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

Thirty-day Outcomes of EmergencyDepartment Patients Undergoing ElectricalCardioversion for Atrial Fibrillation or FlutterFrank Xavier Scheuermeyer, MD, MHSc, Eric Grafstein, MD, Rob Stenstrom, MD, PhD,Grant Innes, MD, Iraj Poureslami, PhD, and Maziar Sighary

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

Methods: A two-center cohort of consecutive ED patients undergoing cardioversion for atrial fibrillationor flutter between January 1, 2000, and September 30, 2007, was retrospectively investigated. This cohortwas probabilistically linked with both a regional ED database and the provincial health registry to deter-mine which patients had a subsequent ED visit or hospital admission, stroke, or thromboembolic eventor died within 30 days. In addition, trained reviewers performed a detailed chart abstraction on 150 ran-domly selected patients, with emphasis on demographics, vital signs, medical treatment, and predefinedadverse events. Hemodynamically unstable patients or those whose condition was the result of anunderlying 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 theED. Of the 400 eligible patients undergoing direct-current cardioversion (DCCV), no patients died, had astroke, or had a thromboembolic event in the following 30 days (95% confidence interval [CI] = 0.0 to0.8% for all outcomes). A total of 141 patients were included in the formal chart review, with fivepatients (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% to6.6%) admitted to hospital at the index visit.

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

ACADEMIC EMERGENCY MEDICINE 2010; 17:408–415 ª 2010 by the Society for Academic EmergencyMedicine

Keywords: atrial fibrillation, atrial flutter, electric countershock, treatment outcome

A trial fibrillation is the most common arrhythmiapresenting to the emergency department (ED)and is a public health problem associated with

significant morbidity and mortality.1–3 Chronic atrialfibrillation increases stroke risk 5- to 17-fold,4–6 andpatients with paroxysmal atrial fibrillation may have a

similar prognosis.7,8 In the latter group, the 2006 Ameri-can Heart Association guidelines conclude that rhythmcontrol using direct-current cardioversion (DCCV) is anacceptable strategy to restore sinus rhythm.9 This treat-ment is limited to patients whose arrhythmia has per-sisted less than 48 hours, unless they are appropriatelyanticoagulated.9,10 However, up to 12% of patients withchronic atrial fibrillation may not experience traditionalsymptoms such as palpitations, dyspnea, and chestpain.11 Therefore, it is unclear what proportion of EDpatients presenting with atrial fibrillation or flutter have‘‘silent’’ atrial fibrillation for greater than 48 hours thatmay put them at risk for stroke, thromboembolic events,or death.

Retrospective cohort studies have concluded thatelectrical and chemical cardioversion in the ED are safeand efficient, but these studies reported only immediateor short-term (£7-day) outcomes.12–14 A prospective ED

ISSN 1069-6563 ª 2010 by the Society for Academic Emergency Medicine408 PII ISSN 1069-6563583 doi: 10.1111/j.1553-2712.2010.00697.x

From the Department of Emergency Medicine, St Paul’s Hospi-tal 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.Received August 5, 2009; revisions received September 15 andSeptember 19, 2009; accepted September 21, 2009.Address for correspondence and reprints: Frank Xavier Sche-uermeyer, MD; e-mail: [email protected] at the Canadian Association of Emergency Physici-ans’ Annual Conference, Ottawa, ON, Canada, June 7–11, 2008.

trial (n = 56) demonstrated no 6-month strokes, throm-boembolic events, or deaths.15 Recent meta-analyseshave concluded that patients undergoing electrical car-dioversion for atrial fibrillation lasting ‡7 days have atendency toward increased stroke risk compared topatients treated with rhythm control alone.16,17 Thelatter studies were not conducted in an ED setting andmay involve a different group of patients than thoseencountered in the ED. Furthermore, they are con-founded by differing rates of anticoagulation in thecardioversion- and rhythm-control groups. The com-parative intermediate- and long-term risk of ED cardio-version is therefore unknown.

