Trend in Mortality after Stroke with Atrial Fibrillation

The American Journal of Medicine (2007) 120, 47-53

CLINICAL RESEARCH STUDY

AJM Theme Issue: Cardiology

Trend in Mortality after Stroke with Atrial Fibrillation Lars Frost, MD, PhD,a Ljubica Vukelic Andersen, MD,a Peter Vestergaard, MD, PhD, DMSc,b Steen Husted, MD, DMSc,a Leif Spange Mortensen, MScc a

Department of Cardiology A, and bDepartment of Endocrinology and Metabolism C, Aarhus University Hospital, and cUNI-C, The Danish Information Technology Centre for Education and Research, Aarhus, Denmark. ABSTRACT PURPOSE: To evaluate trend in mortality in stroke associated with atrial fibrillation, we examined mortality trend after stroke with atrial fibrillation by calendar year period (1980-1984, 1985-1989, 1990-1994, 1995-1999, and 2000-2002). We estimated trends separately for each sex in unadjusted analyses. We also adjusted for age, comorbid conditions, and general trend in mortality in the background population. METHODS: We identified all individuals, aged 40-89 years, with an incident diagnosis of stroke of any nature (ischemic or hemorrhagic) and no history of heart valve disease and a previous or concomitant diagnosis of atrial fibrillation or flutter in the Danish National Registry of Patients. Subjects were followed in the Danish Civil Registration System for emigration and vital status. We used multivariate Cox proportional hazards regression analysis to estimate trend in mortality. RESULTS: Incident stroke with a previous or concomitant diagnosis of nonvalvular atrial fibrillation or flutter was diagnosed in 24,470 subjects (11,554 men and 12,916 women). During 34,405 years of observation, 9237 men died, and during 35,381 years of observation, 10,827 women died. The hazard ratio for mortality after stroke in the last 3-year period compared with the first 5-year period was .65 (95% confidence interval [CI], .61-.71) in men and .69 (95% CI, .64-.74) in women. CONCLUSIONS: We observed a substantially better survival in men and women after stroke associated with atrial fibrillation or flutter in Denmark during the years 1980 to 2002. However, we could not control for changes in admission practice, diagnostic performance, or treatment. © 2007 Elsevier Inc. All rights reserved. KEYWORDS: Arrhythmia; Cohort study; Epidemiology; Mortality

Stroke is the third leading cause of death in developed countries such as the United States and many European countries.1 Reports from the United States and Europe have indicated that stroke-associated mortality has decreased during the period from 1970 to 2000 in developed countries.2 Atrial fibrillation is an important cause of stroke, especially in elderly subjects.3 The outcome (ie, disability and mortality) in stroke with atrial fibrillation is worse than in stroke without atrial fibrillation.4-6 Embolization from the heart, in particular from the left atrium in patients with atrial Reprint requests should be addressed to Lars Frost, MD, PhD, Department of Cardiology A, Aarhus University Hospital, Tage Hansens Gade, DK-8000 Aarhus C, Denmark. E-mail address: [email protected].

0002-9343/$ -see front matter © 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.amjmed.2005.12.027

fibrillation, accounts for 17%-27% of all strokes.2 Approximately 17% of all deaths among stroke victims are attributable to atrial fibrillation.6 We do not have much information on trend in mortality in stroke with atrial fibrillation, but investigators from the Mayo Clinic recently reported that survival after stroke with atrial fibrillation in Olmsted County, Minnesota, United States, has remained unchanged over time in the period from 1980 to 2000 in men and women.7 However, the study from Olmsted County was based on 363 deaths during follow-up in 446 subjects with incident stroke and may, therefore, have been statistically underpowered due to the relatively low number of outcomes. We examined national trends in mortality by calendar year period of stroke in patients with atrial fibrillation in

48

The American Journal of Medicine, Vol 120, No 1, January 2007

Denmark (1980-1984, 1985-1989, 1990-1994, 1995-1999, and 2000-2002). We estimated trends separately for each sex in unadjusted analyses. We also adjusted for age, stroke severity, comorbid conditions, and general mortality in the background population.

