A. Anticoagulation

Administer an anticoagulant regardless of whether AF is paroxysmal or permanent. AF often recurs, and asymptomatic recurrences are 12 times more common than symptomatic recurrences.¹⁴ In the AFFIRM study, the risk of ischemic stroke was strongly related to absent or suboptimal anticoagulation even in the rhythm-control strategy.¹⁵ Drug therapy makes AF recurrences shorter and slower, hence less symptomatic; the stroke risk, however, persists.

B. Rate control

Three classes of drugs may be used: β-blockers, non-dihydropyridine calcium channel blockers (CCBs) (diltiazem or verapamil), and digoxin.

β-blockers are effective as monotherapy in up to 70% of patients with AF,¹⁶ and are first-line therapy in compensated systolic HF. They also have an antiarrhythmic effect and may convert adrenergically mediated AF into sinus rhythm.

Digoxin is less effective for rate control during exertion. Digoxin is only used as monotherapy in decompensated systolic HF, when the acute initiation of β-blockers is not possible. Digoxin is effective in combination therapy: the combinations digoxin–β-blocker and digoxin–CCB are at least as effective and safe as the combination β-blocker–CCB.¹⁶

Two modern trials compared rate-controlling drugs:¹⁷ (i) RATAF compared monotherapy with CCBs vs. β-blockers; CCBs improved exercise capacity, symptoms, peak VO2, and NT-pro-BNP levels vs. β-blockers, despite a similar rate reduction. (ii)RATE-AF compared monotherapy with digoxin vs. bisoprolol (age ~75); digoxin improved functional class and NT-proBNP vs. bisoprolol, with a similar rate reduction and dramatically less adverse events and hospitalizations.

First-line treatment: CCB or β-blocker. Second-line: add digoxin or combine β-blocker + CCB. CCB is not an appropriate option in systolic HF. The combination of β-blocker and verapamil has a strong negative inotropic effect and should be avoided in all patients. Triple combination is required in ~15–20% of patients but increases the risk of excessive pauses. Pharmacological therapy can achieve rate control in >80% of patients.¹⁶ The remaining patients cannot be rate-controlled with drugs and require rhythm control, provided that long-term success can be expected, or failing that, atrioventricular nodal ablation with ventricular pacing. Patients with AF, including fast AF, may alternate with AF pauses, sinus pauses (immediately after AF converts to sinus rhythm), or sinus bradycardia; this limits the use of rate-controlling agents and requires tight rhythm control or pacemaker placement .

In symptomatic patients, the goal of therapy was to reduce heart rate to <80 bpm at rest, <110 bpm with moderate activity, and <100 bpm (average) on 24-hour Holter monitoring. This goal was adopted in the AFFIRM and RACE trials which established the efficacy of a rate-control strategy. However, in the RACE II trial, this strict rate control did not offer any benefit over a more lenient control (a resting heart rate <110 bpm at rest) in patients with permanent AF and mild or no AF-related symptoms.¹⁸ There was no significant difference in hard outcomes (death, hospitalization for HF, stroke) or in symptom control at 3-year follow-up. Note, however, that most patients in the lenient group had a heart rate <100 bpm (85 ± 14 bpm), which implies that 100 bpm may be a better lenient goal than 110 bpm.

Rate control in HF- A wealth of data indicates that strict rate control improves outcomes in systolic HF with sinus rhythm, but not in systolic HF with AF and not in HFpEF. A higher rate is necessary to compensate for the loss of atrial contribution to stoke volume; also, nocturnal pauses that accompany tight rate control may be particularly harmful in HF (induce VT, low flow). For AF associated with HF, the target rate control remains <100 bpm rather than <80 bpm (based on RACE-II substudy, substudies of β-blockers in HF, and Swedish HF registry).¹⁹ To improve HF outcomes, consider AF ablation rather than strict rate control.

