Thirty-day Outcomes of Emergency Department Patients Undergoing Electrical Cardioversion for Atrial Fibrillation or Flutter


  • Frank Xavier Scheuermeyer MD, MHSc,

    1. From the Department of Emergency Medicine, St Paul’s Hospital and the University of British Columbia (FXS, EG, RS, IP, MS), Vancouver, BC; and the Division of Emergency Medicine, Foothills Hospital and the University of Calgary (GI), Calgary, AB, Canada.
    Search for more papers by this author
  • Eric Grafstein MD,

    1. From the Department of Emergency Medicine, St Paul’s Hospital and the University of British Columbia (FXS, EG, RS, IP, MS), Vancouver, BC; and the Division of Emergency Medicine, Foothills Hospital and the University of Calgary (GI), Calgary, AB, Canada.
    Search for more papers by this author
  • Rob Stenstrom MD, PhD,

    1. From the Department of Emergency Medicine, St Paul’s Hospital and the University of British Columbia (FXS, EG, RS, IP, MS), Vancouver, BC; and the Division of Emergency Medicine, Foothills Hospital and the University of Calgary (GI), Calgary, AB, Canada.
    Search for more papers by this author
  • Grant Innes MD,

    1. From the Department of Emergency Medicine, St Paul’s Hospital and the University of British Columbia (FXS, EG, RS, IP, MS), Vancouver, BC; and the Division of Emergency Medicine, Foothills Hospital and the University of Calgary (GI), Calgary, AB, Canada.
    Search for more papers by this author
  • Iraj Poureslami PhD,

    1. From the Department of Emergency Medicine, St Paul’s Hospital and the University of British Columbia (FXS, EG, RS, IP, MS), Vancouver, BC; and the Division of Emergency Medicine, Foothills Hospital and the University of Calgary (GI), Calgary, AB, Canada.
    Search for more papers by this author
  • Maziar Sighary

    1. From the Department of Emergency Medicine, St Paul’s Hospital and the University of British Columbia (FXS, EG, RS, IP, MS), Vancouver, BC; and the Division of Emergency Medicine, Foothills Hospital and the University of Calgary (GI), Calgary, AB, Canada.
    Search for more papers by this author

  • Presented at the Canadian Association of Emergency Physicians’ Annual Conference, Ottawa, ON, Canada, June 7–11, 2008.

Address for correspondence and reprints: Frank Xavier Scheuermeyer, MD; e-mail:


Objectives:  While the short-term (<7-day) safety and efficiency of electrical cardioversion for emergency department (ED) patients with atrial fibrillation or flutter have been established, the 30-day outcomes with respect to stroke, thromboembolic events, or death have not been investigated.

Methods:  A two-center cohort of consecutive ED patients undergoing cardioversion for atrial fibrillation or flutter between January 1, 2000, and September 30, 2007, was retrospectively investigated. This cohort was probabilistically linked with both a regional ED database and the provincial health registry to determine which patients had a subsequent ED visit or hospital admission, stroke, or thromboembolic event or died within 30 days. In addition, trained reviewers performed a detailed chart abstraction on 150 randomly selected patients, with emphasis on demographics, vital signs, medical treatment, and predefined adverse events. Hemodynamically unstable patients or those whose condition was the result of an underlying acute medical diagnosis were excluded. Data were analyzed by descriptive methods.

Results:  During the study period, 1,233 patients made 1,820 visits for atrial fibrillation or flutter to the ED. Of the 400 eligible patients undergoing direct-current cardioversion (DCCV), no patients died, had a stroke, or had a thromboembolic event in the following 30 days (95% confidence interval [CI] = 0.0 to 0.8% for all outcomes). A total of 141 patients were included in the formal chart review, with five patients (3.5%, 95% CI = 0.5% to 6.6%) failing cardioversion, six patients (4.3%, 95% CI = 0.9% to 7.6%) having a minor adverse event that did not change disposition, and five patients (3.5%, 95% CI = 0.5% to 6.6%) admitted to hospital at the index visit.

Conclusions:  Cardioversion of patients with atrial fibrillation or flutter in the ED appears to have a very low rate of long-term complications.

