JACC Vol. 19 . No . 4Starch 15 . 1992:951-5

Role of Prophylactic Anticoagulation for Direct Current CardioversionIn Patients With Atria[ Fibrillation or Atrial FlutterANITA ZEILER ARNOLD, DO. MATTHEW J. MILK . MD, ROBERT P. MAZUREK, MD,FLOYD D . LOOP, MD, FACC, RICHARD G . TROHMAN, MD, FACCCleveland, Ohm

The need for prophylactic anticoagulation f prevent embolismbefore direct current cardloverstan Is performed For alrial fibril-lation or alrial flutter is controversial . To examine this issuefurther, a retrocpw"ive review was undertaken to assess theincidence of embolic complications after cardioversion, The re'view involved 454 elective direct current cardioversion per-formed for atrial fibrillation or atria) flutter over a 7 year period .

The incidence rate of embolle mmplicatons was 1 .3250 (sixpatients); the eomplica!ions ranged from minor visual dislurvhelices to a fatal cerehrnvasculat event . All six patients had alrialfibrillation, and none had been on anticoagulant therapy (p =0.026), The en .oOvn ar nrr,nl fibrillation was < I week in five ofthe six patterns who had embolic complications .

Baseline characteristics of patients with a posteardioversion

Atrial fibrillation increases the risk for ischemic stroke by afactor of 5 over a lifetime. It has been estimated that onethird of ali patients with alrial fibrillation will eventually havea cerebrovascular accident ID, Atrial fibrillation has aiczbeen implicated in up to 37% of asymptomatic cerebralinfarcts diagnosed by computed tomographic scans (2-4).Because of the incidence of stroke, some investigators (5-7)recommend that certain subgroups of patients with atnolfibrillation receive long-term anticoagulant therapy. Al-though the risk of embolism may be decreased by anticoag-ulant therapy, there is a lack of consensus regarding the needfor such therapy before direct current cardioversion isperformed in patients with atriat fibrillation or atria) flutter(8-10). To further assess the need for anticoagulation in thesetting of elective cardioversion we retrospectively reviewedthe results of 628 consecutive cardioversions.

MethodsStudy patients. Adult patients undergoing elective direct

current cardioversion at our institution for atria) fibrillation

Form the tlepanmenls of Cardiology and Cardiocecculot Surgeo.'rheCleveland Clinic Foundation, Cleveland, Ohio .

Man cripl received May 31 . 1991: revised manuscript receivotl Seplem-her 11 . 1991 . accepted Seplemher 24, 1991 .

Address for reodnts : Richard G. Toohman . MD, Department of Cardiol .ogy 17.15 . The Cleveland Clinic Foundation . Cleveland. Ohio 44195.

01992 by the American College of Cardiology

851

embolic event are compared with these of patkats who did rothave an embafie event. There was no difference In the prevalence

of hypertension, diabetes mi llitus or prior stroke between the twogroups, and there was no difference in the number of patients whowere postoperative or had poor kit ventricular functka . Leftatria) sire was similar between the two groups . No Pattern In theembolic group had valvular disease.

No patient with initial Butter badan embolle event regardless ofanticoagulant status; therefore, anticoagulation is act recum .mended fur patients with acrid flutter undergoing cardinversion .Prophylactic ladcmguktion is pivotal in patients undergoingelective direct emrem cardioversion for aeriul fibrillation, eventhose with acid fibrillation of <I week's duration .

(J Am Calf Cordial 1992;191851 5)

or atrial flutter (of at least 48 h duration) between January I,1983 and June 30, 1990 were included. Our analysis excludedunsuccessful cardioversions, because tl c?"rerslun m nor-

mal sinus rhythm is the presumed risk factor for embolic

events, We excluded 90 cardioversions on this basis, alongwith 51 in which atria[ fibrillation or atria) flutter was presentfor <48 h and 33 cardioversion with indeterminate antico-agulant stains . There were 628 procedures performed in thestudy per'od add 454 (in 428 patients) were subsequentlyreviewed. In I77 cardioversions (40.7%) ; the patient wasinking anticoagulant medication (Coumadin in 98 .3%) . Thisreport summaries the characteristics of the patients under-going direct current Cardioversion and the embolic phenom-ena that occurred as a direct consequence of the procedure .

