A study reports a crude relative risk of 1.05 for alcohol intake and bladder cancer. When the data are stratified by smoking status, the relative risk is 2.2 in both smokers and non-smokers, and the smoking-adjusted pooled estimate is 2.2. Smoking prevalence is much higher among heavy drinkers. The crude estimate differs from the adjusted estimate because:

• A Smoking is an effect modifier that changes the alcohol effect across strata
• B Stratification introduced selection bias into the analysis
• C Smoking acted as a negative confounder, masking the true association ✓
• D The crude risk ratio suffered from collider stratification bias

Explanation

Smoking is associated with alcohol intake in the source population and independently causes bladder cancer, so it confounds the crude estimate. Because the crude RR (1.05) lies below both stratum-specific values (2.2), the confounder was pulling the estimate toward the null, i.e., negative confounding. Effect modification is ruled out because the stratum-specific estimates are identical to each other, so a single adjusted summary estimate is appropriate.

Reference: Modern Epidemiology (Rothman), as cited in Indian PSM teaching ed.

High-yield for: NEET PGINI-CETNExTFMGEUSMLEPLABMRCP

Written and medically reviewed by the StethoPrep medical team.