AFFIRM
PubMed • Full text • PDF
Clinical Question
Among patients with atrial fibrillation and a high risk of stroke or death, what are the effects of rate control versus rhythm control on mortality?
Bottom Line
In patients with nonvalvular AF, there is no survival benefit between rate and rhythm control, but rhythm trends toward increased mortality.
Major Points
The 2002 Atrial Fibrillation Follow-up Investigation of Rhythm Management (AFFIRM) trial was the first and largest study to compare rate-control and rhythm-control strategies for the treatment of AF. Among 4,060 patients with non-valvular AF and a high risk of stroke or death, AFFIRM demonstrated no survival advantage between rate-control (using ß-blocker, calcium channel blocker and/or digoxin) and rhythm-control strategies. All patients were anticoagulated on warfarin initially, but patients in the rhythm control arm who maintained normal sinus rhythm for at least 4 consecutive weeks could stop. In fact, there was a nonsignificant trend toward a decrease in mortality associated with the rate-control strategy, particularly those aged ≥65 and those without a history of HF. However, rates of ischemic stroke were similar between both groups, at approximately 1% per year.
RACE II (2010) directly addressed the issue of optimal rate control for patients with permanent AF.
Guidelines
AHA/ACC/HRS AF (April 2014, adapted)[1]
- In patients with nonvalvular AF with prior stroke, TIA, or CHA2DS2-VASc score ≥2, recommend oral anticoagulation with:
- Warfarin, goal INR 2-3 (class I, level A)
- Dabigatran (class I, level B)
- Rivaroxaban (class I, level B)
- Apixaban (class I, level B)
- In patients with nonvalvular AF unable to maintain INR 2-3 with warfarin, recommend dabigatran, rivaroxaban, or apixaban (class I, level C)
- In patients with nonvalvular AF with moderate or severe CKD with CHA2DS2-VASc score ≥2, consider treatment with reduced doses of dabigatran, rivaroxaban, or apixaban, although safety has not yet been clearly delineated (class IIb, level C)
- In patients with ESRD, dabigatran and rivaroxaban are untested and are not recommended (class III, level C)
- In patients with a mechanical heart valve, do not use dabigatran (class III, level B)
- Rate control with a beta blocker or non-DHP CCB for patients with paroxysmal, persistent, or permanent AF (class I level B)
- HR <80 BPM is reasonable for symptomatic AF management (class IIa level B)
- HR <110 is reasonable if asymptomatic and LV systolic function is preserved (class IIb level B)
- Oral amiodarone may be useful for rate control when other interventions are unsuccessful or contraindicated (class IIb level C)
- Do not use non-DHP CCBs in patients with decompensated HF (class III level C)
- Do not use digoxin, non-DHP CCBs, or IV amiodarone in the presence of AF with pre-excitation, as it may result in VF (class III level B)
Design
- Multicenter, parallel-group, randomized, controlled trial
- N=4,060 patients with nonvalvular atrial fibrillation
- Rate-control strategy (n=2,027)
- Rhythm-control strategy (n=2,033)
- Setting: 213 clinical sites and their satellite sites, including University of Washington
- Median follow-up: 3.5 years
- Analysis: Intention-to-treat
- Primary outcome: All-cause mortality at 5 years
Population
Inclusion Criteria
- Age ≥65 years with AF that was likely to be recurrent
- AF likely to cause illness or death for the participant
- Long-term treatment of AF was warranted
- Other risk factors for stroke or death
Exclusion Criteria
- Contraindication to anticoagulation therapy
- Ineligible to undergo trials of ≥2 medications in either treatment strategy
Baseline Characteristics
From both groups combined except where specified. Comparisons are rate vs. rhythm control.
- Demographics: Age 70 years, female 40.6% vs. 37.9% (P=0.08), ethnic minority 11%
- PMH: HF 23%
- Predominant cardiac diagnosis: CAD 26%, cardiomyopathy 5%, HTN 51%, valvular disease 5%, other 1%, none apparent 12%
- Qualifying AF longer than 2 days: 69%
- First AF episode: 35%
- Pre-randomization failure of any antiarrhythmic: 18%
- Echocardiographic data:
- LVEF: 55% (normal LVEF 74%)
- Normal left atrium size: 35%
Interventions
Rate-Control Strategy
- Therapeutic target for heart rate at rest (<80 bpm) and during activity (<110 bpm), which usually consisted of six-minute walk test.
- Drugs acceptable for use included:
- Class II agents: beta-blockers
- Class IV agents: calcium-channel blockers (verapamil and diltiazem)
- Class V agents: digoxin
- Anticoagulation with warfarin (goal INR 2-3)
Rhythm-Control Strategy
- Anti-arrhythmic agent chosen by treating physician. Attempts to maintain sinus rhythm could include cardioversion as necessary.
