Fig. 1.

Causal diagram representing causal relations between obesity, ICU admission, and mortality. The circle around variable U indicates that it has not been measured. The box around variable ICU indicates that the analysis conditions on it by design, since the study cohort only comprised patients admitted to the intensive care unit. This causal diagram has been simplified for the purpose of illustrating collider stratification bias. ICU, intensive care unit; U, unmeasured factors