A cohort study reports that heavy coffee drinkers have a crude relative risk of 2.0 for lung cancer. When the analysis is stratified by smoking status, coffee drinkers and non-drinkers show identical lung cancer rates within smokers and within non-smokers. The most appropriate interpretation is:

• A Coffee causes lung cancer through a mechanism unmasked only after stratification
• B Coffee drinking is an effect modifier of the smoking-lung cancer relationship
• C Smoking is a confounder, being associated with coffee drinking and independently causing lung cancer ✓
• D Stratification has introduced Berkson's bias into the analysis

Explanation

A variable is a confounder when it is associated with the exposure, independently influences the outcome, and is not on the causal pathway. Here smoking correlates with coffee consumption and strongly causes lung cancer, producing a crude association that disappears completely within strata, the hallmark of confounding. Effect modification means the true association differs across strata, which is not the case since rates are identical within each stratum. Berkson's bias arises from selecting cases and controls from hospitals and cannot be created by stratification.

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.