Atrial Fibrillation: Current Evidence and Management Strategies During the Perioperative Period

Kunal Karamchandani, MD, FCCP; Ashish K. Khanna, MD, FCCP, FCCM; Somnath Bose, MD; Rohesh J. Fernando, MD, FASE; Allan J. Walkey, MD, MSc


Anesth Analg. 2019;130(1):2-13. 

In This Article

Patient-related Interventions

Addressing modifiable patient factors can also help prevent the development of POAF. Perioperative β-blockade effectively reduces POAF in patients after cardiac surgery,[65] and continuing β-blockers in patients chronically on β-blockers likely has some effect on POAF. β-Blockers have shown mixed results as chemoprophylaxis against perioperative AF. The 2008 POISE trial randomly assigned over 8000 patients undergoing major noncardiac surgery to receive either metoprolol or placebo before surgery.[18] Clinically significant new POAF was recorded in 2.2% of patients in the metoprolol group versus 2.9% in the placebo group (hazard ratio [HR], 0.76; 95% CI, 0.58–0.99). In light of an increased risk of mortality and stroke in POISE, and criticisms surrounding the dose of β-blockade in the treatment arm, the clinical relevance of this reduction in POAF with β-blockers is unclear. Early resumption of β-blockers in patients chronically taking these medications remains of critical importance as well. In a 2018 review of >8000 surgical patients receiving β-blockers chronically, the risk of de novo and paroxysmal POAF was reduced if β-blockade was resumed by the end of postoperative day 1 after noncardiothoracic and nonvascular surgery.[66] However, resuming β-blockers on postoperative day 0 was not associated with a decreased risk of POAF. This lack of effect with day 0 β-blockade could have been due to the low event rate on postoperative day 0, or an extension of the effects of long-acting preoperative β-blockers. Further study is needed to better understand how the postoperative timing of β-blockade resumption may affect POAF in the postoperative period (Table 2).

In patients with OSA, use of continuous positive airway pressure (CPAP) ventilation prevents obstructive respiratory events, reducing sympathovagal activation and reversing atrial remodeling.[67,68] Moreover, CPAP decreases the risk of transition from paroxysmal to persistent AF among OSA patients.[69] Hence, in patients with OSA, postoperative CPAP may reduce the risk of POAF by addressing mechanistic factors known to play a role.

Management of POAF

Preoperative Period. Perioperative physicians often wrestle with canceling surgery for additional workup in patients who present for surgery with AF. Because paroxysmal AF is common and often undetected in the general population,[70] it is also often unclear if the arrhythmia is new or preexisting. Existing American College of Cardiology/American Heart Association guidelines recommend that new-onset arrhythmias in the preoperative setting should prompt investigation into underlying causes, including cardiopulmonary disease, ongoing myocardial ischemia or myocardial infarction, drug toxicity, and endocrine or metabolic derangements.[71] However, they also clarify that the paucity of studies prevents specific evidence-based recommendations. If time and resources permit, cardiology consultation may help identify high-risk patients. Ultimately, the decision to cancel or postpone for workup of AF should be made on a case-by-case basis and include discussions with the surgical team. In patients with preexisting AF and RVR, IV diltiazem or β-blockers are reasonable choices for heart rate control and, if unsuccessful, delaying elective surgery should be considered in patients with other comorbidities such as hemodynamic instability, acute myocardial ischemia/infarction, congestive heart failure, or pulmonary embolism (PE).[72] Cardiology consultation may be helpful in identifying underlying pathology and managing these complex patients. Patients with clinically and hemodynamically stable rate controlled AF generally do not require modification of medical management, special evaluation in the perioperative period, or delay of surgery.[71] Patients with AF who have been cardioverted in the past may benefit from an electrocardiogram (ECG) before surgery for detection of recurrence, which may be as high as 50%.[73]

