In a study of combined oral contraceptive use and venous thromboembolism, the overall relative risk is 3.0. Stratification reveals a relative risk of 8.0 in women above 35 years and 1.5 in those aged 20 to 25 years. The most appropriate conclusion and analytic step are:

• A Confounding by age is present, so report the Mantel-Haenszel adjusted estimate alone
• B Age is on the causal pathway, so adjustment is mandatory
• C The results reflect chance, so pool all subjects and ignore stratification
• D Effect modification by age is present, so report separate stratum-specific estimates ✓

Explanation

When stratum-specific estimates differ substantially from each other, the exposure effect varies across levels of the third variable, which defines effect modification rather than confounding. In confounding the stratum-specific values are similar to each other but differ from the crude value. With effect modification, a single pooled or adjusted estimate is misleading and hides clinically important differences, so separate estimates for each stratum must be reported.

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.