Usefulness of anticoagulant therapy in the prevention of embolic complications of atrial fibrillation

The medical records of 254 patients with atrial fibrillation were reviewed to determine the incidence of embolic events in relation to type of cardiovascular disease, duration of atrial fibrillation, and use of anticoagulants. During a total follow-up of 833 patient-years in atrial fibrillation, there were 32 instances of systemic embolism: 21 involved the cerebral circulation and 11 were extracerebral. Thirty of these events occurred during 549 patient-years of follow-up without anticoagulation therapy (5.46 of 100 patient-years), while only two embolic events occurred during 284 patient-years on anticoagulants (0.7 of 100 patient-years). Thus, the incidence of embolism was eight times more frequent during the unanticoagulated period of observation in atrial fibrillation (p < 0.002). The incidence of embolism during follow-up without anticoagulants was the same regardless of the presence or absence of mitral valve disease and regardless of whether atrial fibrillation was chronic or paroxysmal. The rate of serious hemorrhagic complications on anticoagulants was acceptably low (2.11 of 100 patient-years). We conclude that in this study population anticoagulant therapy reduced the risk of embolic complications of atrial fibrillation. The results also indicate that the use of anticoagulants should not be limited to patients with atrial fibrillation due to mitral valve disease. (AM HEART J 1986;112:1039.)

Denis Roy, M.D., Etienne Marchand, M.D., Pierre Gagn6, M.D., Michel Chabot, M.D., and Richard Cartier, M.Sc. Montreal, Quebec, Canada

The prevention of cardiac embolism is a common and important clinical problem for both the inter- nist and cardiologist. It has long been demonstrated that atria1 fibrillation in association with rheumatic heart disease predisposes to systemic embolism.1-6 The degree of risk of embolism in atria1 fibrillation not related to rheumatic heart disease is controver- siaL7-14 Some investigators10-14 contend that patients with nonvalvular atria1 fibrillation are also at increased risk of embolism. Hence, it is unclear if anticoagulants should be administered routinely to patients with atria1 fibrillation of any origin or selectively to those with coexisting mitral valve disease.‘” Furthermore, the current clinical practices with regard to anticoagulation therapy in any group of patients with atria1 fibrillation are based on relatively limited data.‘**“* ‘w* In an attempt to provide further information on this important sub-

From the Ikpartment of Medicine. Montreal Heart institute and the llniversity 01 Montreal Medical School.

Received for publication Sept. lti. 1985: revision received Feb. 10, 1986; zaccrpted March 14, 1986.

Reprint requests: Denis Roy, M.D.. Montreal Heart Institute, 5000 East Manger St.. Montreal. Quebec HIT lC8. Canada.

ject, we retrospectively examined the incidence of embolic events in 254 patients with atria1 fibrillation due to a variety of causes who were followed either with or without anticoagulation therapy.

METHODS

Patients. The records of patients admitted to the Montreal Heart Institute between ,January. 1979, and December, 1983, with a primary or secondary diagnosis of atria1 fibrillation were reviewed for the following informa- tion: age, sex, type of cardiovascular disease, duration of atria1 fibrillation, use of anticoagulants, and details sur- rounding embolic events and bleeding camplications. The presence of atria1 fibrillation was accepted only on the basis of ECG documentation. Patients with alternating atria1 flutter and atria1 fibrillation were included, but those with pure atria1 flutter were excluded from the study. Two hundred fifty-four patients were identified. There were 123 men and 131 women, with a mean age of 54 t 11 years (range 18 to 86 years).

Etiology and type of atrial fibrillation (Table I). Cardio- vascular disease was classified into seven categories based on results of clinical evaluation, chest x-ray examination (all patients), M-mode echocardiography (n = 225 patients), and hemodynamic catheterization (n = 118 patients). The category of mitral valve disease included 120 patients (47 “6) with either hemodynamir, echocardio-

1039

1040 Roy et al. November 1986

American Heart Journal

Table I. Types of cardiovascular disease in 254 patients with atria1 fibrillation

