Bias Flashcards
Definition of bias
Systematic error in the design, conduct or analysis of a study that results in a deviation from the truth
Random deviation (error) definition
- RE’s are equally and randomly distributed across the comparator groups and are independent of exposure or disease status
- Tend to lead to a smaller observed effect size
- Deviates in different directions
- Not precise (deviating from actual value)
- -> The result of random error would be an imprecise estimate of person’s ____
Definition of selection bias
Commonly defined in two ways:
- Error due to systematic differences in characteristics between those who are selected into a study and those who are not
- > Selection bias INTO A STUDY => the study population cannot be compared to the general population - Error due to systematic differences in characteristics between those who are selected into one of the groups within an analytical study (those who are selected into each comparison group within an alaytic study)
- > Selection bias into groups WITHIN an analytic study => the groups cannot be VALIDLY compared with each other
“Occurs when the way in which groups of subjects are selected for a study means that the groups cannot be validly compared with each other/ the general popn - diffs between baseline characteristics of he groups that are compared”
Definition of sampling bias
- T ype of selection bias
- A systematic error that occurs when each potential member of the population being studied does not have an EQUAL chance of selection
- Error due to systematic differences in chara. between those who are selected into a study and those who are not
- Sampling bias and selection bias can sometimes be used interchangeably
Definition of allocation bias
- Type of selection bias
- Term used in RCTs to refer to bias as a result of systematic differences in the chara. between those who are in one trial arm (study group) and those in another trial arm.
- Occurs when all participants do not have the same probability of entering any branch in the trial
- Bias in how parts. are assigned to treatment and comparison groups in clinical trial
Definition of spectrum bias
- Type of selection bias
- Systematic error in diagnostic test accuracy studies that arises when specific groups of patients are inappropriately excluded - concerned with the INCLUSION of pts into a DTA study
- e.g. ‘difficult to diagnose’ patients
- Results in index test appearing more accurate than it actually is
Definition of response/ responder/ participation bias
- Systematic error due to difference in chara. between those who choose to volunteer to participate in a study or who choose one study group compared to another, and those who do not
Definition of loss to follow up/ follow up/ attrition bias
- Type of selection bias
- Systematic error due to the diff. in the chara. of those participants who were lost to follow up compared to those who completed the study
- Unequal loss of participants
Definition of healthy worker effect
- Typically seen in observational (occupational epidemiology) studies of occupational exposures with improper choice of comparison group
- Less healthy workers are more likely to reduce their workplace exposures
- makes exposed workers appear healthier than control popn - using general popn wrong because it includes workers and non-workers
- describes the deficit of both morbidity and mortality when workers
and the general population are compared. - Workers usually exhibit lower overall death rates
than the general population because the severely ill and chronically disabled, who form part
of the generally population, are usually not in employment.
Definition of information/ observation/ measurement bias
A systematic error in the way that data is gathered from the comparison groups that results in differential quality (Accuracy) of information between comparator groups
- The source of measurement bias is either the measurer or those being measured
“Occurs when the way in which the data are obtained from groups being compared differs systematically”
Definition of interviewer bias
- Type of information bias
- Aspects of the interviewers and the way in which they ask questions and respond to answers
- error due to interviewers’ subconscious/conscious gathering of data that
differs systematically between cases and controls. If the interviewer is aware of whether the
person being questioned about exposure is a case rather than a control, they might probe
the cases more for evidence of exposure to the factor of interest.
Definition of detection bias
- Type of information bias
- Systematic diff. between groups in how outcomes are determined / assessed (this could relate to ascertainment of outcome, diagnosis or verification of outcomes, for example)
- can occur in trials when groups differ in the way outcome info is collected or the way outcomes are verified
- can cause over or underestimate of size of effect
- need to blin - blinding of outcome assessors can be especially important for assessment of subjective outcomes, e.g. degree of post op pain
Definition of performance bias
- Systematic diff between groups in the care that is provided, or in exposure to factors other than the intervention of interest
- This type of bias particularly affects RCTs
- one group gets more attention from investigators than another group
Definition of reporting bias
- Type of information bias
- Systematic diff. between groups of reported and unreported findings
- Within a published report, those analyses with stats sig diffs between intervention groups are more likely to be reported than non-sig diffs
Definition of verification bias
- Systematic error in diagnostic test accuracy studies that arises when not all the participants receive both the index test and the reference standard
- This is most likely to occur with patients with negative index test results