ACADEMIC EMERGENCY MEDICINE 2010; 17:408–415 © 2010 by the Society for Academic Emergency Medicine

Atrial fibrillation is the most common arrhythmia presenting to the emergency department (ED) and is a public health problem associated with significant morbidity and mortality.1–3 Chronic atrial fibrillation increases stroke risk 5- to 17-fold,4–6 and patients with paroxysmal atrial fibrillation may have a similar prognosis.7,8 In the latter group, the 2006 American Heart Association guidelines conclude that rhythm control using direct-current cardioversion (DCCV) is an acceptable strategy to restore sinus rhythm.9 This treatment is limited to patients whose arrhythmia has persisted less than 48 hours, unless they are appropriately anticoagulated.9,10 However, up to 12% of patients with chronic atrial fibrillation may not experience traditional symptoms such as palpitations, dyspnea, and chest pain.11 Therefore, it is unclear what proportion of ED patients presenting with atrial fibrillation or flutter have “silent” atrial fibrillation for greater than 48 hours that may put them at risk for stroke, thromboembolic events, or death.

Retrospective cohort studies have concluded that electrical and chemical cardioversion in the ED are safe and efficient, but these studies reported only immediate or short-term (≤7-day) outcomes.12–14 A prospective ED trial (n = 56) demonstrated no 6-month strokes, thromboembolic events, or deaths.15 Recent meta-analyses have concluded that patients undergoing electrical cardioversion for atrial fibrillation lasting ≥7 days have a tendency toward increased stroke risk compared to patients treated with rhythm control alone.16,17 The latter studies were not conducted in an ED setting and may involve a different group of patients than those encountered in the ED. Furthermore, they are confounded by differing rates of anticoagulation in the cardioversion- and rhythm-control groups. The comparative intermediate- and long-term risk of ED cardioversion is therefore unknown.

The objective of this study was to describe the 30-day outcomes of a large cohort of ED patients undergoing DCCV for atrial fibrillation or flutter. We hypothesized that the risk of stroke, thromboembolic event, or death in ED patients undergoing DCCV for atrial fibrillation or flutter is small.


Study Design

This was an administrative database search and retrospective chart audit to identify consecutive ED patients who underwent DCCV for atrial fibrillation or flutter at two urban Vancouver EDs. This study was approved by the Ethics Committee of Providence Health Care and the University of British Columbia.

Study Setting and Population

St. Paul’s Hospital is an urban, tertiary care center with an annual census of 60,000 that has a dedicated cardiac electrophysiology service. Mt. St. Joseph’s Hospital is an affiliated community center that has 26,000 annual patient visits and internal medicine specialist availability. The Vancouver Coastal Health region has six hospitals with approximately 250,000 annual ED patient encounters. Regional ED information, including patient demographics, visits, diagnoses, and visit disposition, can be linked through the use of a unique patient provincial health care number, to monitor patient visits across the entire region. At our institutions, all ED cardioversions were performed by emergency physicians (EPs). During the study period, there was no standard algorithm for the ED management of atrial fibrillation or flutter. Physicians varied in their preferences for rate or rhythm control, chemical, or electrical cardioversion, and antiarrhythmic medications.

Consecutive eligible patients were identified by a query of the St. Paul’s Hospital and Mt. St. Joseph’s Hospital ED databases from January 1, 2000, to September 30, 2007. ED protocol mandates that physicians create an electronic discharge summary for each patient listing diagnoses and procedures. Patients with a diagnosis of “atrial fibrillation” or “atrial flutter” (International Classification of Diseases 10th Revision [ICD-10] code I48.0 or I48.1 respectively), who had DCCV listed under the “procedures” category of the ED discharge summary and had not been enrolled in the previous 30 days, were included. To ensure appropriate enrollment, 10% of all charts with a discharge diagnosis of “atrial fibrillation” or “atrial flutter” were reviewed to verify accuracy of physician coding with respect to DCCV and the presence of study eligibility criteria.

Study Protocol

Data Collection (Overall Cohort).  After subject identification, the ED administrative database was interrogated to capture individual patient demographics, triage acuity, and presenting complaint. In addition, this database contains all laboratory, imaging, and medication orders; consultations; patient disposition; and admitting and discharge times. To identify long-term outcomes, a linked regional ED database was queried to identify any patient visit to a regional hospital within 30 days. Subsequent visits with a triage code of “chest pain,”“dysrhythmia,”“dyspnea” or “respiratory distress,”“stroke” or “extremity weakness” or “dizziness,” any extremity pain, and “cardiac arrest” were treated as potentially related to the index visit and investigated. The subsequent ED discharge diagnosis and disposition for second and further visits was also obtained. The British Columbia Vital Statistics database was cross-referenced to identify any patients who died within 30 days of the index visit, and the cause of death was elucidated. For the purpose of our outcomes, stroke was defined as an acute neurologic deficit of cerebrovascular cause, while a thromboembolic event referred to peripheral arterial embolic disease.

