ATRIAL FIBRILLATION (AF) IS COMMON IN PATIENTS following major surgical procedures of any type (1). This is particularly true after open heart surgery in adults in whom there is about a 20% to 50% incidence of supraventricular arrhythmias, regardless of the presence or absence of preoperative arrhythmias (2). Information and attitudes concerning all aspects of postoperative AF, including epidemiology, clinical significance, pathogenesis, prophylaxis and management, are remarkably controversial. This paper focuses on AF occurring as a complication of cardiac surgery and briefly outlines current views on AF following noncardiac surgical procedures.


EPIDEMIOLOGY

Incidence: The reported incidence of AF occurring after cardiac surgery has varied considerably (2-4). This is due to differences among studies in the definition of the arrhythmia, the methods used to detect it, the underlying cardiac disease and the surgical procedure performed. In most series, only patients undergoing coronary bypass surgery have been considered, and many types of supraventricular arrhythmias have been included in the data analysis. Hashimoto et al (5) reported an incidence of supraventricular arrhythmias of 23% in 800 patients undergoing coronary bypass grafting at the Mayo clinic, but AF accounted for only 61% of these cases. In a meta-analysis of 31 randomized trials on prophylactic regimens, Andrews et al (4) reported an incidence of 26.9% of supraventricular arrhythmias in the pooled proportion of patients in the control group of all trials, varying from 41.3% in trials using a period of Holter monitoring to 19.9% in those not using Holter monitoring. These authors pointed out that Holter monitoring is likely to detect more episodes of arrhythmias, but many of them will be shorter in duration and of questionable clinical significance. More recently, Creswell et al (3), using continuous bedside monitoring, reported a global incidence of atrial arrhythmias of 34.6%, ranging from 9% to 92% according to the type of surgery performed. The distribution of atrial arrhythmias in their series was as follows: 9% after excision of cardiac tumour; 11% after cardiac transplantation; 15% after closure of atrial septal defect; 32% after coronary bypass alone; 42% after mitral valve replacement; 49% after aortic valve replacement; 62% after combined coronary bypass and valve replacement; and 92% after combined coronary bypass and double valve replacement.

Clinical significance: AF is, therefore, one of the most common complications of cardiac operations. Despite that, many surgeons believe that it is a benign complication without long term sequelae. Several repoorts (2) yielded conflicting results, especially with respect to hemodynamic consequences (hypotension and heart failure). While the adverse hemodynamic effect of AF after cardiac operations is always a concern, it has not been clearly documented in the literature, and appears to be rather unusual. On the other hand, the association with embolic stroke must be of higher concern. In their large retrospective study, Creswell et al (3) showed an increased incidence of stroke in patients with postoperative AF (3.3% versus 1.4%). Another study from the Massachusetts General Hospital showed that the odds ratio for stroke with postoperative AF was 3.0 (6). Although these data may not be conclusive, they raise the possibility that an association between stroke and postoperative AF may exist. The study by Creswell et al (3) also noted a longer stay in the intensive care unit (5.7 versus 3.4 days) and in the postoperative ward (10.9 versus 7.5 days), an increased incidence of ventricular tachycardia and fibrillation (9.2% versus 4%), and an increased need for implantation of a permanent pacemaker (3.7% versus 1.4%) in patients with postoperative AF, but not an increased rate of operative mortality. Landymore et al (7) showed that AF rarely recurred over the long term and concluded that treatment for more than three weeks was not recommended.

Clinical correlates: Precise knowledge of the conditions which increase the likelihood of postoperative AF may avoid subjecting all patients undergoing cardiac operations to unnecessary prophylactic measures and may also give insight into the pathophysiology. A number of investigators attempted to identify the risk factors associated with postoperative AF, but those studies have yielded conflicting results, likely due to the small sample size of most studies (2,3,5,8,9). The only factor reproducibly identified as an independent predictor of postoperative AF was age, with an increasing prevalence in older patients. In at least two studies (10,11), the use of beta-blockers preoperatively has been identified as a significant correlate of postoperative AF. However, this was not the case in the larger study of Creswell et al (3), in which many other factors were identified as independent predictors of postoperative AF: age, history of rheumatic heart disease, preoperative use of digoxin, chronic obstructive pulmonary disease and increasing aortic cross-clamp time. Interestingly, the following factors, often related to the severity of left ventricular dysfunction or the length of operative procedure, were identified by univariate analysis: increasing number of previous myocardial infarctions, unstable angina, previous cardiac surgery, current smoking habit, hypertension, chronic renal insufficiency, decreasing left ventricular ejection fraction, increasing left ventricular end-diastolic pressure and increasing cardiopulmonary bypass time.