Preoperative medications for rate control should be continued until the day of surgery, while decisions on continuation of anticoagulation are patient- and procedure-dependent. In such cases, the potential for perioperative ischemic stroke needs to be weighed against the risk of perioperative bleeding. A large population-based 2017 study found no difference in 30-day mortality rates in patients with AF who underwent either urgent or elective surgeries with or without preoperative anticoagulation.[74] Similarly, 30-day mortality did not differ among patients with AF who were treated with either warfarin or a direct oral anticoagulant (DOAC). Bleeding and thrombotic risks were similar between those treated with either warfarin or DOACs. Perioperative management of anticoagulation in patients with POAF is thus best made on a case-to-case basis with input from surgical and cardiology teams. Evidence-based decision tools may help clinicians with periprocedural management of anticoagulation in patients with nonvalvular AF.[75] One approach on withholding versus continuing anticoagulant therapy, for patients on vitamin-K antagonists (VKAs) and DOACs, based on the 2017 ACC expert consensus, is depicted in Figure 3.

Figure 3.

Preoperative management of anticoagulation in patients taking VKAs and DOACs. *Bleed risk is considered increased in the presence of any 1 of the following: major bleed or intracranial hemorrhage within 3 mo; quantitative or qualitative platelet abnormality, including aspirin use; supratherapeutic INR; prior bleed during previous bridging or similar procedure. DOAC indicates direct oral anticoagulants; INR, international normalized ratio; VKA, vitamin-K antagonists.

The decision to bridge VKAs and DOACs with unfractionated heparin (UFH) or low-molecular-weight heparin (LMWH) is based on the patient's thrombotic risk, bleeding risk, renal function, and the risk of procedural bleeding.[75] The timing of either UFH or LMWH is best determined in consultation with the surgeon. The decision to use UFH rather than an LMWH as the bridging agent depends on renal function, the bridging setting (inpatient versus outpatient), and patient comfort with self-injections. Current expert consensus suggests that if creatinine clearance (CrCl) is <30 mL/min, UFH is preferred over LMWH especially in the periprocedural setting.[75] UFH may be discontinued 4–6 hours before the procedure or earlier based on the activated partial thromboplastin time. If LMWH is used for bridging, discontinuation is recommended at least 24 hours before the procedure.[75] Although perioperative bridging with a parenteral anticoagulant is common, current evidence suggests that such a practice may not decrease the risk of arterial thromboembolism, and may increase the risk of both major adverse cardiovascular events and major bleeding.[76] A simple and standardized perioperative DOAC interruption and resumption strategy were recently tested in patients with AF undergoing elective surgery.[77] This approach, which avoided heparin bridging and coagulation function testing, showed low rates of bleeding as well as arterial thromboembolism and could be a viable alternative in patients taking DOACs.

Intraoperative Period. For AF occurring intraoperatively, immediate management depends first on heart rate and blood pressure. RVR combined with hypotension refractory to vasoconstrictor therapy requires emergent electrical cardioversion. For patients with RVR in whom blood pressure remains adequate, rate control (HR < 110/min) can be attempted using β-blockers or Ca2+ channel blockers. Metoprolol is superior to diltiazem in achieving rate control in critically ill patients with RVR and maybe a better perioperative option.[78] While targeting rate versus rhythm control does not affect outcomes in ambulatory[79] or postcardiac surgery AF,[80] no data exist for noncardiac surgical patients. Patients undergoing surgery and anesthesia may be more susceptible than ambulatory patients to hypotension from rate-control agents and rhythm control may be an option if NSR is documented within the last 48 hours. One strategy for rate control is to assess the clinical response to a short-acting β-blocker such as esmolol. If the patient responds with a decrease in heart rate, and no hypotension, either a small dose of metoprolol or an esmolol infusion can be used.

