A case-control study shows a crude odds ratio of 2.0 for coffee consumption and pancreatic cancer. When the data are stratified by smoking status, the odds ratio is 1.0 among smokers and 1.0 among non-smokers, and smokers are more prevalent among coffee drinkers. This pattern indicates:

• A Effect modification by smoking
• B Selection bias
• C Random error
• D Confounding by smoking ✓

Explanation

Confounding is present when the crude estimate differs meaningfully from the stratum-specific estimates, and the stratum-specific estimates are similar to each other. Here both strata show an odds ratio of 1.0, so the entire crude association of 2.0 is explained by unequal smoking distribution between exposure groups. Effect modification would produce stratum-specific estimates that differ substantially from each other.

Reference: Gordis Epidemiology, 6th ed.

High-yield for: NEET PGINI-CETNExTFMGEUSMLEPLABMRCP

Written and medically reviewed by the StethoPrep medical team.