Cardioversion. All procedures were carried out in a fullyequipped cardiac catheterization or electrophysiology labo-ratory with emergency equipment available . Digitalis prep •orations were withheld the morning of the procedure . andThe patient was fasting. Intravenous sedation (usually a bolusirijection of Brevital 0.5 mg/kg body weight) was given, anda synchronized, direct current shock was administered torestore sinus rhythm . Pre- and postcardioversion rhythmstrips were available to 'crlfy the presence of atria) fihritla-tion or atria) flutter before the procedure and subsequentsinus rhythm after cardioversion. Cardioversioos performedin the patients room or the intensive care unit for thepurpose of terminating hemodynamicalty unstable atria] fi-brillation or atrial flutter were excluded .

0735.1097192155.110

852

.ARNOLD ET AL.ANTICOAGULATION FOR CARDIOVERSION

Data analysis. The medical record was reviewed fordemographic characteristics and underlying diseases (hyper-tension, diabetes mellitus or prior cercbrovascular accident ) .In addition, echocardiographically documented left atrialsize, as well as left ventricular function (documented bycathelerization or echocardiographic analysis) . was noted .The duration of the specific rhythm disturbance was ana-lyzed . If the patient had recently undergone open heartsurgery. the nature of the procedure was noted . and forvalvular surgery, the position and type of valve were iden-tified . Anticoagulant medication was recorded and Inborn-Tory variables were checked to confirm therapeutic activity .A prothrombin time >15 s (L5 times the lower limit of the10-s control value in out laboratory) or an activated partialthromboplastin time of 50 s (twice the lower control limit)was considered therapeutic.

F.11-lip . All patients were followed up for evidence ofembolism for at least 2 weeks, and postoperative patientswere followed up until their 6-week postoperative visit .Embolic events were coded as having occurred in thecerebral, pulmonary or peripheral circulation and were doc-umented by computed tomogrtphy . ventilation-perfusionscanning or angiography .

Statistics. All values are given as the mean valuesI SD . Group comparisons of categoric variables were madeby using the chi-square or Fisher exact test, as appropriate .Continuous variables were compared by using the Student ttest. A p value <11.05 was considered significant .

Baseline characteristics, The predominant rhythm distur-bance requiring cardioversion was atrial fibrillation (atrialfibrillation in 72 .8%. atrial flutter in 27 .2%). The majority168.671 of the patients with atrial fibrillation were male, andthe average age was 64 ± 9.7 years (range 21 to 88) .Concomitant hypertension was present in 36 .6% of patients,diabetes mellitus in 14 .7% .a history of prior cercbrova5cularaccident in 10 .7%a . There were 206 patients (67 .3%) withpostoperative atrial fibrillation at the time of cardioversion .These baseline characteristics are summarized further inFigure I . There were no statistical differences between thecharacteristics of patients who were or were not receivingant)congulant therapy,

In the patients with direct current cardioversion for atrialfibrifation, the duration of arrhyduoia ranged from 48 h to 7years. The duration of atrial fibrillation was <8 days in43-'7,-'/ of cardioversions . 8 to 30 days in 29S%, >30 days but<I year in 25!9%, and > 1 year in 0.9% of cardioversions;Fig. 2) . Prosthetic valves had been implanted in 85 patients'27 .7%) ; the prosthesis was in the mitral valve position in 44patients, the aortic valve position in 32 and in both positionsin 9 .

E<'6liographir data were available for 63 .3% ofpatients undergoing direct current cardioversion for atrialfibrillation . The average left atrial size was 49 ± 8 .2 mm(range 27 to 72). The left atrial size did not correlate with theduration of arrhythmia ; in patients with atria] fibrillation for<30 days the average left atrial size was 49 `- 8.2 mm and in

∎ ns

El ao1 p numnrunAe ® 110-11 ruxuL

JACC Vet . 19. Nn. 4March 15. 1992 : 8 5 1 -5

NYrenTnls1ON

eye

p ac,Naruanr ® No .c,NAroanc

Figure 1, Baseline characteristics of the patients with atrial fibrilla-tion . AC = with anticoagulant therapy ; CVA - cerebrovascularaccident : NO AC - without anticoagulant therapy ; natural -patients who did not have atrial fibrillation related to an operativeintervention ; POSTOP = postoperative .

those with atrial fibrillation for >30 days the average sizewas 50 ± 8 mm .

Assrcsntettt ofleft venrrirnlarfmction was available for96.1% of patients ; 70 .1% were characterized as havingnormal function or mild dysfunction . Of the remainingpatients . 19.7% had moderate and 10.2% had severe leftventricular dysfunction .