METHODS Study Period and Study Population

cian who discharged a patient coded all diagnoses for that patient. A change in ICD coding from ICD-8 to ICD-10 occurred in Denmark in the beginning of 1994: Atrial fibrillation and atrial flutter were coded separately in ICD-8 (codes 427.93 and 427.94), but in ICD-10, atrial fibrillation and flutter have the same ICD code (I48). Therefore, we had to include atrial flutter in our study. CLINICAL SIGNIFICANCE Relevant ICD-8 and ICD-10 codes are listed in the Appendix. ● We do not know much about trend over

time in survival in stroke associated with The study was conducted in DenComorbidity atrial fibrillation. mark from January 1, 1980 to DeWe obtained data on previous or cember 31, 2002 (identification of ● In a national cohort study, a substantial simultaneous—if any— diagnoses incident stroke with previous or improvement in survival in stroke with of hyperthyroidism, diabetes and concomitant nonvalvular atrial fiatrial fibrillation was observed over a cardiovascular diseases (hypertenbrillation or flutter), and to Sepsion, ischemic heart disease, con23-year period. tember 10, 2004 (follow-up on vigestive heart failure, and heart tal status) in Denmark. During the ● Survival in stroke with atrial fibrillation valve disease) from the Danish study period the total population improved in both men and women. National Registry of Patients from increased from about 5.1 million 1977 to the end of 2002. to 5.4 million, and the population of interest for the present study, Validation of Diagnoses of Atrial Fibrillation, namely subjects aged 40-89 years, increased from 2.1 to 2.5 million. The general health and hospital care systems in Atrial Flutter and Stroke Denmark are no charge and nonprofit systems that are To validate the diagnoses, we studied 2 random samples of financed through taxes. 200 patients from the county of Aarhus, Denmark who had a diagnosis of atrial fibrillation or flutter or a diagnosis of Identification of Patients with an Incident stroke that occurred after a diagnosis of atrial fibrillation or flutter in the Danish National Registry of Patients during the Stroke Diagnosis Associated with a Diagnosis study period from 1980 to 2002. (Aarhus, Denmark has of Atrial Fibrillation approximately 500,000 inhabitants and 11 hospitals includWe identified all patients in Denmark, aged 40-89, with a ing 1 university hospital.) For atrial fibrillation or flutter, we first hospital diagnosis of stroke of any nature (ischemic or could retrieve 174 of 200 medical records (index hospitalhemorrhagic) in the Danish National Registry of Patients izations). In 172 of 174 patients, we found documentation from January 1, 1977 through December 31, 2002, or with for atrial fibrillation in an electrocardiogram, a printout from an outpatient hospital clinic diagnosis of stroke during the telemetry, a Holter recording, or an event recorder; this period from January 1, 1994 through December 31, 2002. A corresponds to a verification rate of 99%. primary, as well as a secondary diagnosis of stroke was For a diagnosis of stroke after a diagnosis of atrial fibrilincluded. Among those, we included patients with a diaglation or flutter, we could retrieve 164 of 200 medical nosis of nonvalvular atrial fibrillation that occurred before records (index hospitalizations). Among those, 159 patients or at the time of a diagnosis of stroke for the survival (97%) had a clinical diagnosis of stroke, defined according analysis. The date of the first stroke diagnosis (date of to the World Health Organization’s definition of stroke (ie, admission) was defined as the baseline date and was noted, an acute disturbance of focal or global cerebral function together with information about sex, age at stroke diagnosis, with symptoms lasting more than 24 hours or leading to and comorbid conditions, before or at baseline. To reduce death of presumed vascular origin).9 The mean modified the risk of including patients with prevalent stroke diagnosis, Rankin scale stroke severity score (scale from 0 to 5) was we excluded patients who had a diagnosis of stroke during 3.5 among 151 subjects that could be scored during the the period from 1977 to 1979. Patients with a diagnosis of validation process, and the mean Rankin scale score (score heart valve disease before or at baseline also were excluded. from 0 to 5) was 3.9 (calendar years 1980-1984), 3.9 (calThe Danish National Registry of Patients was established endar years 1985-1989), 4.0 (calendar years 1990-1994), in 1977 and records 99.4% of all nonpsychiatric hospital 3.2 (calendar years 1995-1999), and 3.3 (calendar years contacts in Denmark.8 Data include the civil registry num2000-2002). The subjects that were erroneously coded as ber, dates of admission and discharge, surgical procedures having stroke had transient ischemic attack (4 patients), and performed, and one or several diagnoses per discharge. The one patient had a pituitary gland tumor. Thus, the verificaDanish version of the International Classification of Distion rate of stroke was 97%. Among those, 96% were eases, 8th Revision (ICD-8) was used until 1993, and thereadmitted to the hospital within 30 days of the occurrence of after the national version of ICD-10 was used. The physi-

