Syncope

Author: Christopher T. Bowe, MD, FACEP, Associate Residency Director, Department of Emergency Medicine, Maine Medical Center, Portland, ME.

Peer Reviewer: Mary Jo Bowman, MD, FAAP, FCP, Associate Professor of Clinical Pediatrics, Pediatric Emergency Medicine Fellowship Director, Nationwide Children's Hospital, Columbus, OH.

Introduction

Syncope is the transient loss of consciousness with loss of postural tone, followed by a spontaneous recovery. It is a symptom, rather than a specific disease process. The final shared pathway of syncope is inadequate delivery of vital nutrients to the reticular activating system in the brainstem. The return of perfusion, achieved by supine posture and autonomic stimulation, results in the return of consciousness.

Syncope is a common chief complaint in the emergency department (ED), accounting for 1%-3% of ED visits and up to 6% of admissions. Up to 50% of adults will have a syncopal event in their lifetime.1 Patients rarely require immediate stabilization, as the event is typically transient. Syncope is a challenging complaint for the clinician, as there are multiple causes of syncope. Many common etiologies are benign and do not require an extensive workup, but there remains a subset of patients who may have a potentially life-threatening condition that must be identified. Clinical decision rules have been investigated in an effort to identify patients at higher risk for adverse outcomes.2,3

The initial evaluation of a patient presenting with the complaint of syncope begins with a thorough history and physical examination. An ECG, combined with the history and physical examination, can lead to the diagnosis in up to 69% of cases.4 Historical features can help to differentiate syncope from other similar complaints, including seizure or a fall with associated loss of consciousness. A thorough history can also direct the clinician to other less common but serious causes of syncope, such as pleuritic chest pain and shortness of breath in a patient with a pulmonary embolism. Initial vital signs and orthostatic changes are important parts of the physical examination, as are the neurologic and cardiovascular examination.

An electrocardiogram (ECG) is a noninvasive, inexpensive tool that can help identify many serious causes of syncope. The American College of Cardiology states that, in addition to the history and physical examination, the 12-lead ECG is the procedure of first choice in the evaluation of syncope.5 The ECG can be useful to help predict prognosis and risk-stratify patients, even when it does not result in the diagnosis.

In an effort to improve the sensitivity and specificity of the evaluation of patients with syncope, a number of studies have been offered to assist in the identification of patients at increased risk of adverse outcomes. The American College of Emergency Physicians also has released a Clinical Policy on the critical issues in the evaluation and management of adult patients presenting to an emergency department with syncope.

What Initial Evaluation Findings Can Be Used to Predict Adverse Outcomes in Patients Evaluated for Syncope?

Source: Martin TP, et al. Risk stratification of patients with syncope. Ann Emerg Med. 1997;29:459-466.

Martin et al were one of the first groups to attempt to develop a risk classification for patients presenting to an ED with a complaint of syncope. They undertook two prospective, observational cohort studies of patients presenting to the University of Pittsburgh Medical Center ED, Pittsburgh, PA. The first cohort was designed to identify data from the patients' history, physical examination, and electrocardiogram that correlated with increased risk of arrhythmia or mortality in one year. The second cohort was designed to validate the risk classification system.

The first cohort of patients was enrolled from March 1981 through February 1984, while the second cohort was enrolled from August 1987 through February 1991. Syncope was defined as a sudden transient loss of consciousness associated with an inability to maintain postural tone, followed by a spontaneous recovery. Patients were excluded if they required pharmacologic or electrical treatment before their presentation.

Multiple variables were reviewed, including demographic information, comorbid conditions, prior syncope, and ECG findings. Comorbid conditions evaluated for predictive use included coronary artery disease, hypertension, congestive heart failure, diabetes mellitus, and a history of ventricular arrhythmias. For study purposes, ECGs were reviewed and classified as either normal, nonspecific ST- and T- wave abnormalities (NST) or abnormal. A logistic regression model was developed by using the multivariate predictors of death or arrhythmia in the year after initial evaluation.

All patients were followed at three-month intervals for up to three years. Patients were evaluated for recurrence of syncope, new cardiovascular events, subsequent diagnoses of significant arrhythmias, or mortality. Death was classified for the study outcome as sudden if specific criteria were met, including if death occurred within 24 hours of a patient functioning in his/her normal health, an unwitnessed death, or cardiac arrest that eventually led to death even if the initial resuscitation was successful.

There were a total of 252 patients included in the derivation cohort and 374 patients in the validation cohort. No patients were lost to follow-up in the derivation cohort, and only three patients (.8%) were lost to follow-up in the validation cohort. The predictors of sudden death or arrhythmia were assessed at one year's time.

The combined multivariate model for arrhythmia or death showed four significant factors. They included an abnormal ECG, history of ventricular arrhythmia, history of congestive heart failure, and age greater than 45 years. Patients were then placed into four groups defined by the number of risk factors identified. These groups were then assessed for one-year risk of arrhythmia or death.

In the derivation cohort, patients with no risk factors had a 5.5% risk of arrhythmia, while patients with three or four risk factors had a 63% risk of arrhythmia at one year. In the validation cohort, patients with no risk factors had a 3.3% risk of arrhythmia, while patients with three or four risk factors had a 45.5% chance of arrhythmia at one year.

With regard to mortality, patients in the derivation cohort with no risk factors had a 1.8% incidence of sudden death, while patients with three or four risk factors had a 37% incidence of sudden death. In the validation cohort, patients with no risk factors had a 1.1% incidence of sudden death, while those with three or four risk factors had a 27.3% incidence of sudden death.

Commentary

This study by Martin et al demonstrates that certain risk factors, identifiable at the time of initial evaluation in the ED, can be used to risk-stratify patients for one-year risk of arrhythmia or sudden death. They collected data for both a derivation and then, subsequently, for validation of their initial findings. The study did an excellent job at following patients, with a total of only three patients (0.4%) lost to follow-up. The results were similar in both groups, as both incidence of sudden death and arrhythmia increase as the number of risk factors increase.

This study is a beneficial starting point for an evidence-based evaluation for the risk of serious outcomes of patients presenting to an ED for the evaluation of syncope. It specifically directs providers' attention to the importance of the initial ECG, as well as important historical factors including advancing age, history of congestive heart failure, and history of ventricular arrhythmias. All of these risk factors can be determined at the time of the initial evaluation in the ED.

The limitations of this study are also important to review. The four risk factors were assigned equal value in the risk assessment. Each risk factor had a different odds ratio for serious outcome at one year, but the patients were grouped, for simplicity, based on the number of risk factors. The history of ventricular arrhythmia was associated with an odds ratio of 4.8 for the outcome measures, while age greater than 45 years was associated with an odds ratio of 3.2. One may postulate that a 44-year-old patient with a normal initial ECG in the ED, but a history of both prior ventricular arrhythmia and severe congestive heart failure (two risk factors), is at a higher risk of sudden death than a 44-year-old who presents in atrial fibrillation after a weekend of heavy alcohol consumption (two risk factors). Martin et al did not try to place a value on each risk factor, but rather emphasized the total number of factors present.

An important factor that was not discussed involved whether the underlying etiology of the syncope was determined by the initial evaluation. The ED physician does not need a risk stratification tool for patients presenting with syncope when the underlying etiology is identified. For example, a patient with syncope who has persistent hypotension and black stool that is guaiac positive does not require a decision tool to assist in his/her disposition. Risk stratification is useful in patients where the etiology of syncope is unknown after a complete evaluation. A thorough history and physical exam, combined with the ECG, can lead to diagnosis in up to 69% of patients presenting with syncope.4 This study did not explain which patients had a diagnosis upon completion of their initial evaluation. All of the patients in this study were admitted to the hospital for a minimum of 24-hour Holter or telemetry monitoring. This approach makes the disposition decision clear, but does not add to our understanding of the proper disposition of patients who have an episode of syncope with unknown etiology.

Is There a Clinical Decision Rule That Can Be Used to Identify Syncope Patients at Risk for a Short-term Adverse Outcome?

Source: Quinn J, et al. Derivation of the San Francisco Syncope Rule to predict patients with short-term serious outcomes. Ann Emerg Med. 2004;43: 224-232.

Quinn et al took another step forward in the effort to guide the disposition of patients presenting with syncope to the ED. They performed a prospective cohort study of patients presenting with syncope or near-syncope to a large university teaching hospital and attempted to form an evidence-based decision rule to help guide the decision to admit. This study derived the San Francisco syncope rule to help predict short-term serious outcomes.

Patients were enrolled from July 2000 until February 2002. All presenting patients were prospectively screened by research assistants and brought to the attention of an attending physician who made the final decision to enroll the patient. The attending physician then completed a data form with 50 variables, including historical features, physical exam findings, laboratory data, and radiographic and ECG findings. Patients were then treated without any study intervention.

Patients were followed to determine whether they suffered a serious outcome by day seven. Serious outcomes were defined as death, myocardial infarction, arrhythmia, pulmonary embolism, stroke, subarachnoid hemorrhage, significant hemorrhage, or any condition requiring a return ED visit or hospitalization. All patients received some form of follow up, either with medical record review (48%), direct telephone follow-up (37%), telephone follow-up with the patient's physician (11%), or local hospital checks and review of the death registry (< 4%).

A total of 684 patients were enrolled; a total of 79 (11.5%) patients experienced a serious outcome by day seven. The predictor variables were analyzed individually for association with serious outcomes and then analyzed with recursive partitioning techniques to develop a model to maximize the prediction of serious outcomes. This model was used to establish the San Francisco syncope rule, which identifies five variables that place a patient at increased risk for an adverse outcome. The five variables of the San Francisco syncope rule include an abnormal ECG, complaint of shortness of breath, hematocrit of less than 30%, a triage blood pressure less than 90 mm Hg, and a history of congestive heart failure.

When applying this rule to the 684 patients, 76 of the 79 patients with a serious outcome would have been identified, providing a sensitivity of 96.2%. The application of the rule identified 306 of the 684 patients to be at risk of serious outcomes, placing 45% of the patients in the high-risk category. In the study, disposition decisions were made independent of the predictor variables, and the study group had an admission rate of 55%. Therefore, Quinn et al note that the application of this decision tool could result in a potential 10% absolute reduction in the admission rate found in this cohort.

Commentary

This study by Quinn et al demonstrated the ability to formulate a risk stratification tool to help guide decisions regarding admission in patients presenting to an ED with syncope. The thorough evaluation of 50 predictor variables to formulate a decision rule with a high sensitivity for patients at risk for a serious outcome could be an important advance for the disposition decision.

They replicated the earlier findings of Martin et al that patients with a history of congestive heart failure or abnormal ECG findings are at higher risk for adverse outcomes. Careful review of this paper reveals that there were many factors that were associated with increased risk of serious outcomes, including the prior use of anti-arrhythmic medications, a history of coronary artery disease, and a systolic or diastolic murmur. The decision rule was formulated, however, by taking particular risk factors, in combination, to make the best possible decision tool.

There are important limitations that require discussion. First, as Quinn et al note, this is the derivation study, and validation studies are required before this rule should be implemented into clinical practice. Also, they point out that the rule did not have a sensitivity of 100%, so it is recommended as a risk stratification tool rather than a replacement for clinical judgement.

One of the apparent goals of Quinn et al was to demonstrate the usefulness of the decision tool to decrease admission rates compared to the baseline admission rate. They note that the use of their decision tool could have decreased admission rates from 55% to 45%. This was achieved, however, by decreasing the sensitivity of the decision rule. Quinn et al point out that if they added the factor of age > 75 years, they could have achieved 100% sensitivity but the specificity would have decreased and the admission rate "would be the same, if not slightly higher than, baseline." This appears to be counter to their goal of formulating a decision tool that can be used to predict serious outcomes while still decreasing admission rates. It is not clear that the admission rate that was observed of 55% is "correct," thus formulating a decision tool to decrease this rate may be comparing to a faulty standard.

Can the San Francisco Syncope Rule Findings Be Validated at an External Site?

Source: Sun BC, et al. External validation of the San Francisco Syncope Rule. Ann Emerg Med. 2007;49:420-427.

Sun et al were the first group to publish an effort to validate the findings of the San Francisco Syncope Rule (SFSR) study group. They performed a prospective, convenience study on a sample of patients presenting to a single academic ED with a chief complaint of syncope or near-syncope. Their identified primary outcome was to assess the ability of the SFSR to predict any seven-day serious outcome.

Patients were enrolled from April 2005 until April 2006. All patients presenting between 8 a.m. and 10 p.m. were screened for enrollment in the study. The five predictor variables of the SFSR were prospectively obtained by the treating physician. Physicians then treated and admitted patients in their usual manner, without any study intervention. Direct patient follow-up was completed to identify admissions, or the occurrence of a serious outcome.

A total of 477 patients consented to participate in the study. A total of 56 (11.7%) patients experienced a serious outcome by day seven. Of those 56 patients, only 16 (3.4%) had the event diagnosed after the initial evaluation. Application of the SFSR to all patients revealed a sensitivity of 89% for a serious outcome. However, the sensitivity of the SFSR for identifying patients with a serious outcome, if the underlying cause for the syncope was not made during the initial visit, was only 69%.

The admission rate for the patients enrolled in the study was 58%. Application of the SFSR to guide admission decisions would have decreased the admission rate by 12%, but this would have resulted in not admitting 10% of the patients who experienced serious outcomes.

Commentary

This study by Sun et al was an effort to externally validate the findings of the San Francisco Syncope Rule (SFSR). By prospectively collecting data on patients presenting with syncope, they were able to look at the ability of the tool to detect patients at risk for a serious outcome and whether the risk was identified at the time of the initial evaluation. They also looked at the expected change in admission rate if the SFSR was applied.

Sun found that application of the SFSR was not as sensitive as seen in the derivation study by Quinn. The sensitivity decreased from 96% to 89%. Application of the rule as a guide for admission would have resulted in 10% of patients with a serious outcome not prospectively identified as being at high risk.

There are some differences between the two studies that are important to review. Sun's group obtained patients as a convenience sample and enrolled 76% of potential patients, while the derivation study by Quinn et al attempted to enroll all presenting patients. Quinn et al's sample was also larger, with 684 patients compared to 477 in Sun et al's validation study.

Despite these differences, the groups had a nearly identical incidence of serious outcomes. Eleven percent of patients presenting with syncope experienced a serious outcome within seven days in each group. Both studies were done at a single institution; in each case, it was done at a large, urban teaching hospital. Both study populations had admission decisions made by their attending physician independent of a decision tool, and this resulted in remarkably similar admission rates.

Sun et al also reviewed their data regarding the ability of the SFSR to assist risk stratification in patients who do not have a definitive diagnosis made during the initial evaluation. Whether an underlying etiology was identified by the initial evaluation in the ED was not addressed in the derivation study. This appears to be a critical group, and it appears to be the one that may benefit the most by a risk assessment tool. If an underlying etiology for the syncope is identified during the initial evaluation, no risk-stratification tool is necessary, but rather disease-specific care. Sun et al found that application of the SFSR to patients who did not have a diagnosis after their initial evaluation identified only 69% of patients at risk for a serious outcome. This sensitivity is not adequate for use in the clinical arena.

Did Other External Sites Also Fail to Replicate the Findings of the San Francisco Syncope Rule Study Group?

Source: Birnbaum A, et al. Failure to validate the San Francisco syncope rule in an independent emergency department population. Ann Emerg Med. 2008;52:151-159.

Birnbaum et al published this study detailing the results of their effort to validate the findings of the San Francisco Syncope Rule (SFSR) study group. They undertook a prospective, observational cohort study of patients presenting to a single, urban academic ED with a chief complaint of syncope or near-syncope. Patients were identified by research assistants and enrolled if the emergency physician determined that the reason for the patient's visit was syncope or near-syncope.

Patients were enrolled from January 2005 through December 2006. Four of the five predictor variables of the SFSR were collected prospectively on a structured data sheet by the attending emergency physician. The fifth predictor, an abnormal ECG, was determined by subsequent review by senior physicians blinded to the other variables. Admission and treatment decisions were made by the attending physicians without any study intervention.

A total of 738 patients were enrolled and participated in the study; follow-up was obtained on 718 (97%). A total of 61 (9%) patients experienced a serious outcome by day seven. Of those 61 patients, only 45 were identified by the SFSR as being at high risk for a serious outcome. Application of the SFSR to this study group revealed a sensitivity of 74% for identifying a patient at risk for a serious outcome.

Post hoc analysis of patients with a serious outcome when the diagnosis was not made in the ED was reviewed. Twenty-five of the 61 patients (41%) were not diagnosed in the initial evaluation. Of these 25 patients, the SFSR identified only 17 patients as at-risk for a serious outcome. This resulted in a reported sensitivity of the SFSR to predict serious outcomes in patients not diagnosed in the initial evaluation of 68%.