The objective of this study was to describe the 30-dayoutcomes of a large cohort of ED patients undergoingDCCV for atrial fibrillation or flutter. We hypothesizedthat the risk of stroke, thromboembolic event, or deathin ED patients undergoing DCCV for atrial fibrillationor flutter is small.

METHODS

Study DesignThis was an administrative database search and retro-spective chart audit to identify consecutive ED patientswho underwent DCCV for atrial fibrillation or flutter attwo urban Vancouver EDs. This study was approved bythe Ethics Committee of Providence Health Care andthe University of British Columbia.

Study Setting and PopulationSt. Paul’s Hospital is an urban, tertiary care centerwith an annual census of 60,000 that has a dedicatedcardiac electrophysiology service. Mt. St. Joseph’sHospital is an affiliated community center that has26,000 annual patient visits and internal medicine spe-cialist availability. The Vancouver Coastal Healthregion has six hospitals with approximately 250,000annual ED patient encounters. Regional ED informa-tion, including patient demographics, visits, diagnoses,and visit disposition, can be linked through the use ofa unique patient provincial health care number, tomonitor patient visits across the entire region. At ourinstitutions, all ED cardioversions were performed byemergency physicians (EPs). During the study period,there was no standard algorithm for the ED manage-ment of atrial fibrillation or flutter. Physicians varied intheir preferences for rate or rhythm control, chemical,or electrical cardioversion, and antiarrhythmic medica-tions.

Consecutive eligible patients were identified by aquery of the St. Paul’s Hospital and Mt. St. Joseph’s Hos-pital ED databases from January 1, 2000, to September30, 2007. ED protocol mandates that physicians create anelectronic discharge summary for each patient listingdiagnoses and procedures. Patients with a diagnosis of‘‘atrial fibrillation’’ or ‘‘atrial flutter’’ (International Clas-sification of Diseases 10th Revision [ICD-10] code I48.0or I48.1 respectively), who had DCCV listed under the‘‘procedures’’ category of the ED discharge summaryand had not been enrolled in the previous 30 days, wereincluded. To ensure appropriate enrollment, 10% of allcharts with a discharge diagnosis of ‘‘atrial fibrillation’’

or ‘‘atrial flutter’’ were reviewed to verify accuracy ofphysician coding with respect to DCCV and the presenceof study eligibility criteria.

Study ProtocolData Collection (Overall Cohort). After subject iden-tification, the ED administrative database was interro-gated to capture individual patient demographics,triage acuity, and presenting complaint. In addition,this database contains all laboratory, imaging, andmedication orders; consultations; patient disposition;and admitting and discharge times. To identify long-term outcomes, a linked regional ED database wasqueried to identify any patient visit to a regional hospi-tal within 30 days. Subsequent visits with a triage codeof ‘‘chest pain,’’ ‘‘dysrhythmia,’’ ‘‘dyspnea’’ or ‘‘respira-tory distress,’’ ‘‘stroke’’ or ‘‘extremity weakness’’ or‘‘dizziness,’’ any extremity pain, and ‘‘cardiac arrest’’were treated as potentially related to the index visitand investigated. The subsequent ED discharge diag-nosis and disposition for second and further visits wasalso obtained. The British Columbia Vital Statisticsdatabase was cross-referenced to identify any patientswho died within 30 days of the index visit, and thecause of death was elucidated. For the purpose of ouroutcomes, stroke was defined as an acute neurologicdeficit of cerebrovascular cause, while a thromboem-bolic event referred to peripheral arterial embolic dis-ease.

