In a Directed Acyclic Graph (DAG) used in epidemiology, a 'collider' on the path between exposure and outcome, when conditioned upon, will:
- A Open a previously blocked path and introduce collider stratification bias ✓
- B Block the backdoor path and remove confounding
- C Have no effect on the association between exposure and outcome
- D Always indicate the presence of effect modification
Correct answer: A. Open a previously blocked path and introduce collider stratification bias
Explanation
A collider is a variable that is caused by two other variables on a path (both arrows point into it). In an unadjusted analysis, the path through a collider is blocked. However, conditioning (adjusting/stratifying) on a collider opens this path and induces a spurious association between the exposure and outcome — known as collider stratification bias or Berkson's bias in certain contexts. Confounders are blocked by conditioning, not opened.
Reference: Park's Textbook of Preventive and Social Medicine, 27th ed.
High-yield for: NEET PGINI-CETNExTFMGEUSMLEPLABMRCP
Written and medically reviewed by the StethoPrep medical team.