ResultsAtria) fibrillation . There were 454 cardioversions per-

formed to treat atria] fibrillation or atria[ flutter. Embolicvents occurred in 6 of 179 cardioversions for atrial fibrilla-

tion in the absence of anticoagulant therapy . No embolicevents occurred in the 153 cardioversions for atrial fibrilla-lion in patients who were taking an anticoagulant agent . Thisdifference was statistically significant (p = 0 .026). Anticoag-ulation was used in 47% of patients without embolism and in

Figure 2 . Duration of atrial fibrillation. Abbreviations as in Fig-ure I .

none of the six patients with embolism (p = Rfi2)- Theconsequences of Ihese embolic events ranged from minorvisual disturbances to a fatal cerebrovascular embolus . Inthe group of six patients with embolism, four had hyperten .,ion and one patient had a history of prior cerebrovascularaccident . No significant bleeding episodes related to antico-agulatinn were noted in the study group . The averageduration ofalrial fibrillation in the six patients with embolismwas 6 ± 4 days (range 3 to 19) and two patients (33'7cl werepostoperative. The average left atrial size was 44 . 5 mm ;and left ventricular function was impaired to a moderate orsevere degree in four (67%) of the six (Table 1) .

Cardioversion with aruiroagulaar therapy. There were153 cardioversion performed on patients with laboratoryconfirmation of therapeutic anticoagulation . Postoperativeatrial fibrillation was present in 11)1 cases ; the remaining 5;cardioversion were performed for naturally occurring atria)fibrillation (that is, atrial fibrillation that did not occur in thepostoperative period) . Nineteen percent of the cardlover-sions were performed in patients who had been receivinganticoagulant therapy for >2 weeks before cardioversion (at[except one patient were in the cohort with naturally occur-ring atrial fibrillation) . The duration of anticoagulation didnot appear to affect the incidence of embolic events.

Cardioversion without anricoagulanr therapy. Therewere 179 cardioversions perforated in patients for atria[fibrillation who were not receiving anticoagulant therapy :115 were postoperative and 64 were in the group withnaturally occurring atrial fibrillation. Two rises of embalismoccurred in the postoperative group without anticoagulanttherapy (incidence rate 1 .790) and four occurred in thenatural atrial fibrillation group without anticoagulant therapy(incidence rate 6.2%% ; p = NS). These results are summa-rized in the flow diagram in Figure 3 .

Atria] flutter. We also analyzed data from cardioversionin 122 patients with atrial flutter, Similar m others (8,11), wefound no embolic events in this group, regardless of antico-agulant status (2h% with and 74% without anticoagulantTherapy)_ The number of patients with atrial flutter in ourstudy may be too small to fully detect a low frequency eventsuch as postcardioversion embolism . Our data support thecurrent practice of not giving anticoagulant therapy to pa-tients with atrial flutter who undergo cardioversion .

AF - slant SbIiILianI F - femste. LV - Ld :annicutur . M = r, M.?o.up = p,nrnrrror ,e:'I - rouses.

DiscussionSynchronized electrical cardioversion for the termination

of both atrial and ventricular tachyarrhythmias in a largeseries of patients was first reported by Lown (12) in 1%7 .The most devastating complication of this procedure issystemic embolization, which has been estimated to occur inup to 792 of patients (I3) . The atria) arrhythmia most likely tobe associated with postcardioversion embolism is airiatfibrillation (5,11) . a finding further supported by our results.No patient in our series who underwent cardioversion foratrial flutter had an embolic event .

Mechanism ofembelration . The ntechanismofemboliza-tion in atrial fibrillation appears to be complex . Researchers(6) have documented increased cerebral blood flow afterconversion from atrial fibrillation to normal sinus rhythm,which predisposes to cerebral embolism in the distribution ofthe middle cerebral artery . An intrinsic but unidentifiedproperty of atrial fibrillation may lead to ahnormalitics inclu)ting function, which may contribute to the incidence ofemholieation after cardioversion (14). Atrial fibrillation isassociated with increased levels of airial natrturetic peptideand an increase in hematocrit (15) . These changes may

flint, 1 . lInus diagram !hos5ittg be analysis of 521 eardierensiutsand the rat ents with subsequent embolism. AC = with anticoagu-lant therapy : AF = atrial fibrillation;AFi. = arrant flutter: DCC =direct-current cardioversion : NAT = natural alriat fihrillatiost Iseetext) ; noAC = without amicoagulane therapy; POSTOP = postop-erative .