Data Collection (Chart Review for Patient Character-istics and ED Management). To describe patientcharacteristics and ED management, 150 patients wereselected at random for a detailed chart review by oneof two trained reviewers who were blinded to patientoutcomes. Patient ages and comorbidities were abstrac-ted to generate individual CHADS2 scores.18 This scorehas been used to risk-stratify patients with chronicatrial fibrillation according to their likelihood of sub-sequent stroke, with a score of 0 indicating a 1-yearstroke risk of 2% and a score of 6 denoting a risk of18%. It has not been previously used in an ED popula-tion, and we did not use it for atrial flutter. In addition,time of onset and duration of symptoms, initial and finalvital signs, medications, prior cardiac arrhythmias andtreatment, and all electrocardiograms and cardiacrhythm strips were included. Patients with an underly-ing medical etiology for their atrial arrhythmia (shock,sepsis, pulmonary embolus, thyroid storm or thyrotoxi-cosis, hypertensive emergency, acute valvular disease,acute pericarditis or myocarditis, medication or rec-reational drug overdose, hypothermia), or were a non-resident of the province of British Columbia, wereexcluded, as were patients with a terminal illness andthose who required immediate cardioversion due to he-modynamic instability (syncope, ischemic chest pain,dyspnea, congestive heart failure, or decreased level ofconsciousness). All medications relating to both analge-sia and procedural sedation were recorded, as well asthe number and type of electrical shocks. Electrical car-dioversion was considered successful if normal sinusrhythm was achieved and maintained within threeattempts.

ACAD EMERG MED • April 2010, Vol. 17, No. 4 • www.aemj.org 409

Charts were also abstracted for the following prede-fined immediate complications related to proceduralsedation and DCCV: hypotension requiring intravenousfluids or vasoactive agents within 1 hour of initiation ofprocedural sedation; respiratory compromise requiringintervention (i.e., bag-valve mask, oral airway, noninva-sive positive pressure ventilation, intubation); new ar-rhythmias requiring pharmacologic intervention within1 hour of initiation of procedural sedation; acute aller-gic reactions requiring medications; emesis or nausearequiring antiemetic medications; new confirmedstrokes or thromboembolic events; unplanned admis-sion to the hospital; and death. For all cases where acomplication occurred, charts were further reviewed todetermine whether these led to a change in manage-ment or disposition. Complications were defined asserious if they resulted in an unplanned admission. In-terrater reliability for data abstraction was assessed byhaving the reviewers conduct independent blindedreviews of 30 of 150 charts (20%) and calculating akappa statistic for the following categories: duration ofatrial fibrillation, warfarin use, prior atrial fibrillation orflutter, and adverse events.

Data AnalysisMicrosoft Excel 2008 (Microsoft Corp., Redmond, WA)was used for data entry and basic analysis. Continuousvariables were presented as medians or means withstandard deviations (SDs). Discrete variables werereported as percentages.

RESULTS

From January 1, 2000, to September 30, 2007, a total of1,233 patients had 1,820 visits for atrial fibrillation orflutter. The study flow is presented in Figure 1. Of 182charts randomly selected to check coding accuracy, twohad ‘‘DCCV’’ omitted from their chart summariesalthough they had a successful procedure with unevent-ful ED discharge, for a 1% coding error rate. Overall,409 patients had DCCV, and 150 of these patients wererandomly selected for detailed chart audit to describepatient characteristics and ED management. Interob-server agreement for data abstraction ranged fromgood (j = 0.65 for duration of atrial fibrillation) to excel-lent (j = 1.0 for warfarin use, prior atrial arrhythmias,and adverse events.) Baseline characteristics, includingpresentation, triage codes, vital signs, comorbidities,and medications, are described in Table 1.

Of the 150 patients selected for formal chart review,nine had atrial fibrillation or flutter due to underlyingacute medical issues as listed under Methods or werehemodynamically unstable. Excluding these patients left141 for analysis. The nine patients were also removedfrom 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 ofpatients had prior atrial fibrillation or flutter with a suc-cessful cardioversion. Patients were generally healthy,with the majority having a CHADS2 score of zero, andonly 2% (3 ⁄ 124) having a score of >1. The self-reportedduration of atrial fibrillation or flutter in the study sam-ple 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 anti-coagulation had an attempted ED cardioversion. Table 2shows that few patients had rate control or chemicalcardioversion attempted prior to electrical cardiover-sion. All patients received procedural sedation prior toelectrical cardioversion. Propofol was used as the soleagent nearly half the time and was accompanied byeither a sedative or an analgesic in the majority of theremaining cases. Most cardioversions were accom-plished by a single DC shock. The only adverse eventsinvolved six patients who experienced transient respira-tory depression as a result of procedural sedation; noneresulted in an unplanned admission. Index visit compli-cations relating to procedural sedation or electrical car-dioversion can be reviewed in Table 2. Only five (3.5%)attempted electrical cardioversions failed, but there wasno correlation with age or comorbidity. No variables orcombinations could be elucidated to formulate a predic-tor rule to determine success or failure. Table 3 summa-rizes patient resource use and disposition. Cardiologyconsultation was arranged for 22 of 141 cases (16%)and was typically related to unsuccessful DCCV,although only five patients were admitted. Four patientshad continued uncontrolled atrial fibrillation or flutterthat was refractory to both electrical and chemical car-dioversion and one had severe hypertrophic obstructivecardiomyopathy.

