Atrial Fibrillation and Thromboembolism

Abstract

Atrial fibrillation (AF) is the most common sustained arrhythmia encountered in clinical practice and is present in approximately 2 million people in the United States. Its prevalence increases with age, occurring in 0.5% of the population aged 50-59 yr and in 8.8% of those more than 80 yr of age [1]. AF is associated with a reduced life expectancy and with an approximate doubling of all-cause mortality [2]. Most of the excess mortality associated with AF is secondary to thromboembolic events [2]. Atrial stasis is believed to provide the thrombogenic milieu in which thrombi form and from which emboli originate, although a hypercoaguable state also exists in patients with AF [3]. Recent studies have provided convincing evidence that anticoagulation reduces morbidity in subsets of patients with AF and identifiable risk factors, and current recommendations are that all such individuals receive anticoagulant prophylaxis. Anesthesiologists are now presented with the problem of anticoagulated patients with AF scheduled for surgery. The risk of stopping anticoagulants and exposing the patient to an increased risk of thromboembolism must be weighed against the risk of continuing therapy and surgical bleeding. Thus, it is important for the anesthesiologist to appreciate the risk factors associated with thromboembolism in this population so that prudent recommendations for perioperative management can be made. Thromboembolic risk varies among patients with AF and can be estimated by the presence or absence of a number of recognized risk factors. We present the argument that the perioperative management of anticoagulation in patients with AF should be risk-stratified and cite the literature on patients with prosthetic heart valves receiving anticoagulants to support this hypothesis. We also address the thromboembolic risk of cardioversion of AF to sinus rhythm. Recommendations to reduce this risk and ongoing studies, particularly the use of transesophageal echocardiography (TEE) to screen patients with AF, may further guide patient management. The following discussion concentrates on the issue of cerebral thromboembolism, which is the most common and frequently the most devastating embolic event. However, it should be stated that approximately one third of recognizable systemic thromboemboli in patients with AF are extracerebral [4]. Risk of Stroke Much of our present information concerning the incidence of AF and the associated thromboembolic risk is derived from the Framingham cohort study of more than 5000 individuals followed for more than 30 yr [2]. Compared with age-matched controls without AF, the risk of stroke increased 17-fold in patients with AF associated with rheumatic valvular disease and 5-fold in those with nonrheumatic AF [5]. The risk of stroke varies greatly depending on age and coexisting cardiovascular disease. The annual stroke rate is approximately 7% in patients 80-89 yr old with nonrheumatic AF [6] and, in this age group, AF is the only cardiovascular condition associated with an increased risk of stroke. At least half of the strokes secondary to AF result in severe neurologic deficits or death [6,7]; therefore, these events are far from trivial in terms of their cost to both the individual and society. Although the above information was derived mainly from patients with chronic AF, recent data suggest that the risk of thromboembolism is similar in patients with paroxysmal AF [8,9]. Prevention of and Risk Factors for Stroke In 1989, the American College of Chest Physicians (ACCP) Consensus Conference on Antithrombotic Therapy recommended that patients with AF and a history of embolic events, cardiomyopathy, valvular heart disease, or thyrotoxicosis receive anticoagulant therapy [10]. These recommendations were based on limited data. Since then, six major randomized primary prevention trials have enrolled patients with nonrheumatic AF to evaluate the efficacy of warfarin and aspirin therapy in reducing cerebral thromboembolic complications [8,9,11-14]. Pooled results from these studies demonstrate that, with appropriate anti-coagulant therapy, the risk of cerebral stroke in AF patients could be reduced by as much as 70% (annual stroke rate of 4.5% for control patients versus 1.4% for warfarin-treated patients; P < 0.001) with a minimal increase in the risk of major cerebral hemorrhage (1.0% in the control group versus 1.3% in the warfarin-treated group) [15]. Furthermore, approximately one half of thromboembolic events in the warfarin-treated group occurred in patients who were undertreated or had stopped therapy; therefore, the efficacy of warfarin was probably underestimated. These studies identified several independent risk factors