Data Collection (Chart Review for Patient Characteristics and ED Management).  To describe patient characteristics and ED management, 150 patients were selected at random for a detailed chart review by one of two trained reviewers who were blinded to patient outcomes. Patient ages and comorbidities were abstracted to generate individual CHADS2 scores.18 This score has been used to risk-stratify patients with chronic atrial fibrillation according to their likelihood of subsequent stroke, with a score of 0 indicating a 1-year stroke risk of 2% and a score of 6 denoting a risk of 18%. It has not been previously used in an ED population, and we did not use it for atrial flutter. In addition, time of onset and duration of symptoms, initial and final vital signs, medications, prior cardiac arrhythmias and treatment, and all electrocardiograms and cardiac rhythm strips were included. Patients with an underlying medical etiology for their atrial arrhythmia (shock, sepsis, pulmonary embolus, thyroid storm or thyrotoxicosis, hypertensive emergency, acute valvular disease, acute pericarditis or myocarditis, medication or recreational drug overdose, hypothermia), or were a nonresident of the province of British Columbia, were excluded, as were patients with a terminal illness and those who required immediate cardioversion due to hemodynamic instability (syncope, ischemic chest pain, dyspnea, congestive heart failure, or decreased level of consciousness). All medications relating to both analgesia and procedural sedation were recorded, as well as the number and type of electrical shocks. Electrical cardioversion was considered successful if normal sinus rhythm was achieved and maintained within three attempts.

Charts were also abstracted for the following predefined immediate complications related to procedural sedation and DCCV: hypotension requiring intravenous fluids or vasoactive agents within 1 hour of initiation of procedural sedation; respiratory compromise requiring intervention (i.e., bag-valve mask, oral airway, noninvasive positive pressure ventilation, intubation); new arrhythmias requiring pharmacologic intervention within 1 hour of initiation of procedural sedation; acute allergic reactions requiring medications; emesis or nausea requiring antiemetic medications; new confirmed strokes or thromboembolic events; unplanned admission to the hospital; and death. For all cases where a complication occurred, charts were further reviewed to determine whether these led to a change in management or disposition. Complications were defined as serious if they resulted in an unplanned admission. Interrater reliability for data abstraction was assessed by having the reviewers conduct independent blinded reviews of 30 of 150 charts (20%) and calculating a kappa statistic for the following categories: duration of atrial fibrillation, warfarin use, prior atrial fibrillation or flutter, and adverse events.

Data Analysis

Microsoft Excel 2008 (Microsoft Corp., Redmond, WA) was used for data entry and basic analysis. Continuous variables were presented as medians or means with standard deviations (SDs). Discrete variables were reported as percentages.


From January 1, 2000, to September 30, 2007, a total of 1,233 patients had 1,820 visits for atrial fibrillation or flutter. The study flow is presented in Figure 1. Of 182 charts randomly selected to check coding accuracy, two had “DCCV” omitted from their chart summaries although they had a successful procedure with uneventful ED discharge, for a 1% coding error rate. Overall, 409 patients had DCCV, and 150 of these patients were randomly selected for detailed chart audit to describe patient characteristics and ED management. Interobserver agreement for data abstraction ranged from good (κ = 0.65 for duration of atrial fibrillation) to excellent (κ = 1.0 for warfarin use, prior atrial arrhythmias, and adverse events.) Baseline characteristics, including presentation, triage codes, vital signs, comorbidities, and medications, are described in Table 1.

Figure 1.

 Study flow (*patients excluded from review also excluded from long-term data linkage). DCCV = direct-current cardioversion.

Table 1. 
Baseline Characteristics (From Formal Chart Review, n = 141)
CharacteristicAtrial FibrillationAtrial FlutterCombined
  1. *Except sotalol.

  2. BP = blood pressure; EMS = emergency medical services; CCS = Canadian Cardiovascular Society classification; CTAS = Canadian Triage and Acuity Scale; DCCV = direct current cardioversion; NYHA = New York Heart Association classification; TIA = transient ischemic attack; CHADS2 score = congestive heart failure, hypertension, age, diabetes, stroke.

Number of patients (%)124 (87%) 17 (13%)141
 Age (yr), mean ± SD 57 ± 14 56 ± 13 57 ± 14
 Sex (male), n (%) 92 (74.2%) 13 (76.4%)105 (74.5%)
 EMS arrival, n (%) 13 (10.4%)  4 (23.5%) 17 (12%)
Triage level, n (%)
 CTAS 2 44 (35%)  4 (24%) 48 (34%)
 CTAS 3 80 (65%) 13 (76%) 93 (66%)
Initial vitals on index ED visit
 Heart rate, mean ± sd123 ± 30119 ± 33122 ± 30
 Systolic BP (mm Hg), mean ± SD126 ± 25131 ± 30127 ± 24
 Diastolic BP (mm Hg), mean ± sd 79 ± 14 84 ± 20 80 ± 15
 Respiratory rate (breaths/min), mean ± SD 16 ± 2 18 ± 1 16 ± 2
 Arrhythmia duration (hr), mean ± sd  8 ± 6 12 ± 6  8 ± 6
 Arrhythmia duration (hr), range  1–150  2–100  1–150
 Arrhythmia duration <24 hr, n (%)118 (95%) 15 (88%)133 (94%)
 Arrhythmia duration 24–48 hr, n (%)  4 (3%)  1 (6%)  5 (4%)
 Arrhythmia duration >48 hr, n (%)  2 (2%)  1 (6%)  3 (2%)
 Arrhythmia duration >48 hr and no therapeutic anticoagulation, n (%)  0 (0%)  0 (0%)  0 (0%)
Prior arrhythmias, n (%)
 Prior atrial fibrillation or flutter 82 (66%) 10 (59%) 92 (65%)
 Prior successful DCCV 73 (59%)  9 (53%) 82 (58%)
 Prior successful chemical cardioversion  6 (4%)  0 (0%)  6 (4%)
 Number of prior DCCVs, range  0–6  0–9  0–9
Risk factors, n (%)
 Hypertension 43 (35%)  5 (29%) 48 (34%)
 Coronary artery disease (CCS 3/4) 10 (8%)  0 (0%) 10 (7%)
 Valvular disease 19 (15%)  3 (18%) 22 (16%)
 Thyroid disease  0 (0%)  0 (0%)  0 (0%)
 Diabetes  4 (3%)  0 (0%)  4 (3%)
 Congestive heart failure (NYHA 3/4)  3 (2%)  1 (6%)  4 (3%)
 Prior stroke or TIA  0 (0%)  0 (0%)  0 (0%)
CHADS2 score (only for Afib), n (%)
 CHADS2 score 0 75 (61%)  
 CHADS2 score 1 46 (37%)  
 CHADS2 score >1  3 (2%)  
Concurrent medications, n (%)
 Aspirin 46 (37%)  5 (29%) 51 (36%)
 Clopidogrel  1 (1%)  0 (0%)  1 (1%)
 Warfarin 20 (16%)  5 (29%) 25 (18%)
 Beta blocker* 42 (34%)  5 (29%) 47 (33%)
 Calcium channel blocker 14 (11%)  2 (12%) 16 (11%)
 Propafenone 13 (10%)  1 (6%) 14 (9%)
 Amiodarone  7 (6%)  2 (12%)  9 (6%)
 Sotalol 13 (10%)  3 (17%) 16 (11%)
 Digoxin  2 (2%)  2 (12%)  4 (3%)