Table 4 summarizes 30-day outcomes for all 400patients. Only 22 patients (5.5%) made a visit to a regio-

Figure 1. Study flow (*patients excluded from review alsoexcluded from long-term data linkage). DCCV = direct-currentcardioversion.

410 Scheuermeyer et al. • 30-DAY OUTCOMES OF DC CARDIOVERSION FOR ATRIAL FIBRILLATION OR FLUTTER

nal ED that was potentially related to the index visit,with six patients (1.5%) requiring repeat DCCV (onepatient required two cardioversions). Two patients(0.5%) were admitted. No patient had a documentedstroke, thromboembolic event, or died in the 30 daysfollowing the index visit. Even at the lower confidenceboundary, the rate of stroke, thromboembolic event, ordeath for the overall group is less than 1%.

DISCUSSION

In this study of 400 ED patients who underwent DCCVfor atrial fibrillation or flutter, none suffered significant

complications, stroke, new thromboembolic events, ordeath within 30 days. In contrast with other studies,12

almost all of our patients received primary electricalcardioversion; few had preceding rate control or antiar-rhythmic drug therapy. Six patients required assistedventilations with a bag-valve mask during the sedationprocedure, but no other adverse events were seen, nopatients had a change in disposition as a result of theirtreatment, and only a few patients had repeat ED visits,subsequent DCCVs, or hospitalizations within 30 days.

Increasing adoption of rate-control strategies (basedon data from non-ED populations) has reducedthe impetus to restore sinus rhythm, and a growing

Table 1Baseline Characteristics (From Formal Chart Review, n = 141)

Characteristic Atrial Fibrillation Atrial Flutter Combined

Number of patients (%) 124 (87%) 17 (13%) 141Demographics

Age (yr), mean ± SD 57 ± 14 56 ± 13 57 ± 14Sex (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 visitHeart rate, mean ± sd 123 ± 30 119 ± 33 122 ± 30Systolic BP (mm Hg), mean ± SD 126 ± 25 131 ± 30 127 ± 24Diastolic BP (mm Hg), mean ± sd 79 ± 14 84 ± 20 80 ± 15Respiratory rate (breaths ⁄ min), mean ± SD 16 ± 2 18 ± 1 16 ± 2Arrhythmia duration (hr), mean ± sd 8 ± 6 12 ± 6 8 ± 6Arrhythmia duration (hr), range 1–150 2–100 1–150Arrhythmia 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%)

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


awareness of thromboembolic complications after atrialfibrillation conversion has raised concerns about thesafety and appropriateness of ED cardioversion. Bergerand Schweitzer19 published a retrospective analysis of4,621 patients undergoing elective DCCV, of whom 88developed an embolic episode within 7 days and eightadditional patients within 18 days. Of these 96 events,only three occurred in patients with duration of atrialfibrillation less than 48 hours. Weigner and cowork-ers20 described only three embolic events in a cohort of375 patients admitted to a cardiology unit for atrialfibrillation. In this group, 250 patients (66.7%) also had

spontaneous conversion to sinus rhythm. The only pre-dictor of increased probability of spontaneous cardio-version was atrial fibrillation duration of <24 hours;21

similar results were reported by Tejan-Sie and col-leagues.22 These findings have left uncertainty as to thetime frame when ED cardioversion is appropriate, theoptimal technique of cardioversion, and the best riskstratification tool to apply in determining treatmentapproach