Commentary

This study by Birnbaum et al was unable to validate the findings of the SFSR study group. This was a well performed study that replicated the derivation group. They were able to prospectively collect data on enrolled patients, and follow-up was achieved in similar rates as the derivation study by Quinn et al. The study was performed at a single academic ED. The rate of occurrence of a serious outcome was similar (9% compared to 11%).

Birnbaum found that the application of the SFSR to their study population was not as sensitive as reported in the derivation study by Quinn et al. The sensitivity of the SFSR to identify patients who experienced a serious outcome by day seven was only 74%. This sensitivity suggests that the SFSR is not adequate for clinical use.

In addition, post hoc review was performed to assess the ability of the SFSR to identify patients at risk for a seven-day serious outcome if the diagnosis of an underlying etiology is not made in the ED. This post hoc analysis was likely prompted by the study by Sun et al, who found that the SFSR had a sensitivity of only 69% for this subgroup. Birnbaum et al had similar findings, as their calculated sensitivity for this population was only 68%. Again, this appears to be the critical group where admission decisions are complicated. Unfortunately, the available evidence suggests that the SFSR are unable to assist in risk stratification for these patients.

What Useful Clinical Policy Statements Are Available to Assist in the Evaluation and Management of Patients with Syncope?

Source: Huff JS, et al. Clinical policy: critical issues in the evaluation and management of adult patients presenting to the emergency department with syncope. Ann Emerg Med. 2007;49: 431-444.

The american college of emergency Physicians (ACEP) completed an update of the 2001 ACEP clinical policy on syncope. The policy is not directed at patients in whom the specific diagnosis of syncope is identified. This policy, instead, focuses on assisting the ED physician to address three critical questions:

1. What history and physical examination data help to risk-stratify patients with syncope?

2. What diagnostic testing data help to risk-stratify patients with syncope?

3. Who should be admitted after an episode of syncope of unclear cause?

The ACEP Clinical Policy Subcommittee on Syncope performed a review and critical analysis of the medical literature to use the best available evidence to formulate recommendations regarding these specific clinical questions. All reviewed studies were graded on their strength of evidence and study design, by at least two subcommittee members, as Class I, II, or III. Recommendations were then formulated based upon the level of evidence to support these recommendations.

Level A recommendations are generally accepted principles for patient management that reflect a high degree of clinical certainty based on Class I evidence. Level B recommendations identify management strategies based upon moderate clinical certainty in Class II studies. Level C recommendations offer management strategies that are based on preliminary or inconclusive evidence. Each clinical question was then reviewed and recommendations offered.

The first question reviewed the importance of history and physical exam evidence findings for patients with syncope. Level A recommendations stated that history or physical examination findings consistent with heart failure help identify patients at increased risk of an adverse outcome. Level B recommendations included the need to consider advanced age, structural heart disease, or a history of coronary artery disease to be risk factors for adverse outcomes. They also suggested that nonexertional syncope in younger patients with no signs of cardiac disease or family history of sudden death should be considered at low risk. There were no level C recommendations for this question.

The second question the panel reviewed regarded the importance of diagnostic testing in the risk stratification of patients with syncope. The level A recommendation states that an ECG should be obtained in all patients with syncope. There were no level B recommendations offered. Level C recommendations were that laboratory testing or further investigative testing should be performed if guided by the history or physical exam rather than by routine.

The third question addressed which patients should be admitted after syncope when an underlying etiology is not identified in the ED. The study panel was not able to offer any level A or level C recommendations. Level B recommendations were that patients with evidence of heart failure or structural heart disease should be admitted. Admission also was recommended for patients who have structural heart disease, coronary artery disease, an abnormal ECG, older age (a specific age is not selected), or a hematocrit < 30.

Commentary

The ACEP clinical policy statement clearly delineates that it is not designed to outline the evaluation of patients presenting with syncope with a specific diagnosis. This statement emphasizes the principle that patients presenting with syncope require a complete history and physical examination to identify an underlying etiology. If a specific diagnosis is made, disease-specific care can be offered. However, if an etiology is not determined, the clinical policy attempts to offer evidence to answer common questions.

The expert panel's review of the evidence reached similar conclusions to what one would draw from the articles reviewed above (of note, the articles reviewed above were instrumental in the formulation of these recommendations). In particular, both the history and physical examination, signs suggestive of congestive heart failure, correlate with an increased risk of an adverse outcome and should factor into admission decisions.