Of the 150 patients selected for formal chart review, nine had atrial fibrillation or flutter due to underlying acute medical issues as listed under Methods or were hemodynamically unstable. Excluding these patients left 141 for analysis. The nine patients were also removed from the overall cohort, leaving 400 for 30-day follow-up. There were 342 cases of atrial fibrillation (85.5%) and 58 cases of atrial flutter (14.5%). The majority of patients had prior atrial fibrillation or flutter with a successful cardioversion. Patients were generally healthy, with the majority having a CHADS2 score of zero, and only 2% (3/124) having a score of >1. The self-reported duration of atrial fibrillation or flutter in the study sample ranged from <1 to 120 hours, with the majority being <12 hours. Of the patients with symptom onset >48 hours previous, only patients with therapeutic anticoagulation had an attempted ED cardioversion. Table 2 shows that few patients had rate control or chemical cardioversion attempted prior to electrical cardioversion. All patients received procedural sedation prior to electrical cardioversion. Propofol was used as the sole agent nearly half the time and was accompanied by either a sedative or an analgesic in the majority of the remaining cases. Most cardioversions were accomplished by a single DC shock. The only adverse events involved six patients who experienced transient respiratory depression as a result of procedural sedation; none resulted in an unplanned admission. Index visit complications relating to procedural sedation or electrical cardioversion can be reviewed in Table 2. Only five (3.5%) attempted electrical cardioversions failed, but there was no correlation with age or comorbidity. No variables or combinations could be elucidated to formulate a predictor rule to determine success or failure. Table 3 summarizes patient resource use and disposition. Cardiology consultation was arranged for 22 of 141 cases (16%) and was typically related to unsuccessful DCCV, although only five patients were admitted. Four patients had continued uncontrolled atrial fibrillation or flutter that was refractory to both electrical and chemical cardioversion and one had severe hypertrophic obstructive cardiomyopathy.

Table 2. 
Procedural Sedation and Cardioversion Outcomes (From Chart Review, n = 141)
ParameterAtrial FibrillationAtrial FlutterCombined
  1. Values are %, mean (±SD), or n (%, 95% CI)

  2. DCCV = direct current cardioversion; dBP = diastolic blood pressure; HR = heart rate; sBP = systolic blood pressure.

  3. *Complications at index visit were defined as emesis, nausea requiring antiemetic medications allergic reaction requiring medications, respiratory distress requiring intervention (bag-valve mask, oral airway, noninvasive positive pressure ventilation, intubation), hypotension requiring intravenous fluids or vasoactive agents within 1 hour of initiation of procedural sedation, new bradycardia requiring pharmacologic intervention or pacing within 1 hour of initiation of procedural sedation, unplanned admission, death, or confirmed thromboembolic event.

Chemical cardioversion attempted prior to DCCV at index ED visit11 (8.9)0 (0)11 (7.8)
Amiodarone4 (3.2)0 (0)4 (2.8)
Sotalol1 (0.8)0 (0)1 (0.7)
Procainamide6 (4.8)0 (0)6 (4.3)
Prior rate control attempted in ED6 (4.8)0 (0)6 (4.3)
Beta-blocker3 (2.4)0 (0)3 (2.1)
Calcium channel blocker3 (2.4)0 (0)3 (2.1)
Sedation regime
 Propofol58 (47)10 (59)68 (48)
 Propofol + ketamine15 (12)2 (12)17 (12)
 Propofol + fentanyl33 (27)3 (18)36 (27)
 Propofol + midazolam8 (6)0 (0)8 (5)
 Fentanyl + midazolam10 (8)2 (12)12 (8)
 Received periprocedural anticoagulation8 (6.5)0 (0)8 (5.7)
 Enoxaparin3 (2.4)0 (0)3 (2.1)
 Dalteparin5 (4.0)0 (0)5 (0.35)
Number of shocks to facilitate conversion
 189 (72)12 (71)101 (72)
 219 (15)3 (18)22 (16)
 312 (10)1 (12)13 (9)
 >3 (defined as failure)4 (3)1 (12)5 (3)
Successful cardioversion energy (J)
 <502 (2)1 (6)3 (2)
 5032 (26)4 (24)36 (26)
 10048 (39)8 (47)56 (40)
 15020 (16)0 (0)20 (14)
 20022 (17)4 (24)26 (18)
Final HR (beats/min) 72 ± 1870 ± 2072 ± 18
Final sBP (mm Hg)115 ± 15102 ± 19113 ± 15
Final dBP (mm Hg)70 ± 1059 ± 1569 ± 9
Failed DCCV, n (%, 95% CI)4 (3.2, 0.1–6.3)1 (5.9, 0–17.6)5 (3.5, 0.5–6.6)
ED complications of sedation or cardioversion,*n (%, 95% CI)
 Nausea0 (0, 0–2.4)0 (0, 0–17.7)0 (0, 0–2.1)
 Emesis0 (0, 0–2.4)0 (0, 0–17.7)0 (0, 0–2.1)
 Allergic reaction0 (0, 0–2.4)0 (0, 0–17.7)0 (0, 0–2.1)
 Hypotension0 (0, 0–2.4)0 (0, 0–17.7)0 (0, 0–2.1)
 Bradycardia0 (0, 0–2.4)0 (0, 0–17.7)0 (0, 0–2.1)
 Respiratory compromise5 (4.0, 0.6–7.5)1 (5.9, 0–17.6)6 (4.3, 0.9–7.6)
 Unplanned admission0 (0, 0–2.4)0 (0, 0–17.7)0 (0, 0–2.1)
 Stroke0 (0, 0–2.4)0 (0, 0–17.7)0 (0, 0–2.1)
 Thromboembolic event0 (0, 0–2.4)0 (0, 0–17.7)0 (0, 0–2.1)
 Death0 (0, 0–2.4)0 (0, 0–17.7)0 (0, 0–2.1)
 Total complications5 (4.0, 0.6–7.5)1 (5.9, 0–17.6)6 (4.3, 0.9–7.6)
Table 3. 
Resource Use at Index Visit (From Chart Review, n = 141)
ParameterAtrial FibrillationAtrial FlutterCombined
  1. LOS = length of stay.