PATHOGENESIS

The mechanism of AF is generally explained by the Ômultiple wavelet theoryÕ developed by Moe (12). It consists of multiple small reentrant circuits within the atrium, which lead to disorganized contraction of atrial myocardium. Non-uniformity in the spatial distribution of atrial refractory periods is essential to the initiation and maintenance of these multiple circuits (13). In experimental models, it is generally not possible to induce AF in a normal atrium without intervention to modify atrial refractoriness (13,14). Since postoperative AF frequently occurs on the second to fourth days after the operation and is transient, there must be some temporary alterations in the atrial myocardium inherent to cardiac surgery that modify the distribution of refractory periods and make it susceptible to the development of fibrillation. Such intraoperative events may be related to inadequate myocardial protection (15) and pericarditis (14). Mullen et al (16) showed that postoperative supraventricular arrhythmias were more frequent after crystalloid than blood cardioplegia (63% versus 40%, P<0.05), but other studies yielded conflicting results (17-19). Despite the best available myocardial protection techniques, AF still occurs in a significant number of patients, suggesting that certain patients have an intrinsic pattern of abnormal distribution of refractoriness preoperatively (20) or increased intra-atrial conduction time (21). An increased postoperative sympathetic tone may also contribute to the occurrence of AF (22,23). Kempf et al (24) showed that patients who were on beta-blockers preoperatively had a higher level of beta-adrenoreceptors in atrial tissue excised during cardiac surgery. Moreover, patients with AF postoperatively had the highest level of beta-adrenoreceptors. However, the peak sympathetic activity occurs during the first 24 h after operation (22,23) whereas AF typically occurs a mean of 2.5 days postoperatively (2,3), suggesting that the parasympathetic system might also play a role.


PREVENTION

Several prophylactic pharmacological protocols have been demonstrated to be effective in decreasing the incidence of postoperative AF. In the aforementioned meta-analysis by Andrews et al (4), 31 randomized trials were reviewed to determine the efficacy of digoxin, verapamil and beta-blockers in preventing the occurrence of postoperative supraventricular arrhythmias in patients undergoing coronary bypass grafting. They concluded that beta-blockers had a significant protective effect (8.7% versus 34%), but that neither digoxin nor verapamil were effective. It is interesting to note that several of the studies have shown either no effect or an increase in the incidence of postoperative AF with the use of digoxin (3,25). More recent studies have shown the effectiveness of intravenous procainamide (26), nifedipine and metoprolol (27), sotalol (28), diltiazem (29), acebutolol (30), atenolol (31) and magnesium (32) as prophylactic agents. These protocols appear to be more effective in patients undergoing coronary artery bypass grafting than valvular operations. The most spectacular results were reported with intravenous acebutolol (30). As approximately 75% of coronary bypass patients will not present with postoperative AF, then is it advisable to submit all patients to drugs which would not be otherwise required? Furthermore, patients with bronchial disease, bradycardia, AV block or poor left ventricular function may not tolerate beta-blockers.


MANAGEMENT

Atrial and ventricular epicardial pacing wires, when routinely placed at operation and left for five to 10 days postoperatively, are mostly useful in the diagnosis and treatment of postoperative arrhythmias (33). Although rapid atrial pacing is usually effective for conversion of reentrant arrhythmias such as atrial flutter or atrioventricular junctional tachycardia (AV nodal reentry), it is not applicable to AF. Therefore, it is mandatory to establish a precise diagnosis of any narrow QRS tachycardia occurring in the postoperative periods. Definition of AF was provided in another chapter of this review and the recording of an atrial electrogram may be extremely useful if the diagnosis is unclear on the surface ECG tracing obtained from the bedside monitor.