Frost et al

Stroke with Atrial Fibrillation

symptoms of stroke; the vast majority were admitted within 24 hours from onset of symptoms. Eighty-three of 164 patients with a diagnosis of stroke had a computed tomography cerebral scan or a magnetic resonance cerebral scan. Among those, 92% could be classified as having ischemic stroke and 8% as having hemorrhagic stroke. Ischemic stroke with a hemorrhagic transformation was categorized as stroke of ischemic nature.

Follow-up and Outcome

49 Table 1 Demographic and Clinical Data for 24,470 Men and Women with an Incident Hospital Diagnosis of Stroke and a Previous or Concomitant Diagnosis of Nonvalvular Atrial Fibrillation or Flutter in Denmark, 1980-2002 Characteristic Age (% of total) 40-49 50-59 60-69 70-79 80-89 Condition of comorbidity (% Diabetes Hyperthyroidism Hypertension Ischemic heart disease Congestive heart failure

Men (n ⫽ 11,554) 1.1 5.7 19.3 43.2 30.7 of total) 15.3 2.4 25.1 50.6 30.3

Women (n ⫽ 12,916) 0.3 1.6 9.5 37.6 51.0

Patients were followed in the Danish Civil Registration System until September 10, 2004 (emigration and vital status). This Civil Registration System has electronic records of all changes in status for the Danish population since 1968, including change in address, date of emigration, and date of death. The outcome of interest in this study was time to death after stroke. We linked records from the different registries by using the civil registration number, a unique 10-digit code given to each individual having or having had an address in Denmark since April 1968.

lege Station, Tex). Our study was approved by the Danish Data Protection Agency.

Statistical Analysis

RESULTS

We used a multivariate Cox regression model separately in men and women with forced entry of age (10-year age group), history of diabetes (no, yes), history of hyperthyroidism (no, yes), history of hypertension (no, yes), history of ischemic heart disease (no, yes), history of congestive heart failure (no, yes), and calendar period of diagnosis of stroke. (Categories of calendar year intervals were: 19801984, 1985-1989, 1990-1994, 1995-1999, and 2000-2002.) We performed supplementary analyses by adding product terms, which included age and sex, and age and comorbid conditions, to test for interaction. We also assessed effect modification by stratified analyses (stratification by age 40-74 years vs age 75-89 years, and by conditions of comorbidity). The relevance of a variable in the model was further assessed by the change-in-estimate method.10 To adjust for the general mortality in the Danish population, we introduced a transformed version of observation time into Cox regression analysis, namely, the patients’ expected normal risk of dying during the time of observation, computed from the observed mortality in the general Danish population by sex, age and calendar period.11 To try to adjust for trend in severity of stroke, we assigned the mean modified Rankin scale score found in the validation series in each of the calendar year periods to all individuals within the same calendar year period. The proportional hazards assumption in the Cox models was evaluated using graphical assessment and was found appropriate in all models. We calculated 95% confidence intervals (CI) throughout the analyses. Trends in age and comorbidity by calendar period were evaluated by using Spearman nonparametric correlation analysis. All reported P values are 2-sided. We used SPSS statistical software version 11.5 (SPSS Inc., Chicago, Ill), and STATA statistical software version 7 (Stata Corp, Col-

16.2 9.3 28.2 47.3 29.9

During the study period from 1980 to 2002, we identified 24,470 subjects, aged 40-89 years, with an incident hospital diagnosis of stroke and a previous or concomitant diagnosis of nonvalvular atrial fibrillation or flutter; 11,554 were men and 12,916 were women. The age distribution and the proportion of men and women with a previous or a concomitant diagnosis of an endocrine or cardiovascular disease at baseline are shown in Table 1.

Trend in Age and Diagnosed Comorbidity Table 2 shows trends in age and comorbid conditions in patients with incident stroke by calendar year period. The number of patients diagnosed with stroke and atrial fibrillation increased over the study period. The mean age at time of stroke increased by 0.3 years from 73.9 years to 74.2 years in men, and by 2.0 years from 77.1 years to 79.1 years in women. Among subjects with stroke, the proportion diagnosed with diabetes decreased in women, but not in men. More patients had a diagnosis of hypertension, but fewer men and women had a diagnosis of ischemic heart disease and congestive heart failure during the study period.