Mitral valve disease 120 (47’C ) Aortic valve disease 10 ( 4cCJ) Cardiomyopathy 30 (12°C’) Ischemic heart disease 20 ( 8°C’) Hypertensive heart disease 14 ( 5”r) Other disease 23 ( 9’C’) None 37 (15’i )

graphic, or surgical (prior mitral commissurotomy) evi- dence of mitral stenosis and/or regurgitation. Of these 120 patients, 98 had pure or predominant mitral stenosis and 22 had pure or predominant mitral regurgitation. Ten patients (4Yb) had significant aortic valve disease without coexisting mitral valve disease. Thirty patients (12 “;# ) were diagnosed as having either congestive or hypertro- phic cardiomyopathy. The category of ischemic heart disease comprised 20 patients (89,) with either a prior myocardial infarction or angiographic evidence of a 2 70 O. stenosis of at least one major coronary artery. The group of hypertensive heart disease included 14 hypertensive patients (5:‘;)) with ECG and/or echocardiographic evi- dence of left ventricular hypertrophy. Miscellaneous dis- eases were found in 23 patients (906): atria1 septal defect (seven patients), car pulmonale (eight patients), hyper- thyroidism (four patients), pericarditis (two patients), Wolff-Parkinson-White syndrome (one patient), and myxoma (one patient). The arrhythmia was considered idiopathic in the remaining 37 patients (15”;‘).

Atria1 fibrillation was divided into chronic versus parox- ysmal. The fibrillation was considered chronic when it appeared on all consecutive ECGs during a minimum l-year period.

Follow-up information. The entry point of the study was measured from the time of first documentation by ECG of atria1 fibrillation. The end points were death, mitral valve replacement, and last in- or out-patient visit available for review. The average length of observation was 3.3 years (range 0.1 to 21 years). For each patient, the length of observation was measured in years with or without anticoagulants and in years during which atria1 fibrillation was either paroxysmal or chronic. Embolic events were examined in relation to type of cardiovascular disease, duration of atria1 fibrillation, and use of anticoag- ulants. Because it is often difficult to distinguish throm- botic from embolic events to the brain, all completed strokes or transient ischemic attacks were included in the analysis. However, none of the patients included in this study had either clinical or angiographic evidence of carotid atheromatous disease. Bleeding complications due to anticoagulation therapy were recorded if they required treatment such as cessation of anticoagulants, transfusion, or surgical intervention. Minor bruising and epistaxis were excluded. Control of anticoagulant therapy was assessed by monthly visits at the anticoagulant outpatient clinic or our institution. Prothrombin times were measured with a

rabbit brain and lung thromboplastin reagent. A pro- thrombin ratio of 1.8 to 2.3: 1 was considered t,he desirable therapeutic range.

Analysis of data. Differences among continuous vari- ables were analyzed by the Student’s t test,. Discrete variables were compared with the chi square test. Values are expressed as mean i SD. A p value of less than 0.05 was considered indicative of a significant difference. To test, the equal occurrence of embolic events between periods with or without anticoagulants, a Biomedical Data Package (BMDP) IL’” life table and survival function was used and censored data were analyzed with the general- ized Savage (Mantel-Cox) test.“’

RESULTS

During a total of 833 patient-years’ follow-up of atria1 fibrillation, there were 32 instances of embo- lism: nine limb embolisms, one renal and one mesen- teric embolism, four transient ischemic attacks, and 17 completed strokes. Of the 120 patients with mitral valve disease, 22 were anticoagulated throughout the period of observation, 69 were observed with and without anticoagulants, and 29 were never anticoagulated during the course of follow-up. Of the 134 patients without mitral valve disease, 14 were anticoagulated throughout the peri- od of follow-up, 58 were observed with and without anticoagulants, and 62 were not anticoagulated dur- ing the course of follow-up.

Table II lists the distribution of embolic events according to type of cardiovascular disease during the observation period without anticoagulants. Thirty (94 % ) of all embolic events occurred during the 549 patient-year follow-up (range 0.1 to 21 years, SD -+ 3.3 years) without anticoagulation therapy. Twenty-one involved the cerebral circulation and nine were extracerebral. One half of these events occurred in 14 of 98 (14.3%) unanticoagulated patients with mitral valve disease and the other half occurred in 14 of 120 (11.7%) unanticoagulated patients without mitral valve disease. The incidence of embolism per 100 patient-years without anticoag- ulation was 5.1 for patients with mitral valve disease compared to 5.91 for the remaining patients with atria1 fibrillation of non-mitral origin.

