biases
DESCRIPTION
By Prof. Steven Kassem as part of the 5th Research Summer School by King Abdullah International Medical Research Center (KAIMRC) - Western regionTRANSCRIPT
Steven Moustafa Kassem
Mitigating Researcher Bias
Steven Moustafa Kassem
Thank you!
unknown or unacknowledged error
created during the design,
measurement, sampling, procedure, or
choice of problem studied
What is Bias?
◦ bias is so pervasive because we want to confirm our beliefs
◦ science is organized around proving itself wrong not right
◦ key difference between qualitative and quantitative research is attempts to eliminate bias by quantitative researcher explicit acknowledgement of bias by qualitative
researchers
The Role of Bias in Science
when the study fails to identify the validity problems
when publicity about the research fails to incorporate the researchers cautions
Design Bias
selecting the most or least of anything creates a regression effect
Design Bias 1: Regression Effect
selecting the lowest functioning mentally ill people to study the effects of a therapy program
selecting the poorest people to study the effects of an anti-poverty program
selecting chronic homeless to study the effects of a housing program
Unless your report addresses the problem of regression, it will be a biased report
Study Dropouts may be the ones who needed it most
People also drop out when the program works
Design Bias 2: Attrition effect
Occurs when researcher fails to control for the effects of data collection and measurement
using self report is often biased by social desirability
most clinical research is highly vulnerable to measurement bias
when the person perceives there is something to lose by their answer
Measurement Bias
Differential Bias occurs when one group of study participants is more likely to be misclassified than the other
Misclassification of exposure is non-differential if it is similar among cases and controls i.e. the exposure (mis)classification is not related to the person's disease status
Interviewer bias happens when interviewers ask questions differently in case-control or cohort studies
a histopathologist may be more likely to report on a biopsy specimen as mesothelioma if a history of asbestos exposure is reported (biased follow-up)
Salmonella cases may be more likely to remember exactly what they ate than controls, since they may already have suspected a particular food (recall bias)
Types of Measurement Bias
Occurs when we administer the research interview or questionnaire under adverse conditions
Poor layout, obligatory participation, uncomfortable settings, etc…
Procedural Bias
only cases within a limited range of a disease spectrum are included.
This more commonly occurs with more obvious or advanced disease.
For example, in a study investigating the ability of MR imaging to depict cirrhosis, if only advanced clinical cases are included the sensitivity will be overestimated.
Disease Spectrum Bias
occurs when individual preferences or local practices determine which subjects undergo a certain treatment or imaging study
New treatments or imaging studies that have not been universally accepted into clinical practice are particularly prone to this type of bias.
Referral Bias
Exists when study subjects are self-selected for enrollment for treatment or
imaging
Differences may exist between those who volunteer and those who refuse participation
Volunteers may be more health conscious or even healthier than the general population
Self-Selection Bias
may arise in studies in which subjects are interviewed by an investigator who is also involved in the interpretation of a test result or determination of disease classification.
The investigator may inadvertently “coach” subjects or selectively review entire medical records
Interviewer Bias
results when additional factors or variables are associated with exposure and disease status
Common confounding variables are age and sex
Stratification (separation based on variables) is a common method to address confounding in the analysis phase
Confounding
journals tend to publish studies with positive results or better-quality study designs
does not arise within a given single study but can be seen in a review and analysis of the literature, as in a meta-analysis
leads to overly optimistic results or inflated associations
Publication Bias
retained knowledge of the results of one study influences the interpretation of the second study, potentially leading to a more accurate reading or diagnosis
learning-curve phenomenon
Reader-Order Bias
study subjects are randomly allocated to receive one or other of the alternative treatments under study.
Methods include allocation concealment and blinding
Mitigating Biases #1:Randomization
Allows for consideration of time factors
i.e.: differences in patients who visit an emergency room in the morning versus those who visit during the night
Mitigating Biases #2:Consecutive recruitment
Information such as pertinent test results, demographic data, or disease status, which may affect an investigator's test interpretation or assessment of an outcome, is not available to the investigator
A double-blinded study refers to one in which both the investigator and study subject are blinded to group assignment
Mitigating Biases #3:Blinding
Good Intentions
Mitigating Biases #4: