Skip main navigation

Stroke Severity in Atrial Fibrillation

The Framingham Study
Originally published 1996;27:1760–1764


    Background and Purpose Stroke occurring with atrial fibrillation (AF) is more likely to be fatal or more severe than non-AF stroke based on clinical series, but data from prospective epidemiological studies are sparse and inconsistent.

    Methods Over 40-year follow-up of the original 5070 Framingham cohort, 501 initial ischemic strokes, including 103 with AF, were analyzed. Stroke severity was rated as none, mild, moderate, severe, or fatal. Since 1981, functional status indicated by the Barthel index has been evaluated acutely and at 3, 6, and 12 months. Severity and functional status of AF strokes were compared with non-AF strokes using χ2 test and Student's t test. Thirty-day mortality was assessed by logistic regression analyses.

    Results AF was associated with increased stroke severity (P=.048). Thirty-day mortality was greater in AF strokes than in non-AF strokes (25% versus 14%). The multivariate-adjusted odds ratio for 30-day mortality for AF subjects was 1.84 (95% confidence interval, 1.04 to 3.27). Since 1981, follow-up was available for 150 initial ischemic strokes, including 30 with AF. Compared with the non-AF group, the AF group had poorer survival and more recurrences during 1 year of follow-up. The AF subjects had lower mean Barthel index scores acutely (29.6 versus 58.6, P<.001) and at 3 months (P=.005), 6 months (P=.003), and 12 months (P=.130) after stroke among survivors.

    Conclusions Ischemic stroke associated with AF was nearly twice as likely to be fatal as non-AF stroke. Recurrence was more frequent, and functional deficits were more likely to be severe among survivors. Since stroke is usually the initial manifestation of embolism in AF, prevention is critical to reducing disability and mortality.

    Atrial fibrillation (AF) is the most prevalent chronic cardiac arrhythmia in the elderly and is associated with a fourfold to fivefold increase in the risk of stroke.123 Whether stroke occurring in the presence of AF tends to be more severe than non-AF stroke is unresolved. Fisher reported in a series of consecutive hospitalized stroke patients with AF that more than 70% of the strokes were lethal or resulted in an unacceptable neurological deficit.4 However, in more recent comparison studies, although some suggested a possible role of AF in the risk for more severe stroke,56789101112 others did not show this association.2131415 This inconsistency might be due to differences in setting (hospital or population), study design (case-series, case-control, or follow-up study), and/or outcome measurement (neurological deficit, functional disability, or death at various time points after stroke). Moreover, few of these reports were population-based prospective epidemiological studies.11121415 We believe that further clarification of the relation between AF and stroke outcome is important to enhance the knowledge and understanding of the burden from AF. Using the long-term prospective follow-up of the original Framingham Study cohort, the present study analyzed the first-ever ischemic strokes occurring over a 40-year period. We studied the influence of AF on stroke severity, mortality, and functional disability by means of systematic and serial assessments in stroke subjects.

    Subjects and Methods

    The Framingham cohort of 5070 men and women between 30 and 62 years of age and free of cardiovascular disease including stroke at study entry has been examined once every 2 years since the time period 1948 through 1950. The details of the study design, implementation, and diagnosis criteria have been published previously.16 Briefly, at each routine biennial examination, physical examinations and laboratory work including a 12-lead electrocardiogram (ECG) were performed, and risk factors for cardiovascular diseases were systematically evaluated. In addition, detailed medical information surrounding all interim illness was sought and reviewed. AF was identified both at the time of biennial evaluation and on interim hospitalizations or outside examinations. Onset of AF was defined as the time of the first documentation on ECG with no prior history of AF. Prophylactic anticoagulation was rarely used for prevention of stroke or systemic embolism in these Framingham AF subjects. In the present study, other risk factors of interest documented before stroke included coronary heart disease, valvular heart disease (defined as the presence of a loud cardiac murmur), hypertension, antihypertensive therapy, congestive heart failure, cigarette smoking, and diabetes.

    Stroke surveillance was maintained by daily monitoring of all admissions to the only general hospital in the town. Since 1968, whenever possible, a study neurologist has examined subjects with suspected stroke at the time of acute hospitalization. The clinical records for each cerebrovascular event were reviewed by a panel of investigators that included a neurologist. Prior to the availability of CT scan, most stroke subjects had lumbar puncture, brain scan, electroencephalogram, and skull radiograph. In recent years, at least one CT scan of the brain has been obtained in 85% of stroke cases. On the basis of the detailed clinical, laboratory, and radiological data, stroke subtypes were determined using uniform criteria. Ischemic stroke was considered present if there was sudden onset of a localizing neurological deficit lasting for 24 hours, in the absence of intracranial hemorrhage (assessed by CT scan or in prior years by clinical and laboratory findings) and other disease processes causing focal brain deficits. Stroke severity was evaluated on the basis of the neurological deficits present during the hospitalization. These deficits were classified into four categories: none, no deficit/impairment; mild, a deficit was present in visual, communication, motor, and/or sensory realms, but the impairment was not sufficient to interfere with functional independence; moderate, the deficit was severe enough that the patient required assistance in any one of these domains; and severe, the patient was functionally dependent on others in two or more domains. Stroke resulting in death within 30 days after onset was considered fatal.

    Since 1981, a protocol for standardized serial assessment of stroke subjects has been used. Stroke patients were examined in the hospital within 48 hours after onset and again at 3, 6, and 12 months after stroke. Detailed neurological examination, determination of functional status, and evaluation of cognitive functions were conducted at each examination by a neurologist and a neurological nurse. Functional performance was measured using a modified Barthel index (BI) for activities of daily living (ADL).1718 The composite BI is scaled over a 100-point range, with 100 representing complete independence and 0 corresponding to complete dependence. To express disability more explicitly, three levels of ADL were determined according to the BI: A score of 40 or less defined a group of patients with severe dependence in ADL; a score of 85 or above indicated mild to no dependence in ADL; and an intermediate score implied moderate dependence in ADL. During the follow-up, medical data for each recurrent stroke were reviewed at the regular panel meetings, and the serial assessment continued despite the recurrence.

    In the present study, only the first-ever ischemic stroke events were analyzed. Stroke subjects associated with AF were defined as those persons whose AF was documented before or at the time of stroke onset. Otherwise the strokes were considered to be non-AF events. For this study, 40 years of follow-up data (1950 through 1991) were available for analyses.

    Statistical Analysis

    Comparison was made between stroke subjects with and without AF. Continuous data were compared with Student's t test and categorical data with the χ2 test. Logistic regression modeling was applied to analyze the impact of AF on 30-day mortality. A bivariate model, adjusting for age, was first used to examine each potential predictor of outcome. Variables with a value of P<.1 in the bivariate model were entered into a multivariate model with AF and age as forced-in variables, and then a backward stepwise method was used to select the final model. Survival and recurrence after stroke were estimated by the Kaplan-Meier product-limit method, and the log-rank test was used for comparison between groups. The two-tailed significance level was set at P=.05.


    During 40 years of follow-up in the cohort of 5070 subjects, an initial ischemic stroke occurred in 508 (220 men and 288 women). Among the stroke subjects, 104 (20%) had AF documented before (n=85) or at the time of (n=19) stroke onset. Seven subjects (6 non-AF and 1 AF) whose stroke severity could not be determined were excluded, leaving 398 non-AF and 103 AF strokes for analyses. The AF subjects differed significantly from the non-AF subjects in several respects: They were older, they had a higher prevalence of congestive heart failure and valvular heart disease, and fewer smoked cigarettes (Table 1).

    Stroke Severity

    Severity of stroke was related to the presence of AF and was more likely to be severe or fatal in the AF subjects (P=.048, Table 2). Since age is an important determinant of stroke outcome, we compared stroke severity between subjects with and without AF in three age strata: <65, 65 to 74, and ≥75 years. With decreased sample sizes after stratification, the differences in stroke severity were not statistically significant (Table 2). However, AF strokes were more likely to be fatal.

    30-Day Mortality

    To analyze the influence of AF on the poststroke 30-day mortality, AF was entered into a bivariate logistic regression model adjusting for age. The age-adjusted odds ratio for AF was 1.76 (95% confidence interval, 1.02 to 3.02). The effect of AF was then evaluated by a multivariate model adjusting for age, coronary heart disease, and cigarette smoking, the latter two of which were selected with a value of P<.1 in age-adjusted bivariate models. In the multivariate model, age, cigarette smoking, and AF were each significantly associated with the risk of death within 30 days (Table 3). Mortality was increased 1.84-fold in the presence of AF.

    Survival and Functional Status After Stroke in Subjects Since 1981

    During the 10-year period since 1981, post-stroke follow-up was available for 150 subjects with first-ever ischemic stroke, including 30 associated with AF. The AF subjects were older than the non-AF subjects (mean±SEM, 81.1±1.6 versus 77.6±0.7 years, P=.025). Comparison of other clinical features between the AF and non-AF strokes for these 150 subjects was similar to that for all the stroke subjects shown in Table 1.

    By 1 year of follow-up, 63% (19/30) of the AF subjects and 34% (41/120) of the non-AF subjects had died. Stroke recurred in 23% (7/30) of the AF group and 8% (9/120) of the non-AF group. Most recurrent strokes in AF subjects occurred within 30 days (5/7). Recurrence within 30 days was less common in non-AF subjects, occurring in only 2 of 9. Thus, survival was significantly poorer and the recurrence rate was significantly higher in subjects with AF compared with those without AF (Figure). The 30-day mortality after stroke in association with AF was 30% (9/30) compared with 17% (20/120) in non-AF subjects; this was similar to the 30-day mortality in the entire cohort, 25% for AF strokes and 14% for non-AF strokes (Table 2).

    Acutely, subjects with AF had a significantly lower mean BI score compared with subjects without AF (Table 4). After adjustment for other potential confounders including variables listed in Table 1, this association persisted with no substantial change in the magnitude of difference. Nearly three quarters of AF subjects were severely dependent in ADL, compared with about one third of non-AF subjects. Among the survivors at 3 months and 6 months after stroke, the AF group continued to have significantly lower mean scores than the non-AF group. At 12 months of follow-up, subjects with AF performed at a lower functional level than the subjects without AF, although the difference was no longer significant.


    We systematically evaluated the influence of AF on the outcome of ischemic stroke. Stroke constitutes the most common lethal and disabling neurological disease of adult life.19 Most of the previous studies of stroke outcome emphasized only mortality. In the present study, we evaluated both mortality and disability in stroke subjects. The 30-day mortality is a distinct and commonly used indicator for stroke outcome. Stroke severity based on the number of neurological impairments at the acute stage, and disability in functional status measured by BI, whose reliability and validity are well established,1718 provide an added dimension to outcomes research. One scale categorizes the specific neurological deficits, and the other measures the impact of disability on daily activities in self-care.

    In evaluating the impact of AF on stroke outcome, we selected only prestroke factors as potential confounders to adjust for in the multivariate analyses. Unlike most other outcome studies, we did not include possible “poststroke” prognostic factors that reflected either the general condition of the patients such as level of consciousness and degree of hemiparesis or the site and size of the lesions shown in imaging studies. The rationale for this decision is that currently for stroke, primary prevention provides the greatest potential for reducing the burden of this disease. Identifying not only the predictors associated with higher risk of stroke but also the “prestroke” factors associated with poorer stroke outcome underscores the efforts in modifying the risk factors before the event occurs.

    Our study showed that stroke severity in the AF subjects was greater than in the non-AF subjects. The poorer outcome in the AF subjects was mostly reflected in the high early mortality. The finding of a significantly higher 30-day mortality in AF strokes, 25% versus 14%, was similar to the previous reports with the range of 23% to 35% in AF subjects versus 7% to 14% in non-AF subjects.101113 As stroke was more lethal in older age, it seems that the difference in mortality between AF and non-AF subjects also became more evident in the extreme elderly, although the differences did not reach statistical significance. It is notable that half of the strokes in AF subjects 75 years of age or older were either severe or fatal. This strongly suggests that prevention of stroke with warfarin anticoagulation in elderly subjects with AF is warranted.

    The multivariate-adjusted odds ratio of 30-day fatality in our study was 1.84, an estimate lower than that in a community-based stroke registry (odds ratio, 3.5; with a 95% confidence interval, 1.9 to 6.2, adjusted for age and infarction subtypes).11 This variation might be due in part to the variables selected to adjust for in logistic regression.

    Only a few studies evaluated AF as a prognostic factor for functional outcome after stroke.2467913 Some of these studies were descriptive and did not include analytical comparisons,4613 whereas others combined functional status with mortality in the outcome assessment.79 Moreover, comparison of disability as a stroke consequence among studies is hampered by variability in the selection of subjects, in the rating system or measuring instrument used, and in the time of assessment. In general, most of these reports showed that AF patients were older and more disabled. In our study, comparison of the functional status indicated by BI between subjects with and without AF revealed that the largest difference was at the acute stage: Nearly three quarters of AF subjects, compared with approximately one third of non-AF subjects, were severely dependent in ADL. This difference was not attributable to other factors such as age and heart diseases. Overall, acute stroke severity was significantly greater in AF-associated stroke.

    At 3 months after stroke, 75% of the AF subjects remained moderately or severely dependent in ADL. A pooled analysis of control subjects in five clinical trials of antithrombotic therapy in AF showed that 44% (36/81) of the strokes left a functional deficit 1 to 3 months after the event.20 Older age (10 years older on average) and greater comobidity in our AF subjects than in those controls of the clinical trials might contribute to the poorer functional performance.

    The difference of functional status between AF and non-AF subjects decreased at 12 months after stroke and was no longer statistically significant. This decrease of difference might be explained by the poorer survival for AF subjects during the follow-up period as shown in the Figure. More AF subjects died. Those AF survivors might have better functional performance with higher average BI scores, and thus the disparity with the non-AF subjects diminished. It was also likely that the small number of stroke survivors with AF might have affected the power of comparison.

    Several hypotheses of pathogenesis are of interest with respect to the association of AF with worsened stroke outcome. Some authors speculated that the higher prevalence of ischemic heart disease and congestive heart failure in AF subjects would have contributed to a reduced cerebral blood flow when the cerebral autoregulatory mechanisms are impaired as in an ischemic region.21 In the present study, AF itself seemed to carry an excess risk of early mortality and poor functional outcome independent of the underlying cardiac disease. Other researchers have found that chronic AF may cause a significant reduction in regional cerebral blood flow.22 On the other hand, in terms of the stroke mechanism, collateral circulation in the brain may be less developed in patients with a sudden interruption of blood flow caused by an embolus than in those who suffer a stroke as the result of an underlying chronic arterial atherosclerotic disease.29 These two factors may further compromise the cerebral circulation, increase the infarction size, and retard the recovery of function after ischemia associated with AF.

    Study Strengths and Limitations

    This study was based on a general population sample and thus minimized the bias of case selection encountered in hospital-based series that were more likely to contain an excess of more severe stroke. The prospective design allowed all risk factors to be ascertained before the stroke event and eliminated possible outcome-associated bias. Therefore, our findings provide the least distorted picture of stroke outcome in terms of survival and functional status. However, we did not evaluate social and psychological factors that might be as important as biological factors in determining the functional outcome of these stroke patients.2324 The fact that not every stroke survivor could attend each set time follow-up examination might have introduced some bias into our analyses. Nevertheless, we believe that this bias would not substantially affect the findings since the major reason for subjects not receiving a detailed follow-up neurological examination and functional evaluation was geographical.

    Conclusions and Clinical Implications

    We found that stroke patients with AF had higher mortality, more recurrences, graver severity, and poorer functional status than those without AF. These unfavorable outcomes were particularly prominent in the early poststroke course. Therefore, not only did subjects with AF have increased risk of stroke but the strokes that occurred were more serious. Since stroke is usually the initial manifestation of embolism in AF, primary prevention remains key to reduction of stroke mortality and morbidity in persons with AF.

          Figure 1.

    Figure 1. One-year Kaplan-Meier survival and stroke recurrence curves for 120 non-AF subjects and 30 AF subjects. AF indicates atrial fibrillation.

    Table 1. Clinical Characteristics in Stroke Subjects With and Without Atrial Fibrillation

    AF (n=103)Non-AF (n=398)
    Age, years*†75.4±1.071.2±0.5

    AF indicates atrial fibrillation; CHD, coronary heart disease; CHF, congestive heart failure; VHD, valvular heart disease; HBP, high blood pressure; and HRx, antihypertensive therapy.


    P<.001, χ2 test.

    Table 2. Stroke Severity in Subjects With and Without Atrial Fibrillation

    All AgesAge <65Age 65-74Age ≥75
    Stroke SeverityAF (n=103)Non-AF (n=398)AF (n=13)Non-AF (n=97)AF (n=31)Non-AF (n=140)AF (n=59)Non-AF (n=161)

    AF indicates atrial fibrillation.

    2 test, comparing AF group with non-AF group.

    Table 3. Multivariate Regression Model for the Risk of Poststroke 30-Day Mortality

    VariableOdds Ratio* (95% Confidence Interval)P
    Age (/10 yrs)1.66 (1.27-2.17)<.001
    Smoking1.87 (1.07-3.27).028
    CHD1.74 (0.98-3.08).061
    AF1.84 (1.04-3.27).036

    CHD indicates coronary heart disease; AF, atrial fibrillation.

    Table 4. Barthel Index at Acute Stage of Stroke and at 3-Month, 6-Month, and 12-Month Follow-up

    TimeBarthel IndexAFNon-AFP*
    Dependence level†
    3 Monthsn=12n=49
    Dependence level
    6 Monthsn=11n=57
    Dependence level
    12 Monthsn=10n=55
    Dependence level

    AF indicates atrial fibrillation.

    *Student's t test for Barthel index (BI) as a continuous variable, χ2 test for BI as a categorical variable.

    †For definitions of dependence level, refer to the text.

    This study was supported by grants from the National Institute of Neurological Disorders and Stroke (2-RO1-NS-17950-15) and National Heart, Lung, and Blood Institute (Contract NIH-NO1-HC-38038).


    Correspondence to Philip A. Wolf, MD, Department of Neurology, Boston University School of Medicine, 80 E Concord St, B-608, Boston, MA 02118. E-mail [email protected].


