Related Experiment Videos
Bias arising in case-control studies from selection of controls from overlapping groups
H Austin1, W D Flanders, K J Rothman
1Department of Epidemiology, School of Public Health, University of Alabama, Birmingham 35294.
International Journal of Epidemiology
|September 1, 1989
Summary
Using overlapping control categories in case-control studies, like friend controls or caliper matching, introduces significant selection bias. Researchers should avoid these methods to ensure study validity and reliable results.
Area of Science:
- Epidemiology
- Biostatistics
Background:
- Case-control studies are crucial for investigating disease etiology.
- Selection bias is a major threat to the validity of case-control studies.
- Common control selection methods may inadvertently introduce bias.
Purpose of the Study:
- To demonstrate how overlapping control categories can lead to selection bias in case-control studies.
- To highlight the practical consequences of using friend controls.
- To provide recommendations for avoiding biased control selection.
Main Methods:
- The study uses three illustrative examples to explain selection bias.
- Examples include the use of friend controls and age-matching (caliper matching).
- The focus is on bias arising from overlapping variable categories.
Main Results:
- Selection bias occurs when control groups are chosen from categories that overlap with the case group.
- Friend controls can introduce substantial bias, a common issue with practical implications.
- Age-matching using a narrow caliper can also result in biased control selection.
Conclusions:
- The selection of controls from overlapping categories, including friend controls and caliper-matched controls, should be avoided.
- Careful consideration of control selection is essential to prevent bias in case-control research.
- Ensuring independent selection of controls is critical for valid epidemiological findings.