Methods explainer

Same table. Different causal questions.

Each regression row looks equally authoritative. Click a row to see why the model was only designed to answer one of them.

Adjusted model Outcome: cardiovascular events

The model's intended question

What is the effect of periodontitis on cardiovascular events?

Smoking confounds this relationship, so adjusting for smoking helps recover the intended exposure effect.

Adjusted estimate 0.51

True effect 0.50

Appropriate interpretation: this row answers the question the adjustment set was built for.