    • 1 Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent risk factor for stroke: the Framingham Study. Stroke.1991; 22:983-988.CrossrefMedlineGoogle Scholar
    • 2 Britton M, Gustafsson C. Non-rheumatic atrial fibrillation as a risk factor for stroke. Stroke.1985; 16:182-188.CrossrefMedlineGoogle Scholar
    • 3 Sherman DG, Goldman L, Whiting RB, Jurgensen K, Kaste M, Easton JD. Thromboembolism in patients with atrial fibrillation. Arch Neurol.1984; 41:708-710.CrossrefMedlineGoogle Scholar
    • 4 Fisher CM. Reducing risks of cerebral embolism. Geriatrics.1979; 34:59-61.MedlineGoogle Scholar
    • 5 Lowe GD, Jaap AJ, Forbes CD. Relation of atrial fibrillation and high haematocrit to mortality in acute stroke. Lancet.1983; 1:784-786.CrossrefMedlineGoogle Scholar
    • 6 Friedman PJ. Atrial fibrillation after stroke in the elderly. Stroke.1991; 22:209-214.CrossrefMedlineGoogle Scholar
    • 7 Allen CMC. Predicting the outcome of acute stroke: a prognostic score. J Neurol Neurosurg Psychiatry.1984; 47:475-480.CrossrefMedlineGoogle Scholar
    • 8 Woo J, Kay R, Yuen YK, Nicholls MG. Factors influencing long-term survival and disability among three-month stroke survivors. Neuroepidemiology.1992; 11:143-150.CrossrefMedlineGoogle Scholar
    • 9 Censori B, Camerlingo M, Casto L, Ferraro B, Gazzaniga GC, Cesana B. Prognostic factors in first-ever stroke in the carotid artery territory seen within 6 hours after onset. Stroke.1993; 24:532-535.CrossrefMedlineGoogle Scholar
    • 10 Gustafsson C, Britton M. Pathogenetic mechanism of stroke in non-valvular atrial fibrillation: follow-up of stroke patients with and without atrial fibrillation. J Intern Med.1991; 230:11-16.CrossrefMedlineGoogle Scholar
    • 11 Sandercock P, Bamford J, Dennis M, Burn J, Slattery J, Jones L, Boonyakarnkul S, Warlow C. Atrial fibrillation and stroke: prevalence in different types of stroke and influence on early and long term prognosis (Oxfordshire community stroke project). BMJ.1992; 305:1460-1465.CrossrefMedlineGoogle Scholar
    • 12 Abu-Zeid AH, Choi NW, Hsu PH, Maini KK. Prognostic factors in the survival of 1,484 stroke cases observed for 30 to 48 months, II: clinical variables and laboratory measurements. Arch Neurol.1978; 35:213-218.CrossrefMedlineGoogle Scholar
    • 13 Candelise L, Pinardi G, Morabito A. Mortality in acute stroke with atrial fibrillation: The Italian Acute Stroke Study Group. Stroke.1991; 22:169-174.CrossrefMedlineGoogle Scholar
    • 14 Krahn AD, Manfreda J, Tate RB, Mathewson FA, Cuddy TE. The natural history of atrial fibrillation: incidence, risk factors, and prognosis in the Manitoba Follow-Up Study. Am J Med.1995; 98:476-484.CrossrefMedlineGoogle Scholar
    • 15 Anderson CS, Jamrozik KD, Broadhurst RJ, Stewart-Wynne EG. Predicting survival for 1 year among different subtypes of stroke: results from the Perth Community Stroke Study. Stroke.1994; 25:1935-1944.CrossrefMedlineGoogle Scholar
    • 16 Shurtleff D. Some characteristics related to the incidence of cardiovascular disease and death: the Framingham Study, 18-year follow-up. In: Kannel WB, Gordon T, eds. The Framingham Study: An Epidemiological Investigation of Cardiovascular Disease. Washington, DC: Dept of Health, Education, and Welfare; 1974: section 30. DHEW publication NIH 74-599.Google Scholar
    • 17 Granger CV, Dewis LS, Peters NC, Sherwood CC, Barrett JE. Stroke rehabilitation: analysis of repeated Barthel index measures. Arch Phys Med Rehabil.1979; 60:14-17.MedlineGoogle Scholar
    • 18 Mahoney FI, Barthel DW. Functional evaluation: the Barthel Index. Md State Med J.1965; 14:61-65.MedlineGoogle Scholar
    • 19 Modan B, Wagener DK. Some epidemiological aspects of stroke: mortality/morbidity trends, age, sex, race, socioeconomic status. Stroke.1992; 23:1230-1236.CrossrefMedlineGoogle Scholar
    • 20 Anonymous. Risk factors for stroke and efficacy of antithrombotic therapy in atrial fibrillation: analysis of pooled data from five randomized controlled trials—Atrial Fibrillation Investigators: Atrial Fibrillation, Aspirin, Anticoagulation Study; Boston Area Anticoagulation Trial for Atrial Fibrillation Study; Canadian Atrial Fibrillation Anticoagulation Study; Stroke Prevention in Atrial Fibrillation Study; Veterans Affairs Stroke Prevention in Nonrheumatic Atrial Fibrillation Study. Arch Intern Med.1994; 154:1449-1457.CrossrefMedlineGoogle Scholar
    • 21 Keller TS, McGillicuddy JE, LaBond VA, Kindt GW. Volume expansion in focal cerebral ischemia: the effect of cardiac output on local cerebral blood flow. Clin Neurosurg.1982; 29:40-50.CrossrefMedlineGoogle Scholar
    • 22 Lavy S, Stern S, Melamed E, Cooper G, Keren A, Levy P. Effect of chronic atrial fibrillation on regional cerebral blood flow. Stroke.1980; 11:35-38.CrossrefMedlineGoogle Scholar
    • 23 Glass TA, Matchar DB, Belyea M, Feussner JR. Impact of social support on outcome in first stroke. Stroke.1993; 24:64-70.CrossrefMedlineGoogle Scholar
    • 24 Kelly-Hayes M, Wolf PA, Kannel WB, Sytkowski P, D'Agostino RB, Gresham GE. Factors influencing survival and need for institutionalization following stroke: the Framingham Study. Arch Phys Med Rehabil.1988; 69:415-418.MedlineGoogle Scholar


    eLetters should relate to an article recently published in the journal and are not a forum for providing unpublished data. Comments are reviewed for appropriate use of tone and language. Comments are not peer-reviewed. Acceptable comments are posted to the journal website only. Comments are not published in an issue and are not indexed in PubMed. Comments should be no longer than 500 words and will only be posted online. References are limited to 10. Authors of the article cited in the comment will be invited to reply, as appropriate.

    Comments and feedback on AHA/ASA Scientific Statements and Guidelines should be directed to the AHA/ASA Manuscript Oversight Committee via its Correspondence page.