Skip to main content
Fig. 4 | Cardiovascular Diabetology

Fig. 4

From: Association of body mass index and long-term mortality in patients from nationwide LIPIDOGRAM 2004–2015 cohort studies: no obesity paradox?

Fig. 4

Collider stratification bias as an explanation of obesity paradox. A directed acyclic graph representing the relations between obesity (BMI), cardiometabolic mediators and mortality. Dashed red lines represent correlations created by conditioning/adjusting on colliders. The true causal effect of obesity on mortality may be largely mediated by cardiometabolic diseases. As those diseases can also be influenced by factors other than BMI, they are colliders. Subgroup analyses involving patients with or without cardiometabolic diseases are subject to collider bias as non-causal correlations between confounders and obesity are created

Back to article page