Trend in All-Cause Mortality During 34,405 years of observation (mean 3.0 years), 9237 men died, and during 35,381 years of observation (mean 2.7 years), 10,827 women died. The age-adjusted cumulative survival (Kaplan-Meier estimate) improved by calendar period in both men and women (Figure 1). We observed a modest excess in risk of mortality in men relative to women, adjusted hazard ratio 1.07 (95% CI, 1.04-1.10) (adjusted by 10-year age group and conditions of comorbidity), but when we adjusted for mortality trend in the general population, we

50

The American Journal of Medicine, Vol 120, No 1, January 2007

Table 2 Trends in Age and Diagnosed Comorbidity by Calendar Period in 11,554 Men and 12,916 Women with an Incident Hospital Diagnosis of Stroke and a Previous or Concomitant Hospital Diagnosis of Nonvalvular Atrial Fibrillation or Flutter in Denmark, 1980-2002 Calendar Year Period

Characteristic

1980-1984

Men Number of subjects 1469 Age (mean ⫾ SD) 73.9 ⫾ 8.5 History and/or presence of (% of total) Diabetes 14.0 Hyperthyroidism 2.2 Hypertension 18.7 Ischemic heart disease 68.5 Congestive heart failure 29.3 Women Number of subjects 1810 Age (mean ⫾ SD) 77.1 ⫾ 7.3 History and/or presence of (% of total) Diabetes 18.4 Hyperthyroidism 6.9 Hypertension 26.1 Ischemic heart disease 66.0 Congestive heart failure 29.4

1985-1989

1990-1994

1995-1999

2000-2002

P Value for Correlation*

1923 74.1 ⫾ 8.6

2518 74.4 ⫾ 8.4

3149 74.6 ⫾ 8.9

2495 74.2 ⫾ 9.5

⬍.001

15.4 2.3 25.3 61.5 35.7 2227 78.2 ⫾ 7.1 18.6 9.2 27.8 58.9 32.7

14.1 2.6 24.3 49.6 32.8 2837 78.5 ⫾ 7.1 16.0 10.0 25.7 47.0 32.3

15.6 2.4 24.4 43.4 28.6 3439 79.0 ⫾ 7.3 14.4 9.6 27.0 39.3 29.2

16.6 2.5 30.6 41.9 26.1 2603 79.1 ⫾ 7.3 15.3 10.1 34.2 35.3 25.9

.12 .55 ⬍.001 ⬍.001 ⬍.001 ⬍.001 ⬍.001 .003 ⬍.001 ⬍.001 ⬍.001

*Spearman nonparametric correlation.

observed that the excess risk of mortality was reduced among men relative to women, adjusted hazard ratio 0.87 (95% CI, 0.82-0.87). The relative magnitude of the improvement in survival by calendar period was estimated by Cox regression analyses, and the results appear in Tables 3 and 4. The adjusted hazard ratio for mortality decreased by 35% (95% CI, 29%39%) in men and 31% (95% CI, 26%-36%) in women during the study period (Table 4). Adjustment for assigned mean Rankin stroke scale score within the calendar year periods did not change the estimated hazard ratio by calendar year period.

DISCUSSION We observed a marked advance in survival after stroke associated with nonvalvular atrial fibrillation or flutter. Several improvements have been introduced in medical treatment during the study period, such as oral anticoagulation, potent lipid lowering drugs, more intensive treatment of hypertension and heart failure, together with the establishment of facilities for thrombolysis, and stroke rehabilitation units. On the other hand, equipment for cerebral computed tomography or magnetic resonance scan has been widely disseminated in our country during the study period, and this may have led to diagnosis of many minor strokes, which would not have been diagnosed during the early part of the study period. This would cause us to overestimate the improvement in survival by calendar period. However, controlling for stroke severity did not change the estimates of improvement in survival.

Consistency with Other Findings We found that an increasing number of patients with stroke associated with atrial fibrillation were reported to the Danish National Registry of Patients during the study period. This is in accordance with findings in Oxfordshire, UK, where a population-based cohort study with ascertainment of incident strokes by referral of any patients with a suspected cerebrovascular event from general practice to the Oxford Vascular Study for clinical assessment showed that the proportion of patients with known previous atrial fibrillation among patients with incident stroke increased from 9.6% in the period from 1981-1984 to 16.8% in the period from 2002-2004.12 We do not know if this increase in the number (or proportion) of patients with stroke associated with atrial fibrillation reflects a true increase in the incidence of atrial fibrillation at the population level, or if it mirrors the increasing interest in atrial fibrillation among health professionals and lay persons.13 We observed a better survival after stroke with atrial fibrillation over time. This is in contrast to findings in Olmsted County, Minnesota, where a community-based cohort study, which ascertained information on atrial fibrillation and stroke from recordings in administrative databases in the Mayo Clinic, showed that the relative mortality hazard in those who sustained a stroke with atrial fibrillation did not change over time in the period from 1980-2000.7 However, the results found in Olmsted County may be affected by lack of statistical power, because the number of subjects with the outcome of interest determines the statistical power. The number of subjects with stroke followed by

Frost et al

Stroke with Atrial Fibrillation

51

Men 1.00

0.75

2000-02 1995-99 1990-94 1985-89 1980-84

0.50

0.25

0.00 0

1

Years

2

3

4

Women 1.00

0.75 2000-02 1995-99 1990-94 1985-89 1980-84

0.50

0.25

0.00 0

1

Years

2

3

4

Figure 1 Cumulative survival (Kaplan-Meier estimate) adjusted to age 75 years in men and women with stroke associated with nonvalvular atrial fibrillation by calendar period of diagnosis of stroke.

death in the Olmsted County study was 363, whereas the number of subjects with stroke followed by death exceeded 20,000 in our study. We believe that progress in medical treatment and rehabilitation of patients with stroke should translate into a better chance of surviving a stroke. For example, the user rate of oral anticoagulant therapy has increased during the study period in our country,14,15 and an observational study has shown that oral anticoagulation in patients with atrial fibrillation reduces not only the frequency of ischemic stroke but also its severity and the risk of death from

stroke.16 However, such an observational study should be interpreted with caution due to risk of confounding by indication. Patients with stroke associated with atrial fibrillation are at a substantially increased risk of another stroke, the risk of a second stroke being approximately 10% annually.17 Treatment with oral anticoagulation in patients with stroke associated with atrial fibrillation is very effective in the prevention of another stroke, and the user rate of oral anticoagulation after stroke with atrial fibrillation has increased in our country.15

52

The American Journal of Medicine, Vol 120, No 1, January 2007

Table 3 Effect of Calendar Period of Stroke on Risk of Dying in 24,470 Men and Women with an Incident Hospital Diagnosis of Stroke and a Previous or Concomitant Hospital Diagnosis of Nonvalvular Atrial Fibrillation or Flutter in Denmark, 1980 –2002 Men (n ⫽ 11,554) Hazard Ratio (95% CI) Model 1.Unadjusted estimates 1980-1984 (reference) 1985-1989 1990-1994 1995-1999 2000-2002

1.00 0.92 (0.85-0.99) 0.85 (0.79-0.91) 0.73 (0.68-0.78) 0.63 (0.58-0.68) P value for trend ⬍.001 Model 2.Adjustment for 10-year age group and conditions of comorbidity* 1980-1984 (reference) 1.00 1985-1989 0.91 (0.85-0.97) 1990-1994 0.83 (0.77-0.88) 1995-1999 0.72 (0.67-0.77) 2000-2002 0.62 (0.57-0.67) P value for trend ⬍.001

Women (n ⫽ 12,916) Hazard Ratio (95% CI) 1.00 0.98 (0.92-1.04) 0.91 (0.85-0.97) 0.82 (0.77-0.87) 0.73 (0.68-0.79) P value for trend ⬍.001 1.00 0.94 (0.88-1.00) 0.87 (0.82-0.92) 0.78 (0.73-0.83) 0.70 (0.65-0.75) P value for trend ⬍.001

CI ⫽ confidence interval. *Adjusted for 10-year age group, hypertension, diabetes, hyperthyroidism, ischemic heart disease, and congestive heart failure.

