Previous studies have not confirmed associations between some current performance measures for inpatient heart failure processes of care and postdischarge outcomes. It is unknown if alternative measures are associated with outcomes.

Using data for 20,441 Medicare beneficiaries in OPTIMIZE-HF from March 2003 through December 2004, which we linked to Medicare claims data, we examined associations between hospital-level processes of care and patient outcomes. Performance measures included any β-blocker for patients with left ventricular systolic dysfunction (LVSD); evidence-based β-blocker for patients with LVSD; warfarin for patients with atrial fibrillation; aldosterone antagonist for patients with LVSD; implantable cardioverter-defibrillator for patients with ejection fraction ≤35%; and referral to disease management. Outcome measures were unadjusted and adjusted associations of each process measure with 60-day and 1-year mortality and cardiovascular readmission at the hospital level.

Adjusted hazard ratios for 1-year mortality with a 10% increase in hospital- level adherence were 0.94 for any β-blocker (95% CI, 0.90-0.98; P = .004), 0.95 for evidence-based β-blocker (95% CI, 0.92-0.98; P = .004); 0.97 for warfarin (95% CI, 0.92-1.03; P = .33); 0.94 for aldosterone antagonists (95% CI, 0.91-0.98; P = .006); 0.92 for implantable cardioverter-defibrillator (95% CI, 0.87-0.98; P = .007); and 1.01 for referral to disease management (95% CI, 0.99-1.03; P = .21).

Several evidence-based processes of care are associated with improved outcomes, can discriminate hospital-level quality of care, and could be considered as clinical performance measures.