Cardiology consult, n (%) 20 (16) 2 (12) 22 (16%)
Cardiology admission, n (%)  4 (3.2) 1 (5.9)  5 (3.5)
ED echocardiography, n (%)  0 (0) 0 (0)  0 (0)
Mean (±SD) ED LOS (hr)  4:43 ± 2:30 7:55 ± 3:41  5:07 ± 2:15
Median ED LOS (hr)  3:55 4:30  3:59
Mean (±SD) hospital LOS for admitted patients (days)  2 ± 2 (4 patients   admitted) 2 (1 patient admitted)  2 ± 2 (5 patients  admitted)

Table 4 summarizes 30-day outcomes for all 400 patients. Only 22 patients (5.5%) made a visit to a regional ED that was potentially related to the index visit, with six patients (1.5%) requiring repeat DCCV (one patient required two cardioversions). Two patients (0.5%) were admitted. No patient had a documented stroke, thromboembolic event, or died in the 30 days following the index visit. Even at the lower confidence boundary, the rate of stroke, thromboembolic event, or death for the overall group is less than 1%.

Table 4. 
30-day Outcomes (From Linked Regional and Provincial Databases, n = 400)
ParameterAtrial FibrillationAtrial FlutterCombined
  1. *For visit related to atrial fibrillation or flutter.

  2. †One patient had two DCCVs for atrial fibrillation within 30 days.

  3. DCCV = direct current cardioversion.

New stroke, n (%, 95% CI)0 (0, 0.0–0.9)0 (0, 0.0–5.2)0 (0, 0.0–0.8)
New thromboembolic event, n (%, 95% CI)0 (0, 0.0–0.9)0 (0, 0.0–5.2)0 (0, 0.0–0.8)
Death, n (%, 95% CI)0 (0, 0.0–0.9)0 (0, 0.0–5.2)0 (0, 0.0–0.8)
Return to ED, n (%)*10 (2.9)2 (3.4)12 (3.0)
ED DCCV, n (%)5† (1.5)2 (3.4)7 (1.8)
Admission to hospital, n (%)*2 (0.6)0 (0)2 (0.5)


In this study of 400 ED patients who underwent DCCV for atrial fibrillation or flutter, none suffered significant complications, stroke, new thromboembolic events, or death within 30 days. In contrast with other studies,12 almost all of our patients received primary electrical cardioversion; few had preceding rate control or antiarrhythmic drug therapy. Six patients required assisted ventilations with a bag-valve mask during the sedation procedure, but no other adverse events were seen, no patients had a change in disposition as a result of their treatment, and only a few patients had repeat ED visits, subsequent DCCVs, or hospitalizations within 30 days.

Increasing adoption of rate-control strategies (based on data from non-ED populations) has reduced the impetus to restore sinus rhythm, and a growing awareness of thromboembolic complications after atrial fibrillation conversion has raised concerns about the safety and appropriateness of ED cardioversion. Berger and Schweitzer19 published a retrospective analysis of 4,621 patients undergoing elective DCCV, of whom 88 developed an embolic episode within 7 days and eight additional patients within 18 days. Of these 96 events, only three occurred in patients with duration of atrial fibrillation less than 48 hours. Weigner and coworkers20 described only three embolic events in a cohort of 375 patients admitted to a cardiology unit for atrial fibrillation. In this group, 250 patients (66.7%) also had spontaneous conversion to sinus rhythm. The only predictor of increased probability of spontaneous cardioversion was atrial fibrillation duration of <24 hours;21 similar results were reported by Tejan-Sie and colleagues.22 These findings have left uncertainty as to the time frame when ED cardioversion is appropriate, the optimal technique of cardioversion, and the best risk stratification tool to apply in determining treatment approach