for stroke in patients with nonrheumatic AF: hypertensive heart disease (HTN) (relative risk [RR] 1.6), diabetes mellitus (DM) (RR 1.7), a history of transient ischemic attack (TIA) or stroke (RR 2.5), and increasing age (RR 1.4 per decade) [17]. Although not significant in the multivariate analysis, the risk of stroke was also increased by the presence of congestive heart failure (CHF) and clinical coronary artery disease (angina or myocardial infarction). The stroke rate in patients with either of these disorders was approximately threefold higher than in patients with no risk factors. Echocardiographic predictors of thromboembolism were left atrial enlargement, left ventricular dysfunction [18], and mitral annular calcification [9]. Based on these studies, we know that patients with AF who are <65 yr old without HTN, CHF, DM, or a history of TIA/stroke have a low risk of thromboembolism (approximately 0.5%-1% per year) and derive no significant benefit from anticoagulant therapy. In patients with AF, warfarin therapy is beneficial in patients >65 yr old and patients <65 yr old with one or more of the aforementioned risk factors. The efficacy of aspirin in preventing stroke is less than that of warfarin in patients >65 yr old or those with clinical risk factors. Table 1 summarizes the recommendations of experts based on pooled data from these six clinical trials; echocardiographic data were not considered. Unfortunately, a recent survey of health-care practices demonstrates significant deviations from those recommendations, with a particular bias on the part of care-givers to withhold warfarin treatment in the elderly [19].Table 1: Recommendations for Anticoagulant Therapy in Patients with Atrial FibrillationPatients with nonrheumatic AF are therefore not homogenous in terms of their risk for thromboembolism. Based on the risk factors mentioned above and the recommendation of others [20], patients with AF can be categorized as being at high, intermediate, or low risk of cerebral thromboembolism (Table 2). For example, in the absence of anticoagulation, patients with AF and two or more clinical risk factors or a history of TIA stroke within the last year have an annual risk of stroke of 18% or 12% per patient-year, respectively, and are considered at high risk of thromboembolic complications [21,22].Table 2: The Annual Rate of Cerebral Thromboembolism in Patients with Atrial Fibrillation Categorized by Overall RiskPerioperative Management of Anticoagulation The above-mentioned studies examined the annualized stroke rate from which an overall estimate of thromboembolic risk may be derived. No study, however, has specifically examined the perioperative management of anticoagulation in patients with AF and the risk of stopping anticoagulation for this period. In a review, Kearon and Hirsh [23] stated that, in patients with nonrheumatic AF, oral anticoagulants may be stopped before surgery and that the risk of thromboembolism was not sufficiently high to justify either pre- or postoperative prophylaxis with IV heparin. They did not address the perioperative management of patients with AF associated with valvular disease and, although recognizing the marked variability in the risk of thromboembolism among patients with nonrheumatic AF, they did not recommend stratifying perioperative management based on risk. From the previous discussion and the data presented in Table 2, a "one rule fits all" approach may not be appropriate. In support of this argument, we cite the published experience of the perioperative management of anticoagulation in patients with mechanical heart valves for noncardiac surgery. Overall, patients with a mechanical heart valve have an average rate of thromboembolism, without anticoagulation, of approximately 8% per annum [24]; however, risk depends on the type of prosthetic valve and its position. For example, in patients with a St. Jude mechanical valve, the thromboembolic rate varies from 6% with the prosthetic valve in the aortic position to 17% with mitral valves [25]. It should be noted that the risk of thromboembolism in patients with moderate to high-risk AF is comparable to that of patients with a mechanical heart valve in the aortic and mitral position, respectively. The variable risk for thromboembolism in patients with mechanical heart valves is reflected in the results of studies that examined the perioperative management of anticoagulation in such patients for noncardiac surgery. Tinker and Tarhan [26] performed a retrospective study of 159 patients with mechanical heart valves, mostly aortic, undergoing noncardiac surgery. They concluded that stopping warfarin 1-3 days before surgery and allowing the prothrombin time (PT) to recover to within 20% of normal, then restarting