C. Rhythm control

Rhythm control is generally the least important goal. Long-term rhythm control, as compared with rate control, did not reduce mortality, stroke rate, or HF hospitalizations in patients at high risk of stroke or AF recurrences (AFFIRM and RACE trials),^20,21 and in stable HF patients with EF <35% (AF-CHF trial).²² In fact, rhythm control was associated with a higher rate of hospitalization for recurrent AF and drug-related brady- and tachy-arrhythmias. In addition, in the AFFIRM trial, rhythm control was associated with increased non-cardiac mortality, particularly when amiodarone was used (pulmonary and cancer mortality).^23,24 Amiodarone was the most commonly used drug in the AFFIRM trial, while sotalol was the most commonly used drug in the RACE trial. These studies enrolled persistent AF (RACE), and paroxysmal or persistent AF (AFFIRM and AF CHF), and AF duration was mostly <1 year.

This failure of rhythm control was partly related to the withdrawal of anticoagulation, the marginal efficacy of antiarrhythmic drugs, and their toxicity: only 62% of patients in the rhythm-control arm of AFFIRM and 39% of the rhythm-control arm of RACE were in sinus rhythm on follow-up.²¹ Secondary analyses of these trials showed that the presence of sinus rhythm on follow-up was associated with improved quality of life (RACE)²⁵ and survival (AFFIRM),²³ regardless of the treatment strategy or the use of antiarrhythmic drugs. Antiarrhythmic drugs, per se, were associated with reduced survival.²³

Conversely, modern trials have shown that a rhythm control strategy is superior in 2 settings: HF or, less dramatically, early AF (≤ 1 year, symptomatic or not, paroxysmal or persistent); as long as a safer and more effective rhythm control regimen is used, focused on AF ablation, over a background of anticoagulation and comprehensive risk factor management:

• Compared to rate control or rhythm control with drugs, rhythm control with catheter ablation significantly and dramatically reduced both mortality and HF hospitalizations in systolic HF patients (CASTLE AF).²⁶
  
• Catheter ablation also showed a trend toward improvement in cardiovascular outcomes in all comers with AF, albeit modest (CABANA trial).²⁷
  
• In new AF ≤ 1 year, a stepwise rhythm control strategy focused on class Ic drugs (44%), dronedarone, AF ablation (20%), and less so, amiodarone (11–20%) slightly but significantly improved cardiovascular mortality by 0.3% per year, the overall cardiovascular outcomes by 1.1% per year, and stroke, compared to rate control. Sinus rhythm was maintained at 2 years in 82% of patients. This early AF control attenuates or reverses atrial cardiomyopathy.²⁸ (EAST AFNET4 trial: mean age 70, median BMI 28, median AF duration 36 days, LA diameter 4.4 +/- 0.8 cm, asymptomatic AF in 30% of patients, first AF episode in 38%, paroxysmal AF in 36%, HF in 28%).
  
• In ATHENA trial, dronedarone significantly reduced cardiovascular mortality (0.7%/yr), hospitalizations, and stroke compared to rate control.

In patients with BMI >27 kg/m², weight loss >10%, by itself, dramatically reduces the recurrence of AF. Weight loss, by itself, is almost as effective as antiarrhythmic drugs in preventing AF recurrence, and it allows interruption of antiarrhythmic drugs in many patients (~50% freedom from AF and antiarrhythmic drugs over several years of follow-up, vs. <20% if no weight loss). Also, weight loss is synergistic with AF ablation and antiarrhythmic therapy (87% freedom from AF over several years of follow-up after ablation, vs. <40% if no weight loss). This is enhanced by aggressive control of HTN, diabetes, lipids, sleep apnea, and smoking (LEGACY and ARREST-AF studies).^29,30

IV. Management of a patient who presents with acute, symptomatic AF

1. Direct-current cardioversion (DCCV) is indicated emergently in cases of severe shock, acute HF with severe pulmonary edema, or myocardial ischemia attributed to AF. Typically, to attribute severe hemodynamic compromise to AF and to justify urgent cardioversion, heart rate must be >150 bpm; or > 130 bpm in systolic HF. Look for potential causes of shock that might have triggered both hypotension and AF.
   
2. Acute rate control is achieved with β-blockers or CCBs (= first-line agents) in the absence of hypotension or decompensated HF. These agents are given intravenously if AF is severely symptomatic, with an initial target <110 bpm (Table 10.4).
   