Cardioversion flow diagram

tvAF

Araaa

AcJ rsa `

ftru _Pastor NAT POMP MT, 5

0

a FMFGLI

Ia0

a~

~

t4 EMBOLI

a,saa • ones -'

Piso.

ApeIY'Y

GcndcrAnti-EAhn1

Tharar, R1Yrhmn-4 ititAF IJa..I

IVFunction Pawl F"-

Tiatw--r., W.,,

S91F No AF Normal Yes Pedpfrcrnlrm5olhr, I10117 No AF mud-- Y., OCmur,I .mtr,Ihm 75590 Na AF S- if Na Fulmowtyr .b,tism 017171 No AF 4 Sto tretr Na Qi, , tr ,ir"k r; 3

S 67im N o AP t Store Na Ceoemlaabalom tTrM v o SF S hlaa No Crrrbrelembotism -

JACC V& 19, No. 4Marry I5 . 1992-.a'11-5

o ET AL .

853:t'OICO4GULATION FOR CAROIOVERSION

Table I • Charaelerislics of Six Patients With an Embo'ic Evcat

854

ARNOLD ET AL .ANTICOAGULATION FOR CARDIOVERSION

account for the increase in stroke with chronic atria) fibril-lation.

One of the suggested mechanisms of embolism is theformation during atrial fibrillation of a clot that is expelledwith the sudden resumption of mechanical atrial systole.Obarski et al. (16) detected a spontaneous echo contrast,consistent with blood stasis, during transesophageal echo-cardiography in a patient with atrial fibrillation, enlarged leftatrium and a normal mitral valve . The study was repeatedafter the patient was in sinus rhythm . The echo contrast(residual stasis) was localized to the left atrial appendage,whereas the remainder of the left atrium was clear . Obarskiet al. (16) theorized that this residual blood stasis mayaccount forembolic events several days after sinus rhythm isrestored. However, this residual stasis does not account forthe lower incidence of embolism with paroxysms of atrialfibrillation compared with the incidence with chronic atrialfibrillation . Short paroxysms may not allow the hormonalchanges that result in clot formation to occur .

Incidence of emholizatimt . Although it has been suggestedthat patients with atrial fibrillation should receive anticoag-uiant therapy before undergoing direct current cardioversion(11,12), there is a lack of consensus on such treatment. Dataarc sparse on the incidence of embolism after cardioversionfor specific arrhythmias. In a 1967 review, Lown (12) exam-ined 456 cardioversion attempts and reported a 1 .2% inci-dence rate of embolization . No patient received anticoagu-lant therapy at the time of the procedure . In 1969,Bjerkelund and Orning (9) reported the effect of anticoagu-lation on the rate of stroke after cardioversion in a nonran-domized population of 437 patients . They reported thatcardioversion without anticoagulation resulted in a 5 .3%incidence rate of embolism compared with a rate of 0 .8%with anticoagulation (p = 0 .016) . Although the study wasnonrandomized, the results are important as the patientsreceiving anticoagulant therapy were at higher risk forembolization because of their underlying rheumatic heartdisease, congestive heart failure and history of prior embolicevents . Unfortunately, because the patients were receivinglong-term anticoagulant therapy, the effect of prophylacticanticoagulation before cardioversion cannot be addressed .Also, subgroups were not analyzed by rhythm (atria] fibril-lation or atrial flutter), which is reported to be an indepen-dent risk factor for embolic phenomena after cardioversion(11,17,18) .

More recently, Mancini and Weinberg (19) reviewed thedata from the University of Michigan over a 10-year periodand analyzed embolic events after cardioversion for atrialfibrillation . They found no events in the group with antico-agulant therapy, whereas 7% of the group without suchtherapy had embolic complications . Although the numbersare small, the results support well managed, short-termanticoagulant treatment in high risk patients (defined by theauthors as patients >55 years old with coronary arterydisease, cardiomyopathy or hypertension and with a dura-tion of atrial fibrillation of >I year) .

JACC Vol . 19, No. 4March 15, 1992:951-5

Although our 1 .32% incidence rate of emboli : eventsafter cardioversion for atrial fibrillation is similar to that ofother reports (20-22), there may be a subset of patients whoare at higher risk for embolic events : patients with naturallyoccurring atrial fibrillation who do nut receive anticoagulanttherapy. These patients had an incidence of embolism thatwas nearly four times that of the postoperative group with-out anticoagulant therapy . Although this difference in inci-dence did not achieve statistical significance, further inves-tigation in a larger series of patients seems warranted .

