AF-CHF - Rhythm vs rate control in AF with HFrEF

Roy D, et al. Rhythm control versus rate control for atrial fibrillation and heart failure. N Engl J Med 2008;358:2667-77.

Bottom-line: In individuals with both AF & HFrEF, a rhythm-control strategy is not superior to an aggressive rate-control strategy targeting resting HR <80 bpm. More patients starting with the rhythm-control strategy will require a strategy change (NNH 9), but neither strategy works for everybody.

 

Patients (n=1376)

  • Inclusion
    • AF
      • Episode with EKG documentation lasting at least 6h or requiring cardioversion in previous 6 months, or
      • Episode lasting 10+ minutes in previous 6 months & previous cardioversion for AF
    • HF
      • NYHA II-IV in previous 6 months, or
      • Hospitalized for HF in previous 6 months, or
      • LVEF 25% or less
    • LVEF 35% or less measured in last 6 months
  • Exclusion
    • Persistent AF >12 months
    • Reversible cause of AF or HF
    • Decompensated HF in previous 48h
    • Use of antiarrhythmics for other arrhythmias
    • 2o-3o AVB with bradycardia <50 bpm
    • Hx long QT syndrome
    • Dialysis-dependent renal failure
  • "Typical" patient
    • Age 66 y
    • Male 78-85%
    • NYHA class III-IV 32%
    • HF etiology: Ischemic (48%), hypertensive (10%), valvular (5%)
    • Prior hospitalization for AF (50%), HF (55%)
    • AF paroxysmal (1/3), persistent (2/3)
    • PMHx
      • Previous stroke/TIA 10%
      • HTN 49%
      • Diabetes 22%
    • AF on EKG (55-60%)
    • LVEF 27%
    • Concomitant meds
      • ACEI 86%, ARB 11%
      • Mineralocorticoid antagonist 45%
      • OAC 85-90%
      • ASA 40%
      • Lipid-lowering 43%
    • ICD 7%

Interventions

  • I: Rhythm control: Aggressive pharmacotherapy + electrical cardioversion to prevent and cardiovert AF
    • Drug of choice: Amiodarone, then sotalol or dofetilide as required
    • Drugs @ 1 year: Amiodarone (82%), sotalol (2%), dofetilide (<1%)
      • Beta-blocker (80%), digoxin (~50%), anticoagulant (88%)
    • Electrical cardioversion
      • 1st recommended <6 weeks after enrollment not converting to NSR with pharmacological rhythm control alone
    • 2nd recommended <3 months after enrollment if still not in NSR
    • Subsequent cardioversions PRN
  • C: Rate control: Adjusted doses of beta-blocker & digoxin to achieve resting HR <80 bpm & <110 bpm during 6-min walk test (tested @ month 4 & 12, then yearly)
    • Drugs @ 1 year: Beta-blocker (88%), digoxin (75%), verapamil/diltiazem (3%)
      • Amiodarone (7%), sotalol or dofetilide (<1%), anticoagulant (92%)
  • Interventions common to both groups:
    • Max-tolerated doses of beta-blockers (for HFrEF management)
    • Anticoagulation

Results @ mean 3 y f/u

  • Death: 32% vs 33% (p=0.68)
    • CV death (primary outcome): 27% vs 25% (p=0.53)
  • Hospitalization: 64% vs 59% (p=0.06)
    • AF hospitalization: 14% vs 9% (p=0.001)
  • Worsening HF: 28% vs 31% (p=0.17)
  • Switched to other intervention: 21% vs 10%
  • AF on EKG at study visit:
    • Month 4, years 1-3: ~20% vs ~60% (during f/u, >55% in rhythm-control group had at least 1 AF recurrence)
    • Year 4: ~25% vs ~70%

Generalizability

  • Representative of individuals with HFrEF and moderately good use of HFrEF medical therapies & low ICD use
  • Rhythm-control intervention consistent with real world use; rate-control intervention similar to "intensive" intervention from AFFIRM trial

Internal validity

  • Unclear risk of allocation bias
    • Allocation concealment not described + some moderately-large baseline differences in certain characteristics (e.g. male 78% vs 85%, AF on baseline EKG 54% vs 61%)
  • Unclear risk of performance & detection bias
    • Predefined treatment protocols accounted for most potential differences in interventions
    • Rhythm-control group required more AF-related hospitalizations, likely cardioversion-related
    • Higher rate of cross-over in rhythm-control group
    • Once outcomes reported, adjudicated by committee unaware of treatment allocation
  • Unclear risk of attrition bias
    • 5-6% loss-to-follow-up, which could be enough to hide differences between groups in main outcomes