In patients with RVR who cannot tolerate β or Ca2+ channel blockers, amiodarone is a reasonable alternative. An initial 150 mg intravenous bolus, followed by a 1 mg/min infusion, can be started. If the heart rate remains high, an additional 150-mg bolus dose can be given. Phenylephrine is another alternative in patients with RVR and low blood pressures because it reflexively reduces heart rate. No data exist to inform the target heart rate that should be achieved during surgery, but a heart rate <110 bpm is usually acceptable if perfusion is not impaired.[28]

If possible, causes of AF and modifiable risk factors should be evaluated while stabilizing heart rate and blood pressure. Volatile agents do not differ from total intravenous anesthesia (TIVA) with respect to AF risk but desflurane is associated with increased sympathetic stimulation and potential for arrhythmias.[81] Fluid status should be assessed and hypovolemia or hypervolemia, if present, should be addressed. Intraoperative transesophageal echocardiography (TEE) may help rule out AMI and guide fluid management in high-risk patients. If possible, checking electrolytes will permit assessment of metabolic abnormalities such as acidosis, hypomagnesemia, and hypokalemia. Although no clear threshold for replacement exists, a 1999 multicenter review of 2402 patients suggests an increase in arrhythmias with potassium levels <3.5 mg/dL.[82] If a central line was recently inserted, the position of the catheter should be confirmed because catheter tip irritation can precipitate AF.

Postoperative Period. If AF persists after surgery or begins in the postoperative period, the first step would again be the preservation of blood pressure and end-organ perfusion and heart rate control if RVR is present. An effort to identify potentially dangerous causes should proceed concurrently with hemodynamic stabilization. Bedside ultrasound may provide an estimate of the left ventricular function and potentially exclude AMI and/or PE as a cause. A high suspicion of AMI and/or PE might warrant further evaluation by checking serum troponin levels and radiographic evaluation. If the patient was chronically taking a β-blocker or nondihydropyridine Ca2+ channel blocker (NDHP-CCB) preoperatively, restarting the patient's home regimen and/or adjusting the dose may also be appropriate.

As with intraoperative AF, β-blockers, NDHP-CCBs (ie, diltiazem and verapamil), digoxin, and amiodarone can be used to achieve ventricular rate control in the postoperative period. The choice of β-blockers versus diltiazem is frequently debated among clinicians with no clear evidence favoring one over the other. In a substudy of the Atrial Fibrillation Follow-up Investigation of Rhythm Management (AFFIRM) trial, β-blockers were more efficacious at achieving goal heart rate than Ca2+ channel blockers.[83] In addition, because vasodilation with diltiazem may cause hypotension in the perioperative setting and the postsurgical period is already associated with an increased adrenergic tone and catecholamine release, β-blockers may represent a physiologically appropriate first-line therapy in patients who are asymptomatic and have preserved ejection fraction.[84–87] Amiodarone is also a viable option for rate control in the postoperative setting, particularly with borderline low blood pressures as the hypotensive effect of amiodarone is less. A loading dose followed by an infusion is usually administered given its long half-life and extensive tissue distribution.[85,87] Amiodarone administration can be associated with significant side effects both acutely as well as with chronic use. Acute pulmonary and hepatic toxicity are rare but potentially life-threatening complications and should be considered in patients with acute respiratory compromise and transaminitis after amiodarone administration. The drug should be discontinued if there is more than a 2-fold elevation.[88] In the long term, the drug affects the respiratory system, liver, and the thyroid gland predominantly, but photosensitivity, corneal deposits, and neurological side effects are also frequent. Amiodarone-induced pulmonary toxicity (APT), with incidence ranging from 5% to 13%, can manifest acutely as organizing pneumonia or acute respiratory distress syndrome (ARDS) and chronically as interstitial pneumonitis or pulmonary fibrosis.[89] Potential risk factors include a high cumulative dose, a daily dose >400 mg/day, duration of therapy exceeding 2 months, increased patient age, preexisting lung disease, and thoracic or nonthoracic surgery.[90–93] Hence, it is preferable to avoid amiodarone in patients with preexisting lung disease and/or those undergoing thoracic surgery. While hypo- and hyperthyroidism and hepatic injury are common side effects of chronic amiodarone use, acute exacerbation of thyroid or liver disease with amiodarone administration has not been described. Amiodarone can be administered to patients with preexisting thyroid or liver disease, albeit with caution and close monitoring of thyroid and hepatic function.