Table III describes the embolic events and bleed- ing complications during the period of observation on anticoagulant therapy. During the 284 patient- year follow-up (range 0.1 to 9.1 years, SD 4 2.1 years), 1 of 91 (1.1%) anticoagulated patients with mitral valve disease and 1 of 72 (1.4%) anticoagu- lated patients without mitral valve disease experi- enced limb embolism. Both patients had previously suffered a peripheral embolism while not anticoagu- lated. One patient died of necropsy-confirmed intra-

Volume 112

Number 5

Table II. Relationship between cardiovascular disease and embolic events in unanticoagulated patients wit,h atrial fibrillation (549 patient-years of observation, range 0.1 to 21 years, SD -+ 3.3 years) --.____ -- -.-- .---

Number (if fmholic fuents -.~ . -.._ _ -.-.-

Intracercbral Cnrdiouascular Number of patients __-

diseasr with embolic euents TIA Stroke Extracerfbrai Total .-. ..-.___ --

Mitral valve disease 14 1 ; 15

Aortic valve disease 1 1 1

Cardiomyopathy c5 - 4 I 5

lschemic heart disease ?I 1 II ,I 4

Hypertensive heart disease 2 1 1 .~~ ‘>

Other 2 ‘, 2

None 1 1 I

Table Ill. Relationship between cardiovascular disease, embolic events, and bleeding complications in anticoagulated patients with atria1 fibrillation (284 patient-years of observation, range 0.1 to 9.1 years, SD + 2.1 years)

-____ ---- __.- .-.. --.- ~~ .__.. - ~-_-

Number of patients Number oj biwdin:’

Numbrr of rmbollc cucxts C’ar!fio~Kw~ular

c~rrmplic~u:;cirrs with embolic or

disetrw hemorrhagic events Intracerebral Estrawrcbrul Intracerrbrai is’rirac.cwbra/ -- ._

Mitral valve disease 3 I :\ Aortic valve disease .-

Cardiomyopathy 1 1 Ischemic heart disease 1 ,

Hypertension heart disease Other None

-. -

1 z 1 L Total 6

-

cerebral hemorrhage and there were five instances of extracerebral bleeding complications in four pa- tients. The incidence of bleeding complications was 2.11 of 100 patient-years during the observation period on anticoagulants.

As summarized in Table IV, the incidence of embolic events from atria1 fibrillation was 5.46 of 100 patient-years’ follow-up without anticoagulation compared to 0.70 of 100 patient-years with anticoag- ulation (p < 0.002). Comparison of paroxysmal with chronic atria1 fibrillation during the unanticoagu- lated period showed equivalent incidence of embolic events (5.33 vs 5.93 of 100 patient-years). The two embolic events recorded during the observation period on anticoagulation were associated with the paroxysmal form of atria1 fibrillation.

Twelve of the 98 patients (12% ) with pure or predominant mitral stenosis and 2 of 22 patients (9%) with predominant mitral regurgitation had embolic events. There was no significant difference in left atria1 size (51 +- 8 mm, n = 12 patients, vs

52 L?I 9 mm, n = 96 patients) measured by echocar- diography and the presence of cardiomegaly by chest x-ray examination (60% vs 68%) between patients with mitral valve disease who had and those who did not have an embolic event. Eight of the 50 patients with either ischemic heart disease or car- diomyopathy had embolic events. No statistically significant difference could be detected in this group between patients with emboli and those who did not have emboli in terms of left atria1 size (42 rt 1 mm, n = 7 patients, vs 45 -t 7 mm, n = 33 patients), left ventricular end-diastolic dimension by echocardiog- raphy (57 -t 13 mm, n = 7 patients, vs 53 + 14 mm, n = 32 patients, ejection fraction by left ventricular radionuclide angiography (29 ? 17?G, n = 3 pa- tients, vs 28 +- 110/o, n = 13 patients), or radiologic evidence of cardiac enlargement (50% vs 60% ).

There were 152 attempted cardioversions during the course of follow-up. Cardioversion was preceded by anticoagulation therapy in 110 cases (24 weeks in 86). The arrhythmia recorded immediately prior

1042 Roy et a[. November 1996

American Heart Journai

Table IV. Incidence of embolic events per 100 patient- years from chronic or paroxysmal atria1 fibrillation with or without anticoagulation therapy

__I- Patient- Number of

year embolic Incidence/100 fOl/OLU-Up ellents patient-years

Without anticoagulation Paroxysmal AF Chronic AF

With anticoagulation Paroxysmal AF Chronic AF Total

AF = atrial fibrillation.

549 30 5.46 431 23 5.33 118 7 6.93 284 2 0.70 188 2 1.06 97 0 0.0

833 32 3.84

to cardioversion was atria1 fibrillation in 122 instances and atria1 flutter in 30. There were two embolic events (1.3%). Both occurred in two of the unanticoagulated patients with atria1 flutter at the time of cardioversion.

DISCUSSION

Previous studies have shown that anticoagulants reduce the risk of embolism in patients with atria1 fibrillation.1~4~5* M-M However, these trials were usual- ly limited to small numbers of patients and antico- agulants were in most instances administered to patients with rheumatic mitral valve disease who had already experienced a cerebral embolism. Infor- mation concerning the incidence of extracerebral embolism and the rate of bleeding complications due to anticoagulants was lacking in most of these studies.