It is critical to point out that the above reports20–22

and the Atrial Fibrillation Follow-up Investigation ofRhythm Management (AFFIRM) trial23 describe patients

Table 2Procedural Sedation and Cardioversion Outcomes (From Chart Review, n = 141)

Parameter Atrial Fibrillation Atrial Flutter Combined

Number 124 17 141Chemical cardioversion attempted prior toDCCV at index ED visit

11 (8.9) 0 (0) 11 (7.8)

Amiodarone 4 (3.2) 0 (0) 4 (2.8)Sotalol 1 (0.8) 0 (0) 1 (0.7)Procainamide 6 (4.8) 0 (0) 6 (4.3)Prior rate control attempted in ED 6 (4.8) 0 (0) 6 (4.3)Beta-blocker 3 (2.4) 0 (0) 3 (2.1)Calcium channel blocker 3 (2.4) 0 (0) 3 (2.1)Sedation regime

Propofol 58 (47) 10 (59) 68 (48)Propofol + ketamine 15 (12) 2 (12) 17 (12)Propofol + fentanyl 33 (27) 3 (18) 36 (27)Propofol + midazolam 8 (6) 0 (0) 8 (5)Fentanyl + midazolam 10 (8) 2 (12) 12 (8)Received periprocedural anticoagulation 8 (6.5) 0 (0) 8 (5.7)Enoxaparin 3 (2.4) 0 (0) 3 (2.1)Dalteparin 5 (4.0) 0 (0) 5 (0.35)

Number of shocks to facilitate conversion1 89 (72) 12 (71) 101 (72)2 19 (15) 3 (18) 22 (16)3 12 (10) 1 (12) 13 (9)>3 (defined as failure) 4 (3) 1 (12) 5 (3)

Successful cardioversion energy (J)<50 2 (2) 1 (6) 3 (2)50 32 (26) 4 (24) 36 (26)100 48 (39) 8 (47) 56 (40)150 20 (16) 0 (0) 20 (14)200 22 (17) 4 (24) 26 (18)

Final HR (beats ⁄ min) 72 ± 18 70 ± 20 72 ± 18Final sBP (mm Hg) 115 ± 15 102 ± 19 113 ± 15Final dBP (mm Hg) 70 ± 10 59 ± 15 69 ± 9Failed 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)

Nausea 0 (0, 0–2.4) 0 (0, 0–17.7) 0 (0, 0–2.1)Emesis 0 (0, 0–2.4) 0 (0, 0–17.7) 0 (0, 0–2.1)Allergic reaction 0 (0, 0–2.4) 0 (0, 0–17.7) 0 (0, 0–2.1)Hypotension 0 (0, 0–2.4) 0 (0, 0–17.7) 0 (0, 0–2.1)Bradycardia 0 (0, 0–2.4) 0 (0, 0–17.7) 0 (0, 0–2.1)Respiratory compromise 5 (4.0, 0.6–7.5) 1 (5.9, 0–17.6) 6 (4.3, 0.9–7.6)Unplanned admission 0 (0, 0–2.4) 0 (0, 0–17.7) 0 (0, 0–2.1)Stroke 0 (0, 0–2.4) 0 (0, 0–17.7) 0 (0, 0–2.1)Thromboembolic event 0 (0, 0–2.4) 0 (0, 0–17.7) 0 (0, 0–2.1)Death 0 (0, 0–2.4) 0 (0, 0–17.7) 0 (0, 0–2.1)Total complications 5 (4.0, 0.6–7.5) 1 (5.9, 0–17.6) 6 (4.3, 0.9–7.6)

Values are %, mean (±SD), or n (%, 95% CI)DCCV = direct current cardioversion; dBP = diastolic blood pressure; HR = heart rate; sBP = systolic blood pressure.*Complications at index visit were defined as emesis, nausea requiring antiemetic medications allergic reaction requiring medi-cations, respiratory distress requiring intervention (bag-valve mask, oral airway, noninvasive positive pressure ventilation, intu-bation), hypotension requiring intravenous fluids or vasoactive agents within 1 hour of initiation of procedural sedation, newbradycardia requiring pharmacologic intervention or pacing within 1 hour of initiation of procedural sedation, unplanned admis-sion, death, or confirmed thromboembolic event.