Digoxin represents another option for achieving rate control. However, it is less effective in states of elevated sympathetic activity such as the perioperative period,[84–87,94] and has a narrow therapeutic/toxic window when compared to β or Ca2+ channel blockers and amiodarone. In addition, because renal function is often compromised in the perioperative setting, digoxin may accumulate as its clearance is predominantly renal.[95] Electrolyte imbalances such as hypomagnesemia, hypercalcemia, hypernatremia, and hypokalemia can also alter the effects of digoxin. Hence, digoxin should generally be considered in the perioperative setting only when other pharmacologic options are unsuccessful or contraindicated.

Most patients with new-onset POAF convert to NSR before hospital discharge and >95% of patients remain in NSR 2 months after surgery.[96] These patients should be discharged on rate-control agents unless contraindicated and outpatient follow-up is indicated to determine if medication regimens need to be adjusted or discontinued. For patients who develop POAF after thoracic surgery, 2014 American Thoracic Society guidelines recommend that antiarrhythmic therapy be continued for a minimum of 1 week and no longer than 6 weeks beyond the time of discharge.[97] However, no data support a specific strategy in patients undergoing noncardiothoracic surgery. In view of the significant side effect profile of amiodarone when used over long term, it may be ideal to either convert to a β-blocker or NDHP-CCB before hospital discharge or schedule a follow-up appointment with a cardiologist to assess discontinuation as soon as possible.

Antithrombotic Management

Antithrombotic therapy should be considered on a case-by-case basis in patients who develop POAF. Patients with paroxysmal AF have a stroke risk similar to those with persistent or permanent AF.[85,98] Although current literature suggests no clear association between the duration of AF and stroke,[99,100] the risk of stroke for secondary AF such as POAF is similar to traditional AF.[101] In patients with an AF burden detectable by clinical examination or 12-lead ECG at multiple time points, anticoagulation therapy reduces stroke risk.[102] Similarly, when appropriate surgically, anticoagulation should be restarted in patients with chronic or paroxysmal AF who were receiving anticoagulation before surgery. Perioperative physicians should consider the risk of bleeding associated with the surgical procedure, the patient's general bleeding risk, and the patient's thromboembolic risk on an individual basis when determining the need for starting or reinitiating antithrombotic treatment. Although surgery-specific bleeding risk is difficult to assess, the HAS-BLED (hypertension, abnormal renal/liver function, stroke, bleeding history or predisposition, labile international normalized ratio, elderly, drugs/alcohol concomitantly) score can be used to determine the general bleeding risk of the patient.[103] A HAS-BLED score of >3 indicates high bleeding risk and higher scores correlate with higher risk of bleeding.[85,104]

Clinical scoring systems may help predict the risk of stroke and systemic embolism in patients with AF so that a risk-benefit calculation can be made with respect to bleeding risks associated with anticoagulant therapy. The Congestive heart failure, Hypertension, Age older than 75 years, Diabetes and Stroke (CHADS2) score is the most widely used, with higher scores correlating with higher risk of thromboembolism.[84,85,87,96,104–106] The more recent Congestive heart failure, Hypertension, Age older than 75 years, Diabetes, Stroke, Vascular disease, female Sex (CHA2DS2-VASc) score adds an extra point each for female sex and vascular disease (which includes both coronary heart disease and peripheral vascular disease), and divides age into 3 categories (<60 years: 0 point, 60–74 years: 1 point, and ≥75 years: 2 points) instead of the 2 categories in the original CHADS2 score. Oral anticoagulation is indicated in men with CHA2DS2-VASc scores ≥1 and women with scores ≥2.[107]

Although the CHADS2, CHA2DS2-VASc, and HAS-BLED tools have been validated in general medical patients, they have not been validated in postsurgical patients. Even though the pathophysiology of thrombus development caused by AF is similar in both settings, surgical patients are at a high risk for bleeding. Because the overall rate of thromboembolism is low in both bridged and nonbridged patients in recent studies,[76,108] waiting for adequate hemostasis before initiating anticoagulation is reasonable and the decision to restart anticoagulation is best made on a case-by-case basis. Recent studies suggest that real-world application of these risk scores is inconsistent and adequate anticoagulation in at-risk surgical patients is often lacking.[109]