As an abnormal ECG is an important predictor of adverse outcomes, an ECG should be obtained on all patients presenting with syncope. Focused laboratory and diagnostic testing should be obtained when indicated by the history and physical examination, rather than for all patients. Finally, admission recommendations remain unclear, but risk factors of advanced age, anemia, structural heart disease, or coronary artery disease are important details that should be factored into the decision.

Conclusions

Syncope is a symptom of many varied disease processes. The search for the underlying etiology begins with a thorough history and physical examination followed by directed, indicated studies as necessary. Patients who present to an ED will benefit when a diagnosis is made, as it can lead to directed therapy of the underlying disease process. The diagnosis can be determined in up to 69% of patients presenting with syncope by the history and physical examination in combination with the ECG.4

Efforts to assist in the initial evaluation of syncope are beneficial. Syncope accounts for 1% to 3% of ED visits and up to 6% of admissions.1 Decision rules that can assist the clinician in identification of patients that can safely be discharged from the hospital can help decrease medical costs while enhancing patient safety. However, as seen in the research reviewed above, a decision rule for syncope remains elusive.

As syncope is a symptom and not a single disease process, it appears nearly impossible to formulate a decision tool which will identify all patients with any serious etiology. A 25-year-old newly pregnant female patient presenting with pelvic pain and syncope, with no complaints of shortness of breath, a systolic blood pressure of 92, no past medical history, and an initial hematocrit of 34% would be identified by the San Francisco syncope rule as not being at risk for a serious outcome. This patient could be quickly diagnosed with an ectopic pregnancy following a positive pregnancy test and a pelvic ultrasound. She is certainly at risk for a serious outcome if the diagnosis is not made during the initial evaluation and proper treatment is not provided. This simple example demonstrates the complex nature of attempting to develop a decision tool for a symptom, rather than a disease.

The significant benefit of a risk stratification tool would be for patients who fail to have a definitive diagnosis established after the initial evaluation is complete. A proper initial history and physical examination will identify a gastrointestinal hemorrhage presenting with syncope. When the initial evaluation fails to identify an etiology, the remaining concern is often a potential cardiac event, in particular a serious arrhythmia.

The studies discussed in this review by Martin et al and Quinn et al repeatedly identified that patients with a history of coronary artery disease, congestive heart failure, or an abnormal ECG are at increased risk of a serious outcome. A decision tool to assist the examiner in risk stratification for patients who may experience a serious cardiac arrhythmia would be of great benefit for the emergency physician. Unfortunately, in Sun et al's study, attempting to externally validate the SFSR application of the rule to patients who remained undiagnosed after the initial evaluation had a sensitivity of only 69%. This suggests that the SFSR is not sensitive enough to be used in clinical practice on patients with an undiagnosed etiology for the syncope in the ED.

The ACEP clinical policy on the evaluation and management of adult patients with syncope was able to review the available literature and offer general recommendations for patients that did not have a specific diagnosis made upon completion of the initial evaluation. Key recommendations are to obtain an ECG on any patient presenting with syncope and to focus on the history or signs of heart failure as a predictor of adverse outcomes.

After careful review of the available literature, is remains clear that there is no evidence-based answer regarding the safety of discharging patients who present with syncope. The importance of a complete history and physical exam cannot be over-emphasized, as many etiologies of syncope require early, accurate identification to facilitate disease-specific treatment. When the diagnosis remains unclear, history or risk of cardiac disease appear to be the best predictors of potential serious outcomes. Future research efforts should be directed to this particular patient population.

References

1. Miller TH, Kruse JE. Evaluation of syncope. Am Fam Phys. 2005; 72:1492-1500.

2. Martin TP, et al. Risk stratification of patients with syncope. Ann Emerg Med. 1997; 29:459-466.

3. Quinn J, et al. Derivation of the San Francisco syncope rule to predict patients with short-term serious outcomes. Ann Emerg Med. 2004;43:224-232.

4. Sarasin, et al. Prospective evaluation of patients with syncope: a population-based study. Am J Med. 2001;111:177-184.

5. Strickberger SA, et al. AHA/ACCF scientific statement on the evaluation of syncope. J Am Coll Cardiol. 2006;47:473-484.