who are substantially different from those studied inthis and other ED-based cardioversion cohorts.Because rates of stroke, thromboembolism, and deathare more closely related to underlying patient factorsthan to treatment approach, it may be inappropriate toextrapolate findings from non-ED settings to the man-agement of relatively young healthy ED patients withatrial fibrillation or flutter. Our data support the find-ings of Decker and coworkers,15 who reported nostrokes, thromboembolic events, or deaths at 6-monthfollow-up in 153 ED patients with atrial fibrillation, one-third of whom underwent DCCV. Two other ED-basedstudies found similar low adverse event rates.12,13 Inour 8-year review of DCCV for ED patients with atrialfibrillation or flutter, there were few short-term adverseevents and no 30-day complications related to the pro-cedure. The age, sex, and risk profile of our patients isconsistent with prior ED-based studies,14–17 althoughthe CHADS2 score has not been reported previously.

Our patients all had self-reported onset of atrial fibril-lation or flutter within 48 hours or else were appropri-ately anticoagulated. Many had previous successfulDCCVs. No patients had formal transthoracic or trans-esophageal echocardiography to assess ventricularfunction or confirm or refute atrial clot burden, anapproach considered optimal for atrial fibrillation,24 butdebatable for atrial flutter.25 A small minority weregiven single-dose periprocedural anticoagulation, butthe utility of this practice is unclear. Thromboembolicevents from conversion to normal sinus rhythm areknown to occur up to 18 days after cardioversion,19

related to stunning of the left atrial appendage,26,27 so itis improbable that a single dose of anticoagulant wouldbe protective. Of the 141 patients formally reviewed inour setting, only four were discharged with a docu-mented new prescription for aspirin and three forcoumadin, perhaps reflecting the low CHADS2 score.

The ED length of stay in this study ranged from 1.4to 22.8 hours, with a median time of 4 hours. Longerstays were typically a result of prior attempts at ratecontrol or chemical cardioversion, cardiology consulta-tion, and nighttime presentation (with single EP cover-age in a busy department at night, the ability tocardiovert is often limited). However, this is less timethan the 11.8 hours reported by Koenig et al.,28 the10.1 hours reported by Decker et al.,14 and the 9 hoursreported by Domanovits et al.29 and comparable to the5 hours reported by Michael et al..12 Other investiga-tors have found that a greater percentage of patientsundergo a trial of chemical rhythm control prior toelectrical cardioversion, but most patients in our studywere only managed with DCCV. The speed and appar-ent safety of our primary electrical approach may resultin lower system utilization for patients requiring con-version.

Although the study is retrospective, the chart reviewconfirms that the patients in our study are similar tothose in the ED literature, increasing the potential gen-eralizability of our findings. Rigorous use of data link-ages between our ED database, the regional EDdatabase, and the provincial vital statistics registryallows for long-term tracking of individuals.

Table 3Resource Use at Index Visit (From Chart Review, n = 141)


Number 124 17 141Cardiology 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 ± 2:15Median ED LOS (hr) 3:55 4:30 3:59Mean (±SD) hospital LOSfor admitted patients (days)

2 ± 2 (4 patientsadmitted)

2 (1 patientadmitted)

2 ± 2 (5 patientsadmitted)

LOS = length of stay.

Table 430-day Outcomes (From Linked Regional and Provincial Databases, n = 400)


Number 342 58 400New 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)

*For visit related to atrial fibrillation or flutter.�One patient had two DCCVs for atrial fibrillation within 30 days.DCCV = direct current cardioversion.