- Drugs acceptable for use included:
- Class Ia agents: quinidine, procainamide, disopyramide
- Class Ic agents: flecainide, propafenone, moricizine
- Class III agents: amiodarone, sotalol, dofetilide
- Anticoagulation with warfarin encouraged but could be stopped at physician's discretion if sinus rhythm maintained for at least 4, and preferably 12, consecutive weeks.
After the failure of at least two trials of either a rhythm-control drug or a rate-control drug, patients could be considered for non-pharamcologic therapy such as radio-frequency ablation, maze procedure, and pacing techniques.
Outcomes
Comparisons are rate vs. rhythm control strategies.
Primary Outcomes
- 5-year mortality
- 25.9% vs. 26.7% (HR 1.15; 95% CI 0.99-1.34; P=0.08)
Secondary Outcomes
- Composite of death, disabling stroke, disabling anoxic encephalopathy, major bleeding, or cardiac arrest
- 32.7% vs. 32.0% (P=0.33)
- Hospitalization
- 73.0% vs. 80.1% (P<0.001)
- Ischemic strokes
- 5.5% vs. 7.1% (P=0.79), annual rate ~1% per year
- Mostly in whom warfarin had been stopped or in whom INR was subtherapeutic
- Primary intracerebral hemorrhage
- 1.1% vs 1.3% (P=0.73)
- Subdural or subarachnoid hemorrhage
- 0.8% vs. 0.8% (P=0.68)
- Disabling anoxic encephalopathy
- 0.2 vs. 0.4% (P=0.74)
- Torsades de pointes
- 0.2% vs 0.8% (P=0.007)
- PEA, bradycardia
- <0.1% vs 0.6% (P=0.01)
- Hemorrhage not involving CNS
- 7.7 vs. 6.9% (P=0.44)
Therapy
Combinations common; changes in therapy frequent.
- Rate control agents used for initial therapy
- Beta-blocker: 46.8% vs. 21.8%
- Diltiazem: 29.8% vs. 15.6%
- Verapamil: 9.6% vs. 4.4%
- Digoxin: 48.5% vs. 32.9%
- Rate control agents used at any time
- Beta-blocker: 68.1% vs. 49.6%
- Diltiazem: 46.1% vs. 30.0%
- Verapamil: 16.8% vs. 10.0%
- Digoxin: 70.6% vs. 54.4%
- Rhythm control agents used for initial therapy
- Amiodarone: 0.2% vs. 37.5%
- Sotalol: 0.1% vs. 31.2%
- Propafenone: 0.2% vs. 9.3%
- Procainamide: 0 vs. 5.3%
- Rhythm control agents used at any time
- Amiodarone: 10.2% vs. 62.8%
- Sotalol: 4.1% vs. 41.4%
- Propafenone: 2.2% vs. 14.5%
- Procainamide: 1.5% vs. 8.5%
- Use of warfarin
- >85% vs. ~70% (P<0.001)
- Sinus rhythm at 5 years
- 34.6% (>80% rate-controlled) vs. 62.6% (P<0.001)
- Cross-over rate (rate- to rhythm-control vs. rhythm- to rate-control)
- 14.9% vs. 37.5% (P<0.001)
Subgroup Analysis
- Rhythm-control strategy was associated with a higher risk of death than the rate-control strategy among:
- Older patients
- Patients with CAD
- Patients without HF
- After adjustment for prespecified subgroups, trend persisted toward higher risk of death in rhythm-control group than in rate-control group (HR 1.18; 95% CI, 0.99 to 1.41; P=0.07).
Adverse Events
- HF
- 2.1% vs. 2.7% (P=0.58)
- Pulmonary event
- 1.7% vs. 7.3% (P<0.001)
- Gastrointestinal event
- 2.1% vs. 8.0% (P<0.001)
- Bradycardia
- 4.2% vs. 6.0% (P=0.001)
- Prolongation of QT
- 0.3% vs. 1.9% (P<0.001)
Criticisms
- Possible selection bias: Patients with frequent or severe symptoms might have been considered unsuitable for a rate-control strategy and therefore may not have been enrolled by some investigators.
- Use of a single drug could have yielded a different result, but the ability to use multiple drugs increased the chance that any individual patient would maintain sinus rhythm
- Not generalizable to younger patients or to patients without risk factors for stroke (i.e. patients with primary, or "lone", AF), particularly those with paroxysmal AF.
- Those with successful rhythm control could have their anticoagulation stopped. This might have introduced bias in stroke and death outcomes as this led towards less use of anticoagulation in the rhythm control group.
Further Reading
Funding
Funding provided by the NHLBI and Wyeth-Ayerst Laboratories. Authors with significant disclosures.