Biostatistics and epidemiology
Confounding and effect modification
What makes a confounder, how stratification separates confounding from effect modification, the ways to control confounding, and the hormone therapy story as the classic example.
Tier 1Medical sciences concept
Written with AI assistance and fact-checked against the sources below. For exam study, not clinical decisions. How notes are made
At a glance
- A confounder distorts the exposure-outcome association because it is linked to both. It is a real feature of the population, not a measurement error.
- Effect modification (interaction) is a real difference in the effect of an exposure across levels of a third variable. It is something to describe, not remove.
- Stratify to tell them apart. Similar stratum estimates that differ from the crude: confounding. Different stratum estimates: effect modification.
Loading the rest of the note…
Finished reading?
Sources
- Modern Epidemiology (4th ed.)Wolters Kluwer (Lash, VanderWeele, Haneuse and Rothman, 2021), 2021
- Gordis Epidemiology (6th ed.)Elsevier (Celentano and Szklo, 2019), 2019
- The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studiesLancet (von Elm et al., 2007;370:1453-7), 2007
- Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results from the Women's Health Initiative randomized controlled trialJAMA (Writing Group for the WHI Investigators, 2002;288:321-33), 2002
- Manson JE et al. Estrogen plus progestin and the risk of coronary heart disease. N Engl J Med 2003;349:523-34New England Journal of Medicine, 2003
- Last updated
- 14 November 2026
What changed
- 14 November 2026 Personal emails, About story and Australian-style copy