   
   
   • Intravenous digoxin or amiodarone is a first-line option in decompensated systolic HF or borderline low BP (ESC and ACC class I). Amiodarone has the theoretical risk of acutely cardioverting AF in a patient who may have an atrial appendage thrombus, but the short-term (24 hours) likelihood of cardioversion is low, particularly in HF or critical illness. Neither ACC nor ESC guidelines insist on this risk.
     
   • β-Blockers may be used acutely in low EF without clinical HF and in HFpEF.
     
   • The onset of action of IV drugs is fast (a few minutes), except for digoxin (1–5 hours).
   
   
3. Acute AF often spontaneously resolves by 24 hours, in ~60% of the cases (70% at 48 hours), even more so in the absence of heart disease. Thus, a 24-hour wait-and-see strategy is reasonable.³⁴ If AF persists beyond 24 hours:
   
   
   
   1. If a rhythm-control strategy is selected, perform acute cardioversion at 24 hours (DCCV or drugs), after TEE rules out left atrial appendage thrombus. The patient needs to receive acute anticoagulation with heparin, enoxaparin, or NOAC. Another option is to provide oral anticoagulation for 3 weeks then perform DCCV without TEE.

      
      
      
      
      

      Metoprolol 5 mg IV, to be repeated Q5 min up to 15 mg total. Then start oral metoprolol 25–50 mg Q6h if the IV doses prove effective and tolerated. The IV metoprolol effect lasts a few hours. Alternatively, esmolol IV drip may be given (short-acting; effect lasts 10–20 min after the drip is off). Esmolol is the preferred agent if it is uncertain whether β-blockers will be hemodynamically tolerated Diltiazem IV 0.25 mg/kg (~20 mg) bolus. May re-bolus with 0.35 mg/kg (~25 mg) in 15 min if no response to 20 mg, then start 5–15 mg/h IV drip in responders. The IV diltiazem effect lasts a few hours.a In comparison with β-blockers, diltiazem has: more hypotensive effects in general (vasodilatation) similar hypotensive effects in pre-shock, wherein BP depends on both the adrenergic and vascular tone less negative inotropic effect (than verapamil as well), but still not advised in systolic HF Digoxin needs to be given as a load (0.5 mg, then 0.25 mg Q6–8 h × 2 = 1 mg load, oral or IV). The load is followed by 0.125–0.25 mg Qday

      
      

      Acute monotherapy with β-blockers or CCBs is effective in most cases, the target ventricular rate in the acute setting being usually 80–110 bpm (ESC). In case of total non-response to one agent, switch to another; in case of partial response to one, combine two agents, then try to wean off the first one. β-Blockers are more effective in hyperadrenergic states (e.g., postoperatively). Avoid digoxin as a sole agent to control the heart rate, except with decompensated heart failure or low blood pressure.

      
      

      ^a IV diltiazem may be used for a few hours, then oral diltiazem started 3 hours after discontinuation of the IV drip, with a daily oral dose equal to (mg/day): IV drip rate in mg/h x 30 + 30. Short-acting diltiazem may initially be used Q6h.

      
      
   2. Alternatively, for acute AF with symptoms <12–48 hours, immediate pharmacological cardioversion may be attempted with intravenous procainamide (30-min infusion, ~60% success rate, rare hypotensive effect) (RAFF2 trial); or oral flecainide load (300 mg, ~70% efficacy at 3–6 hours).^34,35 This acute cardioversion allows early patient discharge.
      
   3. May allow the progression to a rate-controlled permanent AF. This depends on the severity of AF-related symptoms, their persistence despite rate control, the age of the patient (young → cardiovert), and the likelihood of AF recurrence.
   
   
4. Hospitalize the patient with severe symptoms or HF. Also, hospitalize if DCCV is planned.
   
5. Perform an echocardiogram and look clinically for underlying cardiac or systemic diseases: acute HF, MI, PE, bleeding, thyrotoxicosis. Correcting these factors (e.g., diuresis and vasodilators for HF) may control the rate and eventually convert AF to sinus rhythm. In fact, DCCV is often ineffective in the presence of uncontrolled triggers or uncontrolled HF, as AF quickly recurs after cardioversion.