LIMITATIONS

Results of this two-center study may not be generaliz-able to diverse settings. Retrospective studies are sub-ject to missing or improperly coded data, although ourdata audit indicated an error rate in the range of 1%,which should minimally affect results. Based on ourchart review, 9 of 150 (6%) patients receiving cardiover-sion were excluded due to instability or underlyingacute medical illness. Patients not undergoing a chartreview (n = 259) would not have been scrutinized in thismanner. It is probable that unstable patients were inad-vertently included in the 30-day follow-up and wouldhave been excluded had their charts been scrutinized,but such patients would be expected to have worse out-comes. The duration of atrial fibrillation or flutter isbased on self-reported patient histories and may not beaccurate. DCCV took place in a group of young patientswith low CHADS2 scores who had been in atrial fibril-lation or flutter for only a short time, which may pre-vent extrapolation to sicker patients. While our regionalED database identified all repeat visits to nearby hospi-tals, patients visiting a nonregional ED within 30 dayscould be overlooked. Patients lacking a provincialhealth number or using more than one might escapelinkage as well.

CONCLUSIONS

This study expands on previous work by enlarging thetime frame during which the rates of post-ED cardiover-sion strokes, thromboembolic events, and mortality arevery low. This suggests that ED selection of cardiover-sion candidates is appropriate and that the procedure issafe up to 30 days for both atrial fibrillation and flutter.A larger multicenter trial with a longer follow-up periodwould provide greater confidence in the conclusion thatthere is minimal risk associated with this approach.

References

1. Feinberg WM, Blackshear JL, Laupacis A, KronmalR, Hart RG. Prevalence, age distribution, and gen-der of patients with atrial fibrillation: analysis andimplications. Arch Intern Med. 1995; 155:469–73.

2. Valderrama AL, Dunbar SB, Mensah GA. Atrialfibrillation: public health implications. Am J PrevMed. 2005; 29:75–80.

3. Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation:a major contributor to stroke in the elderly: the Fra-mingham Study. Arch Intern Med. 1987; 147:1561–4.

4. Lakshminarayan K, Anderson DC, Herzog CA,Quereshi AI. Clinical epidemiology of atrial fibrilla-tion and related cerebrovascular events in the Uni-ted States. Neurologist. 2008; 14:143–50.

5. Béjot Y, Ben Salem D, Osseby GV, et al. Epidemiol-ogy of ischemic stroke from atrial fibrillation in Di-jon, France, from 1985 to 2006. Neurology. 2009;72:346–53.

6. Albers GW, Atwood JE, Hirsh J, Sherman DG,Hughes RA, Connolly SJ. Stroke prevention in non-valvular atrial fibrillation. Ann Intern Med. 1991;115:727–36.

7. Hohnloser SH, Pajitnev D, Pogue J, et al., ACTIVEW Investigators. Incidence of stroke in paroxysmalversus sustained atrial fibrillation in patients takingoral anticoagulation or combined antiplatelet ther-apy: an ACTIVE W Substudy. J Am Coll Cardiol.2007; 50:2156–61.

8. Nieuwlaat R, Dinh T, Olsson SB, et al., for EuroHeart Survey Investigators. Should we abandon thecommon practice of withholding oral anticoagula-tion in paroxysmal atrial fibrillation? Eur Heart J.2008; 29:915–22.

9. Fuster V, Rydén LE, Cannom DS, et al. ACC ⁄AHA ⁄ ESC. 2006 guidelines for the management ofpatients with atrial fibrillation: a report of theAmerican College of Cardiology ⁄ American HeartAssociation Task Force on Practice Guidelines andthe European Society of Cardiology Committee forPractice Guidelines (Writing Committee to Revisethe 2001 Guidelines for the Management of PatientsWith Atrial Fibrillation). J Am Coll Cardiol. 2006;48:e149–246.

10. Heilbron B, Klein GJ, Talajic M, Guerra PG. Man-agement of atrial fibrillation in the emergencydepartment and following acute myocardial infarc-tion. Can J Cardiol. 2005; 21(Suppl B):61B–66B.

11. Flaker GC, Belew K, Beckman K, et al., for theAFFIRM investigators. Asymptomatic atrial fibrilla-tion: demographic features and prognostic informa-tion for the Atrial Fibrillation Follow-upInvestigation of Rhythm Management (AFFIRM)study. Am Heart J. 2005; 149:657–63.

12. Michael JA, Stiell IG, Agarwal S, Mandavia DP.Cardioversion of paroxysmal atrial fibrillation in theemergency department. Ann Emerg Med. 1999;33:678–88.