Strengths and Limitations The major advantages of our study derive from the population-based design, the uniformly organized health care system, the identification of incident cases, the validation of a random sample of diagnoses, completeness of follow-up, and the large number of outcomes. Limitations arise from possible errors in assigning stroke severity score and coding of diagnoses. We assigned a stroke severity score to the total study cohort based on a relatively small sample of patients, so this part of our study should be read with caution. Misclassification of comorbid conditions, especially hypertension, may have occurred,18 and we lacked clinical details such as smoking status and severity of comorbid conditions. An earlier study has, together with the internal validation of diagnoses in this study, documented a high validity (ie, a high predictive value) of a diagnosis of atrial fibrillation or flutter in the Danish National Registry of Patients.19 Furthermore, as stroke severity is more pronounced in patients

with atrial fibrillation, we find it very likely that those with a diagnosis of stroke actually had a stroke, and this was documented by the internal validation of stroke diagnoses in our study. A priori, we decided not to exclude patients with hemorrhagic stroke, because many strokes are reported to the Danish National Hospital Registry as unspecified strokes. However, the nature of stroke in patients with atrial fibrillation is predominantly embolic.17,20-25

Future Perspective The health authorities in our country have recently introduced standards for treatment of stroke and a database to which all patients with stroke have to be reported together with information on comorbid medical conditions, stroke severity score, and treatment. Hopefully, those standardizing and monitoring activities will lead to better treatment and improved survival in patients with stroke, but any assessment of the progress in survival in stroke victims caused

Table 4 Effect of Calendar Period of Stroke on Risk of Dying in 24,470 Men and Women with an Incident Hospital Diagnosis of Stroke and a Previous or Concomitant Hospital Diagnosis of Nonvalvular Atrial Fibrillation or Flutter in Denmark, 1980 –2002 with Adjustment for Mortality trend in the Danish Population Men (n ⫽ 11,554) Hazard Ratio (95% CI)

Women (n ⫽ 12,916) Hazard Ratio (95% CI)

Model 3. Adjustment for 10-year age group, conditions of comorbidity and general mortality in the Danish population* 1980-1984 (reference) 1.00 1.00 1985-1989 0.93 (0.87-1.00) 0.93 (0.88-0.99) 1990-1994 0.86 (0.80-0.92) 0.87 (0.82-0.92) 1995-1999 0.76 (0.71-0.81) 0.78 (0.73-0.83) 2000-2002 0.65 (0.61-0.71) 0.69 (0.64-0.74) P value for trend ⬍.001 P value for trend ⬍.001 CI ⫽ confidence interval. *Adjusted for 10-year age group, hypertension, diabetes, hyperthyroidism, ischemic heart disease, congestive heart failure, and mortality trend in the Danish population.

Frost et al

Stroke with Atrial Fibrillation

by monitoring activities initiated by health authorities should take into account that improvements in survival in stroke victims are already being seen.

53

8.

9.

ACKNOWLEDGMENTS This work was supported in part by the Clinical Institute, Aarhus University, and Western Danish Research Forum for Health Sciences.

10. 11.

APPENDIX World Health Organization International Disease Classification (ICD) Codes, ICD-8 and ICD-10 used in the present analysis. Atrial fibrillation and atrial flutter 427.93, 427.94, I48 Hyperthyreoidism 242, E05 Hypertension 400-404, 410.09, 411.09, 412.09, 413.09, 414.09, 435.09, 437.00, 437.01, 437.08, 437.09, 438.09, I10-I15 Diabetes 249, 250, E10-E14 Ischemic heart disease 410-414, I20-I25 Congestive heart failure 425.99, 427.09, 427.10, 427.11, 427.19, 427.99, 428.99, I50 Mitral and/or aortic valve disease 394-396, I05, I06, I08, I34, I35 Stroke 430-434, 436, I60-I64

12.

13.

14.

15.

16.

17.

18.

19. 20.

References 1. Sarti C, Rastenyte D, Cepaitis Z, Tuomilehto J. International trends in mortality from stroke, 1968 to 1994. Stroke. 2000;31:1588-1601. 2. Feigin VL, Lawes CM, Bennett DA, Anderson CS. Stroke epidemiology: a review of population-based studies of incidence, prevalence, and case-fatality in the late 20th century. Lancet Neurol. 2003;2:43-53. 3. Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation: a major contributor to stroke in the elderly. The Framingham Study. Arch Intern Med. 1987;147:1561-1564. 4. Lin HJ, Wolf PA, Kelly-Hayes M, et al. Stroke severity in atrial fibrillation. The Framingham Study. Stroke. 1996;27:1760-1764. 5. Dulli DA, Stanko H, Levine RL. Atrial fibrillation is associated with severe acute ischemic stroke. Neuroepidemiology. 2003;22:118-123. 6. Marini C, De Santis F, Sacco S, et al. Contribution of atrial fibrillation to incidence and outcome of ischemic stroke: results from a population-based study. Stroke. 2005;36:1115-1119. 7. Miyasaka Y, Barnes ME, Gersh BJ, et al. Time trends of ischemic stroke incidence and mortality in patients diagnosed with first atrial

21.