warfarin 1-7 days postoperatively was safe and effective in preventing thromboembolic complications in patients with mechanical heart valves. Katholi et al. [27] also noted that it was safe to stop anticoagulation in the perioperative period in patients with an aortic valve prosthesis undergoing noncardiac surgery, but they reported a high perioperative thromboembolic rate in patients with a mitral valve prosthesis using this approach. In this latter group of patients, the authors recommended a more aggressive approach to thromboembolic prophylaxis with rapid reversal of warfarin, using vitamin K, 24 h before surgery and starting IV heparin 12 h after surgery [28]. These data, albeit limited, form the basis for much of current-day practice and demonstrates that a single management approach does not benefit all patients at risk of thromboembolism. The concept of risk-adjusted intensity of anticoagulation for patients with prosthetic heart valves has been proposed by authors in both the United States [29] and Europe [30]. Although some physicians may consider that the safest approach for all patients at risk of thromboembolism is to hospitalize them before surgery and to continue anticoagulation with heparin until the time of surgery, economic issues should also be considered. Eckman and colleagues [31] demonstrated that the cost of such an approach to all patients with mechanical heart valves undergoing noncardiac surgery in the United States in a given year would be prohibitive. They concluded that the cost-effectiveness of aggressive (defined later in text) anticoagulant therapy is justified only in cases of the most thrombogenic valves or when this therapy can be instituted when the patient is already hospitalized for other reasons. Stopping warfarin before surgery and restarting it postoperatively may represent the most cost-effective approach for the population of individuals with AF as a whole. However, as with patients with prosthetic heart valves, there are high- and low-risk subgroups among patients with AF, and this must be considered when selecting the optimal perioperative management. Given the devastating impact of a stroke in this population [6,7], a more aggressive approach to perioperative anticoagulation may be justified in the high-risk patient with AF. The Patient with AF and a High Risk of Thromboembolism When major surgery (defined as invasion of a body cavity or major orthopedic procedure) is planned in patients with AF and multiple clinical risk factors (HTN, CHF, DM, previous TIA/stroke, age >75 yr, prosthetic heart valve, or mitral stenosis), aggressive anticoagulation throughout the perioperative period should be considered, especially when the patient is already hospitalized. This recommendation is based on data showing an annual risk of stroke of up to 18% in the absence of anticoagulation in this population [21]. Preoperative anticoagulation can be achieved by several means. First, warfarin can be stopped approximately 3 days before surgery and full anticoagulation with heparin therapy substituted until 4-6 h preoperatively. The activated partial thromboplastin time should approach approximately twice normal, correlating with a heparin blood level of 0.2-0.4 U/mL by protamine titration assay [30]. A PT within 20% of normal should be documented before major surgery [26]. A second strategy is to continue warfarin until 24-48 h preoperatively, followed by a small dose of vitamin K (1 mg IV or subcutaneously) [28]. Although vitamin K has the advantage of simplicity and low cost, its dose-response is unpredictable. Overtreatment with vitamin K increases not only the risk of thromboembolism, but also the difficulty of postoperative re-anticoagulation. Patients receiving long-term warfarin therapy may require increased doses of vitamin K or longer warfarin-free periods before surgery [32]. Fresh-frozen plasma (FFP) is an alternative to vitamin K, but its cost ($45 per unit of FFP at the University of Virginia) and transfusion risks must be considered. FFP is indicated for the anticoagulated patient requiring emergency surgery. The correct approach to postoperative anticoagulation management in patients with AF is also contentious. Kearon and Hirsh [23] state that the risk of significant bleeding as a result of postoperative IV heparin outweighs the possible benefit in patients with AF. Although no data dispute this recommendation, a common approach to all patients with AF may not be appropriate. In support of this hypothesis, we cite the previously mentioned experience of the perioperative management of anticoagulation in patients with mitral valve prosthesis for noncardiac surgery [28]. These patients have a risk of thromboembolism similar to that of the high-risk patient with