Bias Flashcards
Types of error
Random Systematic (bias)
Types of systematic error
Measurement bias
Selection bias
Analysis Bias
When is a study said to be internally valid?
If results are close to the truth
When is a study said to be externally valid?
If results are applicable to clinician seeking the evidence
Another name for external validity?
Generalisability
What can compromise the validity of a study
Random error
Systematic error
Confounding
What is a random error?
Due to chance
Why are all studies subject to random error?
They use samples from a population to estimate
What can be used to test probability of random errors?
P values
Confidence Intervals
How can random error be reduced?
Repeating study elsewhere
What is a systematic error/bias?
Error in the way we select patients, measure outcomes or analyse data
What can bias lead to?
False conclusions
Over of under-estimation of an effect
At what stage of a study can bias not be controlled for?
Analysis stage
How can bias be classified?
According to direction of change they produce in study outcome
What is negative bias?
Yields estimates closer to null value i.e. no difference between two groups
What is positive bias?
Higher magnitude of estimates than the true ones
What is switch-over bias?
Change of direction of association
When can switch-over bias occur?
Exaggeration of negative or positive bias
When does selection bias occur?
When selected groups differ in important factors other than study variables
Uneven diagnostic procedures
What reduces selection bias?
Adequate randomisation
Types of selection bias
Berkson bias Neyman bias response bias Unmasking bias Lead-time bias Referral bias Diagnostic purity bias Membership bias
Another name for Berkson bias?
Admission rate bias
What is Berkson bias?
Difference in rates of admission of cases and controls due to influence of exposure
What is another name for Neyman bias?
Incidence-prevalence bias
What is Neyman bias?
When ascertaining causation, one must look for an association between risk factor and incidence, not prevalence.
When does response bias occur?
When person who responds to invitation to participate in a study differs from those who do not.
When is response bias particularly relevant?
When evaluating screening tests
What is unmasking bias?
When risk factor unmasks an event which it is suspected to cause
What is lead-time?
Difference in time between date of dx with screening and date of sx without screening
What cannot be done if lead time is not accounted for?
Survival time cannot be compared to unscreened control group of patients
What can lead-time bias lead to i.e. if lead time is not taken into account?
Increase survival time due solely to advanced date of diagnosis will result in lead time bias
What is referral bias?
Variation in concentration of exposures or diseases between patients in primary and secondary care
What is diagnostic purity bias?
Exclusion of comorbidities resulting in a non-representative sample
When is diagnostic purity bias a problem?
RCTs
What is membership bias?
Case identification using members of patients organisations leading to systematic differences from non-representatives
When do measurement biases occur?
When data is not collected in a uniform fashion
How can measurement bias be reduced?
Blinding
Types of measurement bias
Recall bias Reporting bias Observer bias Surveillance bias Work up/verification bias Misclassification bias Desirability bias Hawthorne effect
What is recall bias?
Subjects often recall risk factors differently depending on their disease status
Which studies are vulnerable to recall bias?
Case-control studies
What is reporting bias?
When a larger percentage of case or control subjects are reluctant to report exposure due to attitudes/perceptions/other concerns
What is observer bias?
When a researcher knowingly or unknowingly evaluates a variable depending on the status of the individual
What is surveillance bias?
Diseases may be better ascertained in monitored population than in general population
What is verification bias?
During assessment of validity of a diagnostic test, execution of gold standard test may be influenced by results of new instrument
What is misclassification bias?
Cases may be misclassified as controls or exposed group may be misclassified as control.
When does misclassification bias only occur?
If the misclassification is one sided
What is non-differential misclassification?
Both sides affected equally
What can lead to non-differential misclassification?
Errors in measurement instruments
What can non-differential misclassification lead to?
Reduction in observed magnitude of association
What is desirability bias?
Patients may choose socially desirable answers during data collection, distorting the picture
Which type of study is Hawthorne effect prevalent?
Cross-sectional surveys using questionnaires
Types of analysis bias
Contamination
Attrition
What is contamination bias?
When participants change group
What is attrition bias?
Patients lost to follow-up
What can minimise analysis bias?
Intention to treat analysis
