Controls who experienced hypothetical causal intermediates should not be excluded from case-control studies

Area of Science:

• Epidemiology
• Cancer Research
• Biostatistics

Background:

• Case-control studies are crucial for investigating associations between exposures like cigarette smoking and diseases such as colorectal cancer.
• Adenomatous polyps are considered an intermediate step in the hypothetical causal pathway between smoking and colorectal cancer.
• A recommendation exists to exclude controls with adenomatous polyps from these studies, claiming it prevents bias toward the null.

Purpose of the Study:

• To evaluate the impact of excluding controls with adenomatous polyps from case-control studies on the association between cigarette smoking and colorectal cancer.
• To determine if excluding intermediate causal steps in case-control studies introduces bias.
• To clarify the correct methodology for handling intermediate variables in epidemiological research.

Main Methods:

• Theoretical analysis of bias in case-control studies.
• Examination of a hypothetical causal pathway involving an exposure, an intermediate, and a disease.
• Demonstration of bias direction under specific assumptions of exposure-intermediate-disease relationships.

Main Results:

• Excluding controls with adenomatous polyps, when these polyps are on the causal pathway, creates a bias away from the null.
• The rationale for exclusion appears to stem from a misunderstanding of case-control study design (trohoc fallacy).
• Case-control studies compare exposure status, not disease status, between groups.

Conclusions:

• Excluding individuals with intermediate conditions like adenomatous polyps from control groups can introduce bias away from the null.
• The practice of excluding such controls in smoking and colorectal cancer studies is methodologically flawed.
• Correct interpretation of case-control studies requires understanding their design as comparisons of exposure distributions.