Survival as a source of confounding, selection bias, and effect modification

Area of Science:

• Epidemiology
• Biostatistics
• Health Research Methodology

Background:

• Survival is a frequently overlooked factor causing confounding, selection bias, and effect modification in research.
• Understanding how survival influences study outcomes is crucial for accurate data interpretation.

Purpose of the Study:

• To describe the mechanisms by which survival introduces bias in research.
• To outline strategies for averting bias caused by survival in epidemiological studies.

Main Methods:

• Directed acyclic graphs (DAGs) were employed to visualize survival-induced confounding and selection bias.
• Selection diagrams illustrated survival's role in effect modification by factors like age and health status.

Main Results:

• Survival acts as a confounder when exposure allocation begins or changes post-recruitment.
• Type 1 selection bias arises when exposure and outcome (or competing risk) affect survival pre-recruitment, potentially altering estimates.
• Bias due to survival is more pronounced in older or less healthy individuals and can be reduced by aligning exposure allocation with recruitment.

Conclusions:

• Consider confounding by survival if exposure allocation starts or changes after recruitment.
• Address Type 1 selection bias from survival when exposure precedes recruitment.
• Prioritize study designs less susceptible to survival-related confounding and selection bias, especially for long-term exposures.