22. 23.

24.

25.

fibrillation in 1980 to 2000: report of a community-based study. Stroke. 2005;36:2362-2366. Andersen TF, Madsen M, Jorgensen J, Mellemkjoer L, Olsen JH. The Danish National Hospital Register. A valuable source of data for modern health sciences. Dan Med Bull. 1999;46:263-268. National Institute of Neurological Disorders and Stroke Ad Hoc Committee. Special report from the National Institute of Neurological Disorders and Stroke. Classification of cerebrovascular diseases: III. Stroke. 1990;21:637-676. Greenland S. Modeling and variable selection in epidemiologic analysis. Am J Public Health. 1989;79:340-349. Kjoller E, Mortensen LS, Larsen ME. Evaluation of prognosis. Dan Med Bull. 1978;25:238-242. Rothwell PM, Coull AJ, Giles MF, et al. Change in stroke incidence, mortality, case-fatality, severity, and risk factors in Oxfordshire, UK from 1981 to 2004 (Oxford Vascular Study). Lancet. 2004;363:19251933. Frost L, Vestergaard P, Mosekilde L, Mortensen LS. Trends in incidence and mortality in the hospital diagnosis of atrial fibrillation or flutter in Denmark, 1980-1999. Int J Cardiol.2005;103:78-84. Frost L, Johnsen SP, Pedersen L, Toft E, Husted S, Sorensen HT. Atrial fibrillation or flutter and stroke: a Danish population-based study of the effectiveness of oral anticoagulation in clinical practice. J Intern Med. 2002;252:64-69. Friberg J, Gislason GH, Gadsboll N, et al. Temporal trends in the prescription of vitamin K antagonists in patients with atrial fibrillation. J Intern Med. 2006;259:173-178. Hylek EM, Go AS, Chang Y, et al. Effect of intensity of oral anticoagulation on stroke severity and mortality in atrial fibrillation. N Engl J Med. 2003;349:1019-1026. Secondary prevention in non-rheumatic atrial fibrillation after transient ischaemic attack or minor stroke. EAFT (European Atrial Fibrillation Trial) Study Group. Lancet. 1993;342:1255-1262. Nielsen HW, Tuchsen F, Jensen MV. [Validity of the diagnosis “essential hypertension” in the National Patient Registry]. Ugeskr Laeger. 1996;158:163-167. Frost L, Vestergaard P. Alcohol and risk of atrial fibrillation or flutter: a cohort study. Arch Intern Med. 2004;164:1993-1998. Petersen P, Boysen G, Godtfredsen J, Andersen ED, Andersen B. Placebo-controlled, randomised trial of warfarin and aspirin for prevention of thromboembolic complications in chronic atrial fibrillation. The Copenhagen AFASAK study. Lancet. 1989;1:175-179. The effect of low-dose warfarin on the risk of stroke in patients with nonrheumatic atrial fibrillation. The Boston Area Anticoagulation Trial for Atrial Fibrillation Investigators. N Engl J Med. 1990;323: 1505-1511. Stroke Prevention in Atrial Fibrillation Study. Final results. Circulation. 1991;84:527-539. Connolly SJ, Laupacis A, Gent M, Roberts RS, Cairns JA, Joyner C. Canadian Atrial Fibrillation Anticoagulation (CAFA) Study. J Am Coll Cardiol. 1991;18:349-355. Ezekowitz MD, Bridgers SL, James KE, et al. Warfarin in the prevention of stroke associated with nonrheumatic atrial fibrillation. Veterans Affairs Stroke Prevention in Nonrheumatic Atrial Fibrillation Investigators. N Engl J Med. 1992;327:1406-1412. Warfarin versus aspirin for prevention of thromboembolism in atrial fibrillation: Stroke Prevention in Atrial Fibrillation II Study. Lancet. 1994;343:687-691.