Type of Bias Flashcards
Spectrum bias
Purposefully excluding ‘difficult to diagnose’ patients
What is the effect of spectrum bias?
The index test appears more successful at diagnosing
Review bias
If interpretation of the index test is not independent and blind to the reference standard
What is the effect of review bias?
The index test appears more accurate
Verification bias
Tendency not to give the reference standard test to patients who were negative on the index test, because the reference standard is more invasive
What is the effect of verification bias?
Over/underestimation of the index test accuracy
Selection bias
Systematic differences in the selection process of exposure groups
What is the effect of selection bias?
Groups cannot be validly compared with each other, or the general population
What are two types of selection bias?
Sampling bias
Response bias
Sampling bias
Certain individuals/groups are systematically more likely to be chosen than others
Response bias
Systematic differences in the characteristics of non-responders vs responders
What is the effect of response bias?
The participants may be representative the type of people who are more likely to respond to a survey, who may share other systematic differences to non-responders, therefore the chosen population is not representative
Information bias
Systematic differences in the method of obtaining data
What are five types of information bias?
Recall bias
Recording bias
Interviewer bias
Lost-to-follow-up bias
Social acceptability bias
Recall bias
Patients with the disease think more about health than healthy patients, therefore have more accurate knowledge/memory of exposure
Recording bias
Patients with the disease have more extensive medical notes and information than healthy patients, therefore more informed conclusions can be made from these patients
Interviewer bias
The interviewer knows whether the patient has the disease or not, and subconsciously alters their questions or question style
Lost-to-follow-up bias
Patients who are lost to follow-up, because of illness, are more likely to have data of interest than control patients
Social acceptability bias
Interviewee tries to make responses ‘socially acceptable’, therefore is dishonest
Healthy-worker effect
The cohort may exclude some individuals in the general population who are not fit enough to be included in the study, therefore this demographic is not represented
Performance bias (control group)
The control group receive the intervention anyway, causing contamination, or they seek complimentary treatments, or HCPs give extra treatment
Performance bias (intervention group)
More contact with HCPs means that health problems might be detected sooner, and additional healthcare advice might be given than in the control group
Detection bias
Systematic differences in how outcomes are determined or how odds ratios are presented, because the analyst is not blinded to the grouping of patients
What is the effect of detection bias?
Over/underestimate of effect