The main objective of treatment for postoperative AF is to achieve a decrease in heart rate (110 beats/min or less), ideally by conversion to normal sinus rhythm. Since AF is often recurrent and usually terminates spontaneously, it is advisable, under most circumstances to begin with pharmacological control of ventricular rate rather than attempting immediate conversion to sinus rhythm. There is general agreement that a more rapid control of ventricular rate is desirable in postoperative patients compared with nonsurgical patients.

Rate control: Strategies aimed at reduction of ventricular rate during AF are controversial. Digoxin is still the most commonly used agent for rate reduction, but its effectiveness in lowering ventricular response is hampered by the high postoperative sympathetic tone (2,22,23,33). Some surgical teams, including those of the authors, advocate rapid complete intravenous digitalization. Intravenous beta-blockers or verapamil are also used in several centres with precautions regarding possible hypotension owing to decreased peripheral resistance or depressed myocardial contractility associated with these agents (2,33). Digoxin appears particularly useful in the following two situations:

• When the use of antiarrhythmic drug aimed at conversion to sinus rhythm, especially procainamide which is devoid of AV nodal depressing effects, results in a faster ventricular rate due to a decrease in the atrial rate during AF.
• In patients with ejection fraction less than 30% and/or at risk of congestive heart failure.

Restoration of sinus rhythm: The efficacy of digoxin alone to convert from AF to sinus rhythm is controversial (2). Most groups believe that DC shock cardioversion is not indicated except if hemodynamic deterioration occurred, which is rather unusual (33). Review of recently published reports indicates that procainamide and propafenone (34,35,38,39) are probably the most effective drugs to achieve pharmacological cardioversion to sinus rhythm (Table 1). Amiodarone and sotalol have also been considered, but with as yet unproven efficacy (38,40). It should be emphasized, however, that postoperative AF usually converts spontaneously, sparing patients from unnecessary antiarrhythmic therapy. Finally, novel approaches are currently being developed. These include direct electrical stimulation of parasympathetic neural elements (36,37) for rate control, and low energy shocks (less than 5 J) with temporary epicardial electrodes (42) for restoration of sinus rhythm.


TABLE 1
Pharmacological restoration of sinus rhythm

^*When two agents are shown, these represent randomized groups.  ^**Percentage conversion to sinus rhythm. CABG Coronary artery bypass grafting; HRÊHeart rate; iv Intravenous; prn As needed; Valves Valvular replacement or repair


ATRIAL FLUTTER

In the postoperative period, atrial flutter should be differentiated from AF owing to the fact that it is usually amenable to pacing therapy with epicardial wires. Two forms of atrial flutter may occur after open heart surgery (43). Both types have a constant atrial rate and uniform morphology of atrial electrograms, but are distinguished by atrial rate and response to rapid atrial pacing. In type I atrial flutter, the atrial rate usually ranges between 280 and 320 beats/min and in type II atrial flutter from 320 to 360 beats/min. The essential difference between these two types is that type I atrial flutter can be interrupted by rapid atrial pacing, but type II atrial flutter cannot. In the authorsÕ experience, the latter is more frequent than type I atrial flutter and is less well tolerated clinically owing to a faster ventricular rate, occasionally up to 180 beats/min. Pacing therapy is not indicated in type II atrial flutter, but if it is sustained and poorly tolerated, rapid atrial pacing at 400 to 600 beats/min can be used to precipitate AF, allowing easier control of ventricular response with pharmacologic agents (33). Rarely, continuous atrial pacing may be required.


ATRIAL FIBRILLATION AFTER NONCARDIAC SURGERY

In contrast to cardiac surgery, the incidence of AF is much lower after major general surgical procedures. In the study by Goldman (1), only 4% of 916 patients developed new postoperative supraventricular arrhythmias, and treatment was not required in 40% of these cases. AF appears to be more frequent after thoracic operations, as Mowry and Reynolds (44) have reported an incidence of 19% after pneumonectomy. Of particular importance in the management of patients with AF after noncardiac surgery is to rule out precipitating factors such as underlying cardiac disorder, respiratory instability and metabolic disturbances. One should also realize that these patients do not have epicardial wires which are useful not only for diagnostic and therapeutic purposes as stated above, but also to provide back-up pacing if bradycardia or asystole should occur as a complication of antiarrhythmic therapy. Despite this precaution, the use of digitalis and/or verapamil appear useful in this setting (33). External cardioversion may be needed for atrial flutter, since this arrhythmia leads to poorly tolerated rapid ventricular rate. If time permits, percutaneous insertion of an endocardial catheter or a transesophageal electrode (45) may be used to perform overdrive pacing with the same protocol as used with epicardial wires.