V. Peri-cardioversion anticoagulation management

The risk of stroke in the peri-cardioversion period is ~8%, reduced to <0.5% with appropriate anticoagulation. For an AF episode that has lasted ≥48 hours:

1. TEE or 3 weeks of anticoagulation is required before DCCV.
   
2. Anticoagulation is required for at least 4 weeks after spontaneous or active cardioversion, regardless of whether the patient’s risk factors dictate long-term anticoagulant therapy.² AF conversion to sinus rhythm is followed by a few weeks of atrial hypocontractility, during which a thrombus may form then embolize once the atria fully recover their contractility. If the patient does not have an indication for chronic anticoagulant therapy, administer an oral anticoagulant for 4 weeks only. If warfarin is initiated, bridging with LMWH is necessary during the critical first few days after cardioversion.

For acute AF<48 hours- Cardioversion of AF that has lasted <48 hours does not mandate TEE or oral anticoagulation before DCCV; heparin or one dose of LMWH is provided peri-cardioversion.^2,31,36 The issue is, however, that AF may have been asymptomatic for some time before symptoms developed; symptoms may correspond to acceleration of AF rather than its onset. In addition, one study has found that as many as 4% of patients with AF <48 hours have LAA thrombus, particularly if underlying heart disease is present.³⁷ A registry analysis found a stroke rate of ~1.1% after cardioversion of AF<48 hours, exclusively in patients with CHA2DS2-VAS ≥2.³⁸ A Finnish study found a similar risk of 1.1% in patients with AF duration of 12–48 hours (cardioversion much safer at <12 hours).³⁹ Thus a safe approach may be to perform TEE even if the presumed AF duration is <48 hours, except in patients with AF <12 hours or CHADS2-VAS of 0 or 1, as per ESC and Canadian guidelines, and to mandate 4 weeks of anticoagulation after DCCV even if AF <48 hours.^31,40,41 ACC guidelines do not mandate TEE when AF is definitely <48 hours, regardless of thromboembolic risk factors (class IIa); ACC and ESC accept forgoing the 4 weeks of anticoagulation when AF <24 hours with CHADS2-VAS of 0 (class IIb).³⁶

If the initial TEE shows LAA thrombus, anticoagulation should be given for at least 3 weeks, followed by a repeat TEE to document thrombus resolution before cardioversion. If thrombus persists, accept a rate-control strategy.³¹

VI. Antiarrhythmic management after the acute presentation

In 25% of patients, DCCV fails or AF recurs after a few seconds or minutes of sinus rhythm; in another 25%, AF recurs within 2 weeks. Distinguish DCCV failure from immediate recurrence: in failure, no sinus conversion occurs at any time, not even for a beat, whereas in immediate recurrence, one or few sinus beats are seen before recurrence of AF. Failure is treated with a higher DCCV energy, a change in pad position/defibrillation vector, IV ibutilide pretreatment, or by pushing down the pads over an obese chest to improve the energy penetration. Immediate or early recurrence may benefit from a second attempt at DCCV after preparation with an antiarrhythmic drug. Antiarrhythmic drugs reduce immediate recurrence but do not help with DCCV failure, as most of them actually increase the atrial defibrillation threshold (except ibutilide, sotalol and dronedarone).

The AF recurrence rate is 70–80% at 1 year after cardioversion. Even when antiarrhythmic agents are used, the recurrence rate is ~35–60% at 1 year (~half),⁴² this rate being higher in patients with multiple risk factors (Table 10.3).^33,43 Except for lone AF, most patients with paroxysmal AF progress to persistent or permanent AF within a few years. This high failure rate underlines why a rate-control strategy is the most realistic strategy in patients with many risk factors for recurrence or progression, even more so if DCCV fails or if AF recurs within 2 weeks despite antiarrhythmic therapy. If a rhythm-control strategy is selected, antiarrhythmic therapy may be considered after a recurrence rather than the first episode in lower-risk patients, and a shorter antiarrhythmic duration is desirable, as most recurrences occur early on (e.g., 1 month of flecainide) (ESC guidelines);³¹ this may be followed by an as-needed, pill-in-the-pocket cardioversion of recurrences with flecainide.

