A case-control study shows a crude odds ratio of 2.8 for passive smoking and childhood asthma. When the data are stratified by household crowding and a Mantel-Haenszel pooled odds ratio of 1.9 is obtained, the investigator concludes that crowding was acting as a:

• A Effect modifier, since the pooled estimate lies between strata
• B Selection bias, since stratum-specific estimates differed
• C Mediator, since it lies downstream of smoking
• D Confounder, since adjustment moved the estimate toward the null ✓

Explanation

When stratum-specific estimates are similar to each other but differ appreciably, here by more than 10 percent, from the crude estimate, the third variable is a confounder and the Mantel-Haenszel pooled estimate is reported. If the stratum-specific estimates themselves differ substantially from one another, that would indicate effect modification, and separate stratum estimates would be presented instead of pooling, making option A incorrect.

Reference: Gordis Epidemiology, 6th ed.

High-yield for: NEET PGINI-CETNExTFMGEUSMLEPLABMRCP

Written and medically reviewed by the StethoPrep medical team.