It is critical to point out that the above reports20–22 and the Atrial Fibrillation Follow-up Investigation of Rhythm Management (AFFIRM) trial23 describe patients who are substantially different from those studied in this and other ED-based cardioversion cohorts. Because rates of stroke, thromboembolism, and death are more closely related to underlying patient factors than to treatment approach, it may be inappropriate to extrapolate findings from non-ED settings to the management of relatively young healthy ED patients with atrial fibrillation or flutter. Our data support the findings of Decker and coworkers,15 who reported no strokes, thromboembolic events, or deaths at 6-month follow-up in 153 ED patients with atrial fibrillation, one-third of whom underwent DCCV. Two other ED-based studies found similar low adverse event rates.12,13 In our 8-year review of DCCV for ED patients with atrial fibrillation or flutter, there were few short-term adverse events and no 30-day complications related to the procedure. The age, sex, and risk profile of our patients is consistent with prior ED-based studies,14–17 although the CHADS2 score has not been reported previously.

Our patients all had self-reported onset of atrial fibrillation or flutter within 48 hours or else were appropriately anticoagulated. Many had previous successful DCCVs. No patients had formal transthoracic or transesophageal echocardiography to assess ventricular function or confirm or refute atrial clot burden, an approach considered optimal for atrial fibrillation,24 but debatable for atrial flutter.25 A small minority were given single-dose periprocedural anticoagulation, but the utility of this practice is unclear. Thromboembolic events from conversion to normal sinus rhythm are known to occur up to 18 days after cardioversion,19 related to stunning of the left atrial appendage,26,27 so it is improbable that a single dose of anticoagulant would be protective. Of the 141 patients formally reviewed in our setting, only four were discharged with a documented new prescription for aspirin and three for coumadin, perhaps reflecting the low CHADS2 score.

The ED length of stay in this study ranged from 1.4 to 22.8 hours, with a median time of 4 hours. Longer stays were typically a result of prior attempts at rate control or chemical cardioversion, cardiology consultation, and nighttime presentation (with single EP coverage in a busy department at night, the ability to cardiovert is often limited). However, this is less time than the 11.8 hours reported by Koenig et al.,28 the 10.1 hours reported by Decker et al.,14 and the 9 hours reported by Domanovits et al.29 and comparable to the 5 hours reported by Michael et al..12 Other investigators have found that a greater percentage of patients undergo a trial of chemical rhythm control prior to electrical cardioversion, but most patients in our study were only managed with DCCV. The speed and apparent safety of our primary electrical approach may result in lower system utilization for patients requiring conversion.

Although the study is retrospective, the chart review confirms that the patients in our study are similar to those in the ED literature, increasing the potential generalizability of our findings. Rigorous use of data linkages between our ED database, the regional ED database, and the provincial vital statistics registry allows for long-term tracking of individuals.


Results of this two-center study may not be generalizable to diverse settings. Retrospective studies are subject to missing or improperly coded data, although our data audit indicated an error rate in the range of 1%, which should minimally affect results. Based on our chart review, 9 of 150 (6%) patients receiving cardioversion were excluded due to instability or underlying acute medical illness. Patients not undergoing a chart review (n = 259) would not have been scrutinized in this manner. It is probable that unstable patients were inadvertently included in the 30-day follow-up and would have been excluded had their charts been scrutinized, but such patients would be expected to have worse outcomes. The duration of atrial fibrillation or flutter is based on self-reported patient histories and may not be accurate. DCCV took place in a group of young patients with low CHADS2 scores who had been in atrial fibrillation or flutter for only a short time, which may prevent extrapolation to sicker patients. While our regional ED database identified all repeat visits to nearby hospitals, patients visiting a nonregional ED within 30 days could be overlooked. Patients lacking a provincial health number or using more than one might escape linkage as well.


This study expands on previous work by enlarging the time frame during which the rates of post-ED cardioversion strokes, thromboembolic events, and mortality are very low. This suggests that ED selection of cardioversion candidates is appropriate and that the procedure is safe up to 30 days for both atrial fibrillation and flutter. A larger multicenter trial with a longer follow-up period would provide greater confidence in the conclusion that there is minimal risk associated with this approach.