The goal of a rhythm-control strategy is symptom control and quality of life improvement; thus, short and infrequent recurrences or mildly symptomatic recurrences of paroxysmal AF are not considered treatment failures in a patient who previously had severely symptomatic paroxysmal or persistent AF. Rate-slowing drugs should be continued throughout the rhythm-control approach to ensure adequate rate control during AF recurrences, unless the baseline sinus rate is slow.

Off-and-on, fast AF is typically more difficult to rate control than permanent AF, as slowing AF spells leads to excessive slowing of the baseline sinus rate; this AF frequently requires tight rhythm control, including catheter ablation, or pacemaker placement.

An alternative to antiarrhythmic therapy is catheter isolation of the pulmonary veins, which successfully prevents AF in 70% of the treated cases over 1 year of follow-up.^2,12,44 It is indicated for the prevention of recurrent, symptomatic, paroxysmal or persistent AF, especially when the left atrial diameter is <4 cm, as second-line or even first-line therapy (Appendix 3).^2,12 If it fails, thoracoscopic surgical ablation is a next option. As a last resort, AV nodal ablation with ventricular pacing may be required.

VII. Decisions about long-term anticoagulation, role of clopidogrel, role of triple therapy

A. Long-term anticoagulation

The annual risk of stroke in patients with AF not receiving any antithrombotic therapy ranges from 1% to 18%, and averages ~5% per year.⁴⁵ The risk is increased in both paroxysmal and permanent AF, although less so if AF burden is low or if AF is paroxysmal, brief and infrequent (Section X). Two validated clinical schemes, CHADS₂ and CHA₂DS₂-VAS scores, help predict this risk (Table 10.5).⁴⁵ CHA₂DS₂-VAS improves the predictive value of CHADS₂ and is particularly useful for further classifying a CHADS₂ score of 0 or 1.^31,46,47 In fact, some patients with CHADS₂ score of 0 or 1 have a high stroke risk. For any level of CHADS₂ score, warfarin reduces the stroke risk by 75% and increases major bleeding ~2.5-fold.⁴⁸The absolute stroke reduction with warfarin only overcomes the bleeding risk when the stroke risk is ≥1.7% per year; the equipoise point is 0.9% with the non-vitamin K oral anticoagulants (NOACs).⁴⁹ Aspirin has no effect on AF-induced stroke, and only reduces vascular-induced stroke by 21% while increasing major bleeding by 60%.^2,31,48

In the ESC and ACC guidelines, no antithrombotic therapy (not even aspirin) is recommended for patients with a CHA₂DS₂-VAS score of 0 (not counting female sex), where the absolute stroke reduction with anticoagulation is marginal (≤0.5%), not justifying the bleeding risk. Anticoagulation is recommended for a score ≥2 (not counting female sex), and NOACs are recommended over warfarin. For a CHA₂DS₂-VAS score of 1 (not counting female sex), ACC and ESC guidelines suggest oral anticoagulation after assessment of bleeding risk (class IIa).^2,31,36 In fact, for a score of 1, the benefit of anticoagulation is marginal, as the absolute stroke risk is 1.5–2% per year; the stroke reduction only slightly outweighs the bleeding risk. Anticoagulation is likely beneficial if the risk factor is age 65–74 (stroke risk ~3% per year), if the risk factor is severe (uncontrolled diabetes, severe HF), or if a NOAC is used.⁵⁰ Aspirin is not indicated for AF purpose in any subgroup (ESC class III).

CHADS₂ score does not apply to patients with valvular AF, who have a high stroke risk and require anticoagulation with warfarin regardless of the score. Warfarin is used in these patients, rather than NOACs. Also, regardless of CHADS₂ score, patients with hypertrophic cardiomyopathy and AF have a high stroke risk that warrants anticoagulation.