13. Burton JH, Vinson DR, Drummond K, Strout TD,Thode HC, McInturff JJ. Electrical cardioversion ofemergency department patients with atrial fibrilla-tion. Ann Emerg Med. 2004; 44:20–30.

14. Stiell IG, Clement CM, Symington C, Perry JJ, Vai-llancourt C, Wells GA. Emergency department useof intravenous procainamide for patients with acuteatrial fibrillation or flutter. Acad Emerg Med. 2007;14:1158–64.

15. Decker WW, Smars PA, Vaidyanathan L, et al. Aprospective, randomized trial of an emergencydepartment observation unit for acute onset atrialfibrillation. Ann Emerg Med. 2008; 52:322–8.

16. Stead LG, Vaidyanathan L. Rhythm control withelectrocardioversion for atrial fibrillation and flut-ter. Ann Emerg Med. 2009; 54:745–7.

17. Mead GE, Elder AT, Flapan AD, Kelman A. Elec-trical cardioversion for atrial fibrillation and flut-ter. Cochrane Database Syst Rev 2005; 20:CD002903.

18. Gage BF, Waterman AD, Shannon W, Boechler M,Rich MW, Radford MJ. Validation of clinical classi-fication schemes for predicting stroke: results fromthe National Registry of Atrial Fibrillation. JAMA.2001; 285:2864–70.

19. Berger M, Schweitzer P. Timing of thromboembo-lic events after electrical cardioversion of atrial


fibrillation or flutter. Am J Cardiol. 1998; 82:1545–47.

20. Weigner MJ, Caulfield TA, Danias PG, SilvermanDI, Manning WJ. Risk for clinical thromboembolismassociated with conversion to sinus rhythm inpatients with atrial fibrillation lasting less than48 hours. Ann Intern Med. 1997; 126:615–20.

21. Danias PG, Caulfield TA, Weigner MJ, SilvermanDI, Manning WJ. Likelihood of spontaneous con-version of atrial fibrillation to sinus rhythm. J AmColl Cardiol. 1998; 31:588–92.

22. Tejan-Sie SA, Murray RD, Black IW, et al. Sponta-neous conversion of patients with atrial fibrillationscheduled for electrical cardioversion: an ACUTEtrial ancillary study. J Am Coll Cardiol. 2003;42:1638–43.

23. Wyse DG, Waldo AL, DiMarco JP. A comparison ofrate control and rhythm control in patients withatrial fibrillation. N Engl J Med. 2002; 347:23. 1825–33.

24. Klein AL, Grimm RA, Murray RD. Use of trans-esophageal echocardiography to guide cardiover-sion in patients with atrial fibrillation. N Engl JMed. 2001; 344:1411–20.

25. Mehta D, Baruch L. Thromboembolism followingcardioversion of ‘‘common’’ atrial flutter. Risk fac-tors and limitations of transesophageal echocardi-ography. Chest. 1996; 110:1001–3.

26. Grimm RA, Leung DY, Black IW, Stewart WJ, Tho-mas JD, Klein AL. Left atrial appendage ‘‘stunning’’after spontaneous conversion of atrial fibrillationdemonstrated by transesophageal Doppler echocar-diography. Am Heart J. 1995; 30:174–6.

27. Harjai K, Mobarek S, Abi-Samra F, et al. Mechani-cal dysfunction of the left atrium and the left atrialappendage following cardioversion of atrial fibrilla-tion and its relation to total electrical energy usedfor cardioversion. Am J Cardiol. 1998; 81:1125–9.

28. Koenig BO, Ross MA, Jackson RE. An emergencydepartment observation unit protocol for acuteonset atrial fibrillation is feasible. Ann Emerg Med.2002; 39:374–81.

29. Domanovits H, Schillinger M, Thoennissen J, et al.Termination of recent-onset atrial fibrillation ⁄ flutterin the emergency department: a sequentialapproach with intravenous ibutilide and externalelectrical cardioversion. Resuscitation. 2000; 45:3.181–7.


Documents

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