Duration of anticoagulation . Atria) fibrillation of longduration is a well documented risk factor for stroke (1). Ithas been recommended (8,11) that patients with atrial fibril-lation of short duration (I 1 week) do not need anticoagula-tion before undergoing cardioversion . In our patients whohad an embolic event, the duration of axial fibrillationranged from 3 to 19 days (average 6 . 4), with five of the sixpatients having atrial fibrillation for e I week . Direct currentcardioversion appeared to be hazardous in our patients withatria] fibrillation of short duration who were not receivinganticoagulant therapy. Our results suggest that low doseanticoagulation with a prothrombin time >15 s was protec-tive against embolic events .

Most cases of embolism have been reported to occurwithin hours to a few days of cardioversion (9,10) . Theoret-ically, however, embolism can occur several weeks afterelectrical conversion when mechanical atrial activity re-sumes (23) . While this theoretical risk of embolism remainsfor up to 3 weeks (24,25), all of the embolic events in ourpatients occurred within 7 days after c ardioversion. it istherefore unclear how long anticoagulant therapy should becontinued after successful conversion to sinus rhythm. Be-cause of the retrospective nature of our data, the length oftime that anticoagulation was continued after cardioversionis not well documented . We cannot conclude that anticoag-ulation at the moment of cardioversion, without its continu-atiun, will prevent a delayed embolism . Manning et al . (23)followed the reccvery of atria] mechanical function in 21patients over 3 months with serial Doppler echocardio-graphic evaluations . They found that peak A wave velocityand percent of atrial contribution to left ventricular filling didnot return to normal until 3 weeks after cardioversion inpatients who maintained sinus rhythm . Continued anticoag-ulation for several weeks is net likely to result in significantbleeding complications (26) and probably should be main-tained until mechanical atrial systole resumes .

Our data raise a number of important issues . Should apatient with atrial fibrillation of >2 days' duration receiveanticoagulant therapy before elective direct current cardio-version? Is the anticoagulant of choice Coumadin . or isheparin adequate therapy? We believe that further investi-gation with randomized, prospective trials should be under-taken to answer these important clinical concerns .

Limitations of study . The limitations of our study arethose inherent in any retrospective review. There was norandomization of treatment arms for anticoagulation, and we

IACC Vol . 19. No . 4March 15 . 1992:851-5

were unable to identify specific high risk subsets that would

benefit from anticoagulant therapy. However, despite theselimitations. all the errrbulic events came from the groupwithout anticoagulant therapy .

Cenelusions. In our series . atrial Burner was never asst,eiated with an embolic event and we du not recommendTontine anticoagulation in these patients . Our results

strongly support the use of anticoagulant treatment in pa-tients with atrial fibrillation of at least 4g-h duration who areundergoing elective synchronized direct current Cardiovef .siou. .It seems prudent to continue anticoagulation untilmechanical atrial systole resumes. Although others 19.13 210have reported a similar risk of embolization with chemicalcardioversion . ear study was not designed to examine thisrisk. Further research is needec for those undergoing ear-dioversion with antiarfhylhmtc agents, as anticoagulationmay be appropriate based on theoretical considerations .

Refererfees

I . HaheninJL, Han RG, Anial06rilldionand stroke : roso ideas, persiuilgdilemmas. Stroke 19&8tt9:937-41 .

2. Peteeten P. Madsen EB, than B. Pederson F, Bovsen G. Silent cerebralinfarotion in wood hbrbldiere. Stroke 1902 :1x:1098-1116,

3 . Kempster PA. Cereals RP. Gales PC. Asymplematic cerebrol intarctionin patients with ataxic amid fibrillation . Stroke 198&:19:955-7.

4. Ralc5ffe PJ, Wikock GK . Cembeovascutew drone in dememia: theimparlance of at541 fibrillation. Poorad Mod 11MS,lel:bh4 .

5, Wolf PA. Dawber lR, Thomas HE. Kaneoet we. Epdemiobeic assess-ment ofetaonic atrial fibrillation and The risk at suoke : the Fr milghamStudy. Ne9rolusy 19181097!-7 .

6. Petersen P . Thrombeembotic complications ofatrialfitrillatioe and theirpresenlton : a maiew . Am 1 Ckrdiol IsSH .ht :24C_8C.