CHADS2 score		CHA2DS2-VAS score		Yearly stroke risk according to CHADS2 scoreb	Yearly stroke risk according to CHA2DS2-VAS scoreb
C: Congestive heart failurea	1	C: Congestive heart failure	1	0: 1.9%	0: <1%
H: Hypertension	1	H: Hypertension	1	1: 2.8%	1: 1.5–2%
A: Age ≥75	1	A: Age ≥75	2	2: 4.0%	2: 2.5–3%
D: Diabetes	1	D: Diabetes	1	3: 5.9%	3: 3.2%
S: prior stroke or TIA	2	S: prior stroke or TIA	2	4: 8.5%	4: 4%
		V: Vascular disease (CAD, PAD, aortic plaque)	1	5: 12.5%	5: ~7%
		A: Age 65≥74	1	6: 18.2%	6–8: ~10%
		S: Sex femalec	1		9: 15%

^a “C” mandates either one of the following (ESC): (i) reduced EF, or (ii) documented HF decompensation, whether EF is reduced or normal (i.e., diastolic HF included).

^b Numbers obtained from references 31 and 45.

^c Female sex conveys increased risk only in patients who already have a high stroke risk, not in lower-risk patients with a non-sex related score of 0 or 1. Hence, female sex is only counted in patients with ≥2 non–sex-related risk factors. ³⁶

VIII. Special situation: atrial fibrillation and heart failure. Optimal heart rate in heart failure

A. General management

In a patient presenting with decompensated HF, AF may be:

1. secondary to HF
   
2. the cause of the cardiomyopathy and HF (tachycardia- or AF-mediated cardiomyopathy)
   
3. the major trigger of HF decompensation in a patient with a previously stable cardiomyopathy

It should be assumed that either (b) or (c) is the case, but at the stage of decompensated HF acute DCCV may not be helpful.

In compensated HF, AF contributes to ~30–40% of cardiac output. Those patients have a large A wave with E/A reversal on echo, confirming the crucial role of atrial contraction. In fast AF, the loss of atrial systole and the reduction of diastolic filling time reduce cardiac output, which increases LA pressure and decompensates HF. However, once HF is decompensated, even if the patient converts back to sinus rhythm, the contribution of atrial systole to cardiac output will be marginal. In fact, decompensated patients have a very small A wave on echo, as the high LV pressure does not allow any filling during atrial contraction (Figure 10.1). At this point:

• Tachycardia of 100–110 bpm may be helpful, as it increases cardiac output and LV emptying by increasing the number of cardiac cycles per minute. Cardiac contractility also increases up to a heart rate of 100–110 (Bowditch phenomenon).
  
• Tachycardia >120 bpm is however counterproductive and reduces the contractility of the energy-depleted myocardium, cardiac output and VO₂ (particularly HFrEF).^73,74 AF needs to be acutely rate controlled to <110–120 bpm (ESC <110 bpm).
  
• Tachycardia >130 bpm with severe hemodynamic compromise (shock, severe pulmonary edema) often needs acute DCCV, particularly in systolic HF, to improve LV energetics and contractility.

In addition, treatment of HF with diuresis and vasodilators slows down AF and possibly converts it (situation (a) above). Once HF is stabilized and diuresed, the goal becomes to slow down AF to a rate <100 bpm at rest and better yet, cardiovert it;¹⁹ at the compensated stage, the atrial kick increases cardiac output and reduces HF symptoms, and sinus rhythm reverses AF-mediated cardiomyopathy.

IX. Special situation: atrial fibrillation with borderline blood pressure or non-AF-related hypotension

In critical illness, a fast rate of up to 120 bpm may be appropriate and tolerated. In patients with borderline BP, or full-blown shock not caused by AF (e.g., septic shock with AF at 120–150 bpm), IV amiodarone may be used for rate control <120 bpm (ACC and ESC). Amiodarone may precipitate hypotension but only during fast boluses (<30 min). Amiodarone has limited rhythm controlling effect in the first 24–48 hours. Digoxin IV may also be used but its vagal effect is ineffective in high catecholamine states.

β-Blockers may be cautiously used in the absence of a pre-shock state, wherein the patient is dependent on the adrenergic tone and wherein β-blockers drop BP precipitously (e.g., severe hypovolemia, sepsis, bleeding). Esmolol IV, being quickly reversible, is the preferred β-blocker in this instance.

AF with clear-cut shock and a rate >150 bpm (or 130 bpm in systolic HF) is often contributive to the shock and DC cardioverted emergently, unless a clear cause of hypotension is identified, such as massive gastrointestinal bleed. IV procainamide may be attempted for acute cardioversion, even in HF, to avert the need for DCCV’s sedation (except in extremis).