Which design feature best reduces selection bias in a cohort study?

• A. Strictly using a single clinic for recruitment.
• B. Random sampling from a defined population and clear inclusion criteria.
• C. Blinding participants to the exposure.
• D. Collecting exposure data via self-report only.

Answer: (B)

Selection bias in a cohort study happens when who gets into the study isn’t representative of the people you want to learn about, often because of how participants are chosen. The best way to prevent this is to sample randomly from a clearly defined population and apply explicit inclusion criteria. Random sampling gives everyone in the defined population an equal chance to be included, so the study group more accurately reflects the target population and the exposure-outcome relationship isn’t distorted by who happened to enroll. Clear inclusion criteria ensure consistency in who is eligible and reduce the risk that people are selected based on their exposure or disease status.

Recruiting from a single clinic can create selection bias because patients at that clinic may differ in important ways from the general population (they might have different health-seeking behaviors or disease prevalence). While blinding participants to their exposure addresses how outcomes are measured, it doesn’t fix how participants were chosen for the study. Collecting exposure data by self-report only can introduce information bias and misclassification, not to the selection process.