AF, and they seem to benefit from IV heparin started postoperatively once hemostasis is assured [28]. When minor surgery is planned in patients at the highest risk of thromboembolism, it is reasonable to consider maintaining either fully or moderately decreased anticoagulation throughout the perioperative period. Minor surgery includes superficial surgery such as biopsies, cataract surgery, and dental extractions. Studies containing small numbers of patients support the safety of performing both dental extractions and cataract surgery in the presence of a PT of 1.5-2.5 times the control value [32-34]. The Patient with AF and a Moderate to Low Risk of Thromboembolism Patients at moderate to low risk of thromboembolism include those with AF at any age with one or no clinical risk factors, including a patient with a remote history of stroke or those with echocardiographic risk factors in the absence of clinical risk factors. Anticoagulation likely can be safely stopped up to 5 days before major surgery in this population and restarted postoperatively. For minor surgery, anticoagulation may be continued, either at a fully or moderately reduced dose, throughout the perioperative period [32-34]. In summary, it seems to be safe to stop perioperative anticoagulation in patients with an annual risk of stroke as high as 6% (e.g., patients with an aortic mechanical prosthetic valve or low- to intermediate-risk AF patients). There is a significant risk of major thromboembolic complications when anticoagulation is stopped in patients who are at high risk of thromboembolism (e.g., patients with mitral prosthetic valves and potentially high-risk AF patients), which may supercede the risk of postoperative hemorrhage. Surgery may be safely performed once the international normalized ratio (INR) reaches 1.5 [23]. Cardioversion of Atrial Fibrillation and Thromboembolic Risk Cardioversion may be performed to optimize cardiovascular function and avoid anticoagulation. Whether the conversion of AF to sinus rhythm is spontaneous or induced by electrical or chemical means, it is associated with a significant risk of stroke (range 0%-7%; average incidence 1.5% per cardioversion) [35]. The risk of thromboembolism depends on the chronicity of AF and coexisting risk factors [35]. The presumed mechanism of cardioversion-related thromboembolism is dislodgement of preexisting thrombus after the resumption of effective atrial contractile activity. It is important that the anesthesiologist, as a perioperative physician caring for patients with AF, appreciates the risks of cardioversion and current techniques to reduce the chance of a thromboembolic event. Recent studies have elucidated several interesting facts regarding cardioversion. Atrial mechanical activity often does not occur when P waves reappear on the electrocardiogram because there is atrial stunning [36-39]. Generally, the longer a patient has been in AF before cardioversion, the longer the period of atrial stunning. It may take more than 3 wk for atrial function to fully recover. Cardioversion is also frequently associated with worsening atrial function, as evidenced by the new formation of or increasing spontaneous echo contrast (SEC) [36,37]. SEC represents areas of blood stasis and rouleaux formation and is an independent predictor of left atrial (LA) thrombus [40]. This may explain why thromboembolic events most often occur on Day 3 postcardioversion [35]. Based on the above data, as well as numerous uncontrolled case series and retrospective trials, the ACCP recommends 3 wk of anticoagulation (INR 2.0-3.0) before elective cardioversion in patients who have had AF >2 days. All patients should receive 4 wk of anticoagulation after cardioversion [17]. Postcardioversion anticoagulation is recommended because of cardioversion-related thrombogenesis, the high rate of relapsing into AF within the first month after cardioversion, and postcardioversion atrial stunning. These recommendations are based on the assumption that thrombus is not present in the atria of patients who have been in AF for less than several days. This assumption has been challenged by Stoddard and colleagues [41], among others, who found that 14% of 143 patients with AF of <3 days' duration had LA thrombi detected by TEE. Predictors of LA thrombus in patients with acute AF are mitral stenosis, left ventricular ejection fraction (LVEF) of 40% or less, and SEC in the LA, whereas independent predictors of an embolic event were mobile LA thrombus, CHF, coronary artery disease, and a decreased LVEF [41]. The findings of early thrombus formation in the atria of patients with AF of short duration by Stoddard et al. [41] should significantly influence the future management of patients who develop postoperative