FIGURE 2.

Mortality risk and β-blocker use. ^aPrimary analysis: β-blocker exposure was determined at baseline and matched pairs were censored upon β-blocker exposure status changes. ^bIntention-to-treat analysis: β-blocker exposure was determined at baseline; patients were not censored upon β-blocker exposure status changes. ^cTime-varying analysis: β-blocker exposure was determined using a time-varying model; patients were not censored upon β-blocker exposure status changes. ^dAdjusted for the following characteristics: atrial fibrillation/flutter, coronary artery disease, chronic obstructive pulmonary disease, myocardial infarction, anticoagulants, ACEi, amiodarone, calcium channel blockers, digoxin, loop diuretics, proton pump inhibitors, statins, inhaled anticholinergics, inhaled β-agonists and inhaled glucocorticoids.