A cohort study reports a crude relative risk of 2.0 for coffee consumption and pancreatic cancer. When the data are stratified by smoking status, the relative risk is 1.1 among smokers and 1.1 among non-smokers, and smokers are far more prevalent among coffee drinkers. The crude estimate most likely reflects:
- A Effect modification by smoking status
- B Random misclassification of coffee intake
- C Confounding by smoking, which should be controlled by adjustment ✓
- D Selection bias arising during follow-up
Correct answer: C. Confounding by smoking, which should be controlled by adjustment
Explanation
Confounding is suspected when the crude estimate differs meaningfully from the stratum-specific estimates, which here are uniform at 1.1. Smoking is associated with coffee drinking and independently with pancreatic cancer, satisfying the classical definition of a confounder. Effect modification is excluded because the stratum-specific estimates are identical rather than materially different, and adjustment, for example by the Mantel-Haenszel method, would yield an estimate close to 1.1.
Reference: Gordis Epidemiology, 6th ed.
High-yield for: NEET PGINI-CETNExTFMGEUSMLEPLABMRCP
Written and medically reviewed by the StethoPrep medical team.