In this retrospective study of 254 patients with atria1 fibrillation, the incidence of embolic events was eight times more frequent during the observa- tion period without anticoagulation (5.46 of 100 patient-years) when compared to follow-up with anticoagulation therapy (0.7 of 100 patient-years) (p < 0.002). The incidence of embolic events associ- ated with chronic atria1 fibrillation was similar to the incidence observed with paroxysmal atria1 fibril- lation, indicating that both forms carry an increased risk of embolism when not treated with anticoagu- lants. This finding is in agreement with recent observations.‘5*21 Comparison of the combined inci- dence of embolic and bleeding complications during anticoagulation therapy with the incidence of embolic events without anticoagulants (2.81 vs 5.46 of 100 patient-years) suggests that in this study population the benefit of treatment outweighs the hemorrhagic risk.

An important finding in this study was the high

incidence of embolism during unanticoagulated patient-years of observation, whether or not atria1 fibrillation was associated with mitral valve disease. This would support the view expressed by some investigators that anticoagulants may be indicated to prevent embolism in nonvalvular atria1 fibrilla- tion.‘j, “‘. “I. ‘?“ It is generally believed that anticoagu- lants should be used only in patients with a prior history of embolism or when the arrhythmia is coupled with rheumatic mitral disease.4. s. I:. z’, “I In 197’7, the Joint Committee for Stroke Resources reviewed 11 trials of anticoagulation for the preven- tion of cerebral embolism from cardiac sources and concluded that the data have been consistent and suggest that anticoagulants reduce the incidence of cerebral embolic episodes, particularly in patients with rheumat,ic heart disease.” The usefulness of anticoagulant therapy is only assumed, since no controlled study has been done to define the advan- tages and risks in any group of patients with atria1 fibrillation. However, on the basis of these earlier reports, it is unlikely that a double-blind trial will be performed to evaluate the usefulness of anticoagula- tion therapy, especially in patients with atria1 fibril- lation due to rheumatic valvular heart disease.

The results of this study suggest that patients with nonrheumatic atria1 fibrillation also benefit from anticoagulation therapy. This observation is of importance, since rheumatic heart disease is now less often a precursor of atria1 fibrillation.‘4 The results also show that the occurrence of embolic events in patients with mitral valve disease, ischem- ic heart disease, and cardiomyopathy is unrelated to the indexes of left atria1 or ventricular size.

Two of the embolic events occurred after cardio- version. The incidence of post cardioversion emboli- zation in our study was 1.3 % , which is similar to what has previously been reported.10~26~2~ It is inter- esting that both cases occurred in unanticoagulated patients with paroxysmal atria1 fibrillation who were in atria1 flutter at the time of cardioversion. A firm belief exists that conversion of atria1 flutter is unlikely to embolize.28 Our observations suggest that cardioversion of atria1 flutter in a patient with previously documented atria1 fibrillation carries a risk of embolization and should be preceded by anticoagulation therapy.

Our data were based on analysis of medical records obtained from a hospital population of patients with atria1 fibrillation. This selection bias probably resulted in a higher incidence of events than would have been observed compared in a general population of subjects with atria1 fibrilla- tion. Another limitation of this study is that the

Volume 112

Number 5

duration of observation off anticoagulant therapy is longer than on anticoagulant therapy, which may have contributed to the higher incidence of embo- lism during the unanticoagulated period of observa- tion. Despite these considerations, the difference in incidence of embolic events between unanticoagu- lated and anticoagulated patient-year follow-up indicates that anticoagulants were useful in this study group. This group was chosen because it permitted review of comprehensive cardiovascular examinations from the same institution, and circum- stances surrounding either embolic or hemorrhagic complications could be properly ascertained. Anti- coagulated patients were more closely followed because they required more hospital visits than the unanticoagulated patients. However, this bias could give more support to our conclusions, since it per- mitted better detection of complications during the course of follow-up on anticoagulants.

We conclude that in this selected population of patients with atria1 fibrillation, anticoagulants reduced the incidence of embolic events and were associated with an acceptably low rate of hemor- rhagic complications. The incidence of embolism in unanticoagulated patients with atrial fibrillation was high irrespective of the presence or absence of mitral valve disease, indicating that anticoagulants should not be reserved solely for patients in whom the arrhythmia is due to rheumatic heart disease.

REFERENCES

1. Askey JM, Cherry CB: Thromboembolism associated with auricular fibrillation. JAMA 1950:144:97.

2.

3.

4.

5.

6.

7.