CONCLUSIONS

• Depending on the type of cardiac surgical procedure, the incidence of postoperative AF varies in the vicinity of 5% for nonthoracic major surgery, 20% for noncardiac major thoracic surgery, 25% for coronary surgery and 80% for valvular operations.
• The strongest independent predictor of postoperative AF is the patientÕs age, although several factors related to the degree of myocardial damage existing preoperatively or occurring intraoperatively may also be predictive.
• The etiology of transient dispersion of atrial refractoriness is unclear as well as the strategies for its prevention.
• No prophylactic or therapeutic interventions universally applicable and optimally effective have yet been established. Although continuous intravenous infusion for 24 to 48 h of beta-adrenoreceptor blockers has been suggested as an effective strategy to prevent the occurrence of postoperative AF, its safety and effectiveness remain to be established. Actually, the patients in whom postoperative AF is likely to be detrimental are not only those who are at the highest risk of having this complication, but are also those who will poorly tolerate antiarrhythmic agents, including beta-adrenoreceptor blockers and calcium channel blockers.
• Novel strategies which will overcome these difficulties are clearly needed.

RECOMMENDATIONS

Detection and diagnosis: Continuous monitoring is useful for rapid diagnosis of postoperative tachyarrhythmias. In cardiac surgical patients, routine placement of epicardial electrode wires at the time of surgery is recommended. They are useful for diagnosis of complex supraventricular rhythms, but, most importantly, provide backup pacing leading to safer use of rate control agents. (Level III, Grade B.)

Prophylaxis after cardiac surgery

• Although postoperative AF is associated with increased morbidity and longer hospital stay, there is no proof that prophylactic measures decrease associated clinical events. (Prophylaxis to decrease associated clinical events: Level V, Grade C.)
• The only drugs with evidence of efficacy in decreasing the incidence of postoperative AF are beta-blockers. They may be used postoperatively if prophylaxis is considered beneficial and indicated, based on the number of recognized clinical predictors found in a given patient. (Prophylaxis with beta-blockers: Level I, Grade A.)
• Methods for absolute prevention have not yet been established. Continuous intravenous administration of beta-blockers during the period at risk appears beneficial, but cost effectiveness of newer short acting beta-blockers has not yet been established. (Prophylactic intravenous beta-blockers: Level IV, Grade C.)

Rate control

Rapid rate control is feasible and safe with aggressive use of digitalis and beta-blockers or verapamil, owing to back-up pacing with temporary epicardial wires. (Rate control with digitalis or verapamil or beta-blockers: Level IV, Grade C.)

Restoration of sinus rhythm

• The decision to convert AF to sinus rhythm is controversial. If conversion is deemed necessary, procainamide and propafenone have been successful with relative safety. External DC shock cardioversion is seldom necessary. (Pharmacologic conversion with procainamide or propafenone: Level II, Grade B.)
• No conclusive evidence allows a consensus on the use of class III antiarrhythmic agents. These appear as suitable second choice agents when others are ineffective and may also be used for rate control. (Class III agents: Level V, Grade C.)

Long term management

• Since postoperative AF is a time limited event in the vast majority of patients, antiarrhythmic therapy is not recommended for more than three weeks after operation. (No long term therapy: Level II, Grade B.)
• Anticoagulation is not indicated for self-limited, short duration AF in patients without preoperative chronic AF. However, it has not been evaluated in clinical trials. Recommendations listed elsewhere apply if AF persists at the time the patient is discharged from the hospital. Some authors recommend cardioversion before discharge. This practice is controversial, since chronic AF developing de novo appears unusual. Anticoagulation therapy should be based on its relative risk of thromboembolic events versus the risk of intrapericardial bleeding (with resultant delayed tamponade). (Anticoagulation: Level V, Grade C.)

REFERENCES