AF. Although anticoagulation should be considered in post-operative patients who develop and remain in AF for more than 48 h [42], there is now an argument for heparinization soon after the diagnosis of AF [43]. When anticoagulation is contraindicated, such as in the early postoperative period, expedient restoration of sinus rhythm by either electrical or pharmacological means may be the optimal approach. Evolving Strategies for TEE in the Management of AF and Cardioversion Many anesthesiologists find TEE to be a useful tool for patient management. It has recently been proposed that TEE can be used to diagnose LA and left atrial appendage (LAA) thrombus and to identify low-risk patients who may safely undergo cardioversion without the need for prior anticoagulation [44]. This would reduce hospital stays by obviating the need for anti-coagulation before cardioversion, decrease patient cost and inconvenience, decrease the total period of anticoagulation with its attendant hemorrhagic risks, and more promptly return patients to the physiologic advantages of sinus rhythm. Although TEE requires sedation and is associated with its own risks, the sensitivity of TEE for identifying LA and LAA thrombus approaches 100% in the hands of a skilled echocardiographer versus a sensitivity of 60% with transthoracic echocardiography (TTE) [45]. The LAA is the most frequent site of thrombus formation in patients with nonvalvular AF but is a site not well visualized by TTE.1 The use of TEE to screen patients before elective cardioversion holds promise, but the existing studies using this strategy have insufficient statistical power to provide definitive conclusions. Furthermore, others have noted a significant false-positive rate with TEE performed immediately before surgery in comparison to the surgical findings [47]. There are also reports of thromboembolic events despite negative TEE examinations before cardioversion [48,49]. (1) Matsumura M, Shah P, Kyo S, Omoto R. Advantages of transesophageal echocardiography for correct diagnosis of small left atrial thrombi in mitral stenosis [abstract]. Circulation 1989; 80(Suppl 2):II678. Although LA SEC is a predictor of LA thrombus, some [50] but not all studies [44] have found this to be a risk factor for thromboembolism. Black et al. [48] reviewed the precardioversion TEE in patients with AF who later suffered a thromboembolic event. They concluded that SEC "has a limited predictive value for embolism in patients undergoing cardioversion." Many of these issues will likely be resolved in the continuing trial Assessment of Cardioversion Utilizing Transesophageal Echocardiography (ACUTE). This multicenter, prospective study is randomizing patients with AF of more than 2 days' duration who were not previously anticoagulated to a conventional arm (treatment according to the recommendations noted in this article) or TEE screening arm. The results of the ACUTE pilot study suggest that the TEE arm may be safer, with fewer bleeding complications and less need for emergent cardioversion [51]. Additionally, recent data support the cost-effectiveness of TEE-guided early cardioversion of AF [52]. This strategy is primarily limited to patients at extreme risk of bleeding complications. Conclusion There have been important changes in the management of patients with AF in the last decade. There has been a greater appreciation of the risks associated with this dysrhythmia and the formulation of a number of recommendations from national societies to reduce the risk of thromboembolism, which is responsible for most of the associated morbidity and mortality. A single, conservative approach to the prevention of perioperative thromboembolism in all patients with AF has been recommended [23]. There have been no clinical investigations to determine the optimal method to prevent perioperative thromboembolism in patients with AF; therefore, the validity of this recommendation is unproven. The evidence to support the opinion presented in this article is, by necessity, indirect. It is supported by the highly variable risk of thromboembolism in patients with AF, depending on the presence or absence of a number of identified risk factors. Furthermore, a precedent for a risk-stratified approach to the perioperative management of anticoagulation has been set by the published recommendations for patients with prosthetic heart valves scheduled for noncardiac surgery. Although patients with AF and patients with prosthetic heart valves are not perfectly analogous populations, they share a similar risk of major thromboembolism and therefore provide a useful comparison. In conclusion, the opinions presented in this article are intended to stimulate further examination of this issue, not to dictate practice.