Daley R, Mattingly TW, Holt CL, Bland EF, White PD: Systemic arterial embolism in rheumatic heart disease. AM HEART J 1951;42:566. Rowe JC, Bland EF, Sprague MB, White PD: The course of mitral stenosis without surgery. Ten and twenty year per- snective. Ann Intern Med 1960:52:741. Szekely P: Systemic embolism and anticoagulant prophylaxis in rheumatic heart disease. Br Med J 1964:1:1209. Coulshed N, Epstein EJ, McKendrick CS, Galloway RW, Walker E: Systemic embolism in mitral valve disease. Br Heart J 1970;32:26. Wolf PA, Dawber TR, Thomas HE, Kannel WB: Epidemio- logic assessment of chronic atria1 fibrillation and risk of stroke. The Framingham Study. Neurology 1978;28:973. Evans W, Swann P: Lone auricular fibrillation. Br Heart ,J 1954;16:189.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24

25.

26.

27.

28.

Anticnagulants and atrial ;Ibrilrntrcm 1043

Weintraub G, Sprecace G: Paroxysmal atria1 fibrillation and cerebral embolism with annarentlv normal ht*art-. N Engl .I Med 1958;259:875. 1 1 - Beer DT, Ghitman B: Embolization from the atria in arterio- sclerotic heart disease. JAMA 1961;177:287. Hurst JW. Paulk EA. Proctor H. Schlant RC: Management of patients with atria1 fibrillation.‘Am J Med 1964;37:728. Vost A, Wolochov DA, Howell DA: Incidence of infarcts of the brain in heart disease. d Path01 Bacterial 1964;88:46:1. Aberg H: Atria1 fibrillation. A study of atrial tthrombosis and systemic embolism in a necropsv material. Ai:ta Med Stand <969;185:373.

. ”

Hinton RC, Kistler JP, Fallon ,JT, Friedlicli %L. Fisher (‘M: Influence of etiology of atria1 fibrillation I~II incidence of systemic embolism. Am J Cardiol 1977;40:509. Kannel WB, Abbott RD, Savage DD, McNamara PM: Epide- miologic features of chronic atria1 fibrillation The Framing- ham Study. N Engl J Med 1982;306:1018. Seizer A: Atria1 fibrillation revisited. ‘\i Entrl ,f Med 1982; 306: 1044. Cosgrift SW: Prophylaxis of recurrent embolism of intracar- diac origin. JAMA 1950;143:870. Carter AB: Prognosis of cerebral rmholinn~. Lancet I!+%: 2:514. Adams GF, Merrett dD, Hutchison WM, I’cttlock AM: Cere- bral embolism and mitral stenosis: Survival with and without anticoagulants. .J Neurol Neurosurg Psych& ry 1974;37:378. Dixon WJ, Brown MB: Biochemical computer programs P-series. Berkeley: IJniversity of California Press. 1979: 741. Mantel N: Evaluation of survival data anti :wo uew rank order statistics arising in its consideration Cancer Cherno- ther Rep 1966;50: 163. Sage JI, Van Uitert RL: Risk of recurrent stroke in patients with atria1 fibrillation and non-valvular heart disease, Stroke 1983;14:537. Wolf PA, Kannel WB. McGee DL, Meeks SL, Bharucha NE, McNamara PM: Duration of atria1 tihrillat.ion and immi- nence of stroke: The Framingham Study. Stroke 1983:14:664. Fowler NO: Cardiac diagnosis and trratmc nt. :lrd ed. New York. Harrow Books, 1980: 1061. Morris DC, Hurst JW: Atria1 fibrillatit,n. l’!rri Probt (‘ardiol 1980;5:44. Genton E, Barnett HJM, Fields WS, (:rnt M, Hoak dC: Report of the ,Jaint Committee on Stroke Resources. XIV. Cerebral ischemia: The role of thrombosis and 11f’ antithrom- bit therapy. Stroke 1977;8:150. Lown B, Perlroth MG, Kaidbey S. Ahe -1’. Harken DE: Cardioversion of atria1 fibrillation. A report on the treatment of 65 episodes in 50 patients. N Engl J Med 1963;269:325. McCarthy P, Varghese PJ, Barritt DW: Prognosis of atria1 arrhythmias treated by electrical counter .;hock therapy. A three year follow-up. Br Heart J 1969:31:496. Mancini GB, Golberger AL: Cardiuversioti of atria1 fibrilla- t.ion: Consideration of embolization. anricoa&ation, prophl lactic pacemaker, and long-term ~\tc~~e,~:~. ,\%I H~AI~I. .l 1982:104:61i.