong/inaccurate assessment of exposure/outcome
-researcher part (unintentionally) or from participants (unintentionally/intentionally)
-instrument bias (fault of the instrument)
Assessor bias
Wrong diagnosis due to clinical error
“Not blinded” to exposure/outcome status of participants
Inaccurate measurements due to faulty instruments
Poor training of assessor
Mistakes during recording of the data and transferring data
Response bias
Wrong/inaccurate answers from participants:
-due to misinterpretation
-intentionally
Wrong/inaccurate answers due to :
-poor recall
-sensitive issue
Recall bias
Participants with a particular outcome or exposure may remember events more clearly or amplify their recollections
Very common in case-control studies
-the primary difference arises more from under-reporting of exposures in the control group
(rather than over-reporting in the case group)
Interviewer bias
A researcher’s knowledge may influence:
-the structure of questions and
-the manner of presentation
Which may influence responders (in any study design)
Observer bias
Researchers may have preconceived expectations of what they should find in an examination
Hawthorne effect
Participants act differently if they know they are being watched
Surveillance bias
The group with the known exposure or outcome may be followed more closely or longer than the comparison group
Misclassification bias
Errors made in classifying either disease or exposure status
Occur in exposure, outcome and covariates
Minimizing bias
Be purposeful in the study design to minimise the chance for bias
-e.g. Use more than one control group
Define a priori who is a case or what constitutes exposure so that there is no overlap
-define categories within groups clearly
Set up strict guidelines for data collection
-train observers or interviewers to obtain data in the same fashion
-optimize questionnaire
How does information bias affect study results
To investigate an association you have to measure the factors accurately
Information bias can occur in the exposure and outcome so association between them will be distorted
More likely to arise from participants or through medical records
Affects mainly studies that rely on self-reports
- Outcomes can be double-checked
- Exposures require using reliable instruments
What must we always assume if a study solely relies on self report?
There is information bias
This compromises the validity of the study results
Validity
The extents to which an assessment tool measures accurately what is intended to measure
Criterion validity
Most common used in medical research
Results from assessment tool are compared to the results of an established assessment tool
Reliability
Aka reproducibility or repeatability
Overall consistency of a measure as regards to producing the same results as when administered under the same conditions in the same group of people
Reliability type
Inter-observer reliability
Intra-observer tool
Inter-observer reliability
The degree of agreement between the results when:
two or more researchers (observers) administer the assessment tool :
-on the same people
-under the same conditions
Intra-observer reliability
Describes the agreement between results when the assessment tool is used by the same researcher (observer) on 2 or more occasions (under the same conditions and in the same test population)
How does bias affect study results
False positive and false negative findings can arise in situations where information bias affects the outcome of interest
If both exposure and outcome are affected by information bias, then results are expected to be particularly biased (erroneus)
Also, all sorts of bias (including selection bias) as well random error can give rise to false positive and false negative findings
Internally valid
Findings of the study were not due to one of the 3 sources of error
The conclusions reached are likely to be correct for circumstance
(Systematic error, assessor bias and response bias)
What must you not do in the goal of generalization?
Do not compromise internal validity
As an invalid result cannot be generalized