7 . Kopecks SL .GelnehhJ .

Mb,elatThenatural501005ofI orKwlekd fihrdtali, e. N Ehgl 1

ModMed 19&7311:669-74,

8 . Manein7GB3.Gotdber0erAL.Cavweersioeeofa.idfibrillatia0:coreet.elation of embol¢Wan . anlicoaglelmion . prophylactic pacemaker. andlong term success . Am Hear S 1982;107 :617-21 .

9. Bierkelund CJ. Orninf OM . The efacasy of onticoegWation etrempy iIipreventing embolism related to three current clecaical ..-in . ofmeal fihrilhtion. tun J Cardiol 1969:23 :2[8-Ifi.

AR-13 Er AL.

855Pvrl000GULATIO4 FOR CARD OVERSION

III, W'ein6crgDid- ManciniGBJ .Anlieeofalefmoh,roerdiooersionofztrietlbeillet on. Am 1 Cod or 1989:63 :745-a.

11 . Corn M- Al-&(J, de &Ira k. Halaer t . Aeeithrombonc therapy inavalthnnador .Closr 1909.95~It83-27S.

12 . LnwnH vlnoencat-ion M~,afiacaahplerO0 BrticarII967....f9.405-fi9.

I7 . Stein B . Holcetio 1L. Foster V. Shouts potions with mrial bhrilialinn beanlictmgnlnled prior 16 and chroeoicatiy folin.ing rardioeersiun? In :Cheillin MD . ed. Dtkmmas in Clinical Cardiowry. Philadelphia: FAU-, 1997.231-49.

!4,Knm5eiK,FukunamiM,Obmodtit .Kito take A.KamadaT.HokiN .Ineraseo erlnrardovesduiar dolling 0 parents wuh chronic atria[ft, . I ik n. J Am Coil Cardiol 1990;16:377-80 .

IJ. Peter,en P. Kaurop J . Villletmsen R. Schalen Hl . Alnal miduelkpeptide In atrial 0bnlfallon befo n and aekr eleemcat card00cniastherapy . Ear Hean 1 I9r9 :9:639_41 .

16, Obarski TP. Sateedo EE. Castle LW. Stewan WJ. Spontaneom echocontrast of The left 05,0110 daring proryamal atoal fibrillation . Am Hood

17. P2tersen P . Gadlfledsen J . Eetodo cnnp5catims in poroaysmel elrialfibulfnian . Stroke 1986;17 :622-6,

10 . Nit-, P 7Lnimbcxmbolk aanpfiemions in ..̂Ida1 fibrillation. Sleoke1910:21 :4-It .

19 . Moecini GBI . Wcinheeg DM, Cardiovers:an ofatlial fibeilhhon : a rein-sp050ooanalysis of the safety and 00100 of anlkrggulation. CerdinvneAce & Rep 1998.1

20- idioms JJ . Poor RH, McIntosh HD. Electrical conversion of atrialtlbnllalien: immediate and imgterm resultsaod selection afpatients . Annlnmn Mod 7966 :6° :216-30.

21 . Rjerkel ltnd C. Orning OM . An evah®lion of DC oboth ectmcm of atrialarrhynhmias . Aria Mod Scald 1964:184l4Rl-91.

22. McCadhy C. Vagane PJ. Poorer DW. Prognosis of atrial aarhythm100401 by eheir!eal ooumershock therapy. Br Head 1 196931 :496-510.

23 . Manning Wl . Leeman DE, Golch PJ. Come PC. Poised Doppler mat-1- o: alrid mechaeot fimraiao afer ekcraicaf ram-f atrialfibrille ;inn . I Am Call Cordial 19&9:[3 :617-23 .

24 . O'Neill PG, hi PR . Both R. Rokey R . Return of atrial nlechaeicalfunmionfollo.bgekdrioal oonveraiondanieldynohylhmios. Am Heart1 t990t12P.3Pr9.

25 . Ikram R. 9igon PGF . Moan T. Left atrial rfonron into, electrical--vi- m shma ah}ihn. Br Hod 1 1968 J k80 I .

26 . LeaineMN.Rasl:ohG.HimhI.Hoinval jccomdicalioesofbn8+armanlooagulationl1Gapy. Chest t101449:1M-d311

27 . Yap ASS . Green 01 . Embolic atrokc fo3 :mng eardooerstun of mrialfibrillation to sinus rhythm with mat amiodarone therapy . Pos1grad Mod 11998 .66:410.