a population-based study of the long-term risks associated with atrial fibrillation: 20-year...

CLINICAL STUDIES

A Population-Based Study of the Long-term RisksAssociated with Atrial Fibrillation: 20-Year

Follow-up of the Renfrew/Paisley Study

Simon Stewart, PhD, Carole L. Hart, MA, David J. Hole, MSc, John J. V. McMurray, MD

PURPOSE: To describe the effect of atrial fibrillation on long-term morbidity and mortality.SUBJECTS AND METHODS: The Renfrew/Paisley Study sur-veyed 7052 men and 8354 women aged 45– 64 years between1972 and 1976. All hospitalizations and deaths occurring duringthe subsequent 20 years were analyzed by the presence or ab-sence of atrial fibrillation at baseline. Lone atrial fibrillation wasdefined in the absence of other cardiovascular signs or symp-toms. Cox proportional hazards models were used to adjust forage and cardiovascular conditions.RESULTS: After 20 years, 42 (89%) of the 47 women with atrialfibrillation had a cardiovascular event (death or hospitaliza-tion), compared with 2276 (27%) of the 8307 women withoutthis arrhythmia. Among men, 35 (66%) of 53 with atrial fibril-lation had an event, compared with 3151 (45%) of 6999 withoutatrial fibrillation. In women, atrial fibrillation was an indepen-

dent predictor of cardiovascular events (rate ratio [RR] � 3.0;95% confidence interval [CI]: 2.1– 4.2), fatal or nonfatal strokes(RR � 3.2; 95% CI: 1.0 –5.0), and heart failure (RR � 3.4; 95%CI: 1.9 – 6.2). The rate ratios among men were 1.8 (95% CI:1.3–2.5) for cardiovascular events, 2.5 (95% CI: 1.3– 4.8) forstrokes, and 3.4 (95% CI: 1.7– 6.8) for heart failure. Atrial fibril-lation was an independent predictor of all-cause mortality inwomen (RR � 2.2; 95% CI: 1.5–3.2) and men (RR � 1.5; 95%CI: 1.2–2.2). However, lone atrial fibrillation (which occurredin 15 subjects) was not associated with a statistically significantincrease in either cardiovascular events (RR � 1.5; 95% CI:0.6 –3.6) or mortality (RR � 1.8; 95% CI: 0.9 –3.8).CONCLUSION: Atrial fibrillation is associated with an in-creased long-term risk of stroke, heart failure, and all-causemortality, especially in women. Am J Med. 2002;113:359 –364.©2002 by Excerpta Medica, Inc.

Atrial fibrillation is a growing public health prob-lem (1–3). The Framingham Heart Study and sev-eral other studies have reported that atrial fibril-

lation was an important cause of cardiovascular morbid-ity, especially stroke (4 –11). There are, however, fewerstudies of the long-term effects of atrial fibrillation onother types of cardiovascular morbidity and on all-causemortality in the general population (1). Consequently,we examined the effect of atrial fibrillation on cardiovas-cular hospitalizations and deaths during 20-year fol-low-up of 15,406 initially middle-aged men and womenwho were first screened between 1972 and 1976 (12). We

have already reported the prevalence and incidence ofatrial fibrillation in the cohort (13).

SUBJECTS AND METHODS

Sample and Baseline DataBetween 1972 and 1976, 7052 men and 8354 women, rep-resenting 80% of subjects aged 45– 64 years residing inthe towns of Renfrew and Paisley, in the west of Scotland,took part in this epidemiologic study (12). Each subject’sdemographic profile and cardiorespiratory health statuswere documented (12,13). Angina pectoris (classified asnone, possible, or definite by the Rose angina question-naire) and chronic bronchitis (determined by the Medi-cal Research Council’s chronic bronchitis questionnaire)were noted (14,15). Past and current medical history andrisk factors for cardiorespiratory disease were docu-mented. Blood pressure, and height and weight (used tocalculate body mass index in kg/m2), were measured. Thecardiothoracic ratio (based on chest radiographs) andforced expiratory volume in 1 second (FEV1) were mea-sured. Cardiomegaly was defined as a cardiothoracic ratio�0.55. An adjusted FEV1 was calculated as a percentageof the “expected” FEV1 (derived from a linear regressionequation of age and height for men and women separately

From the Division of Health Sciences at the University of South Austra-lia (SS), Adelaide, Australia; and the Department of Public Health(CLH), West of Scotland Cancer Surveillance Unit (DJH); and the Clin-ical Research Initiative in Heart Failure (JJVM), University of Glasgow,Glasgow, United Kingdom.

Professor Stewart is supported by the National Heart Foundation ofAustralia. Carole Hart is supported by grants from Chest, Heart, andStroke Scotland and The Stroke Association, London, United Kingdom.

Requests for reprints should be addressed to John J. V. McMurray,MD, Department of Cardiology, University of Glasgow, WesternInfirmary, Glasgow G11 6NT, United Kingdom, or at [email protected].

Manuscript submitted October 24, 2001, and accepted in revisedform May 13, 2002.

©2002 by Excerpta Medica, Inc. 0002-9343/02/$–see front matter 359All rights reserved. PII S0002-9343(02)01236-6

from a healthy subset of the sample who were nonsmok-ers and had no respiratory symptoms) and the actualFEV1 (16). Plasma cholesterol and glucose concentra-tions were measured in a nonfasting blood sample. A six-lead electrocardiograph (ECG) was also obtained andcoded (17). These ECGs were used to identify all subjectswith atrial fibrillation (Minnesota code 8.3) and, withinthe constraints of interpreting data from only six leads,other abnormalities indicative of past myocardial infarc-tion, ST segment changes (possibly indicating myocardialischemia), left bundle branch block, and left ventricularhypertrophy.

Study Follow-upThe Scottish Morbidity Record Scheme (18) was used toretrieve details of all hospital discharges (according to theeighth [a small number of initial episodes] and ninth revi-sions of the World Health Organization International Clas-sification of Diseases) during the 20 years after initial screen-ing (i.e., surviving subjects were aged 65–84 years) (19). Wenoted the occurrence and timing of admissions for myocar-dial infarction, stroke, and heart failure. These three diag-noses, in combination with a number of other less com-monly recorded diagnoses (e.g., angina pectoris and tran-sient cerebrovascular insufficiency), are collectively calledcardiovascular hospitalizations. A recent audit showed thatthese data are approximately 90% accurate in identifying thecorrect discharge diagnosis (20). Deaths and their certifiedcause were obtained from the National Health Service Cen-tral Register for the same period.

Statistical AnalysisAnalyses were performed using SPSS for Windows, ver-sion 10 (Chicago, Illinois). Statistical significance was ac-cepted at the level of P �0.05 (two-sided) for all analyses.

Initial comparisons of the baseline characteristics ofthis cohort involved the chi-squared test, Student t test,and analysis of variance with post-hoc testing of two-waycomparisons, where appropriate. Morbidity and mortal-ity data were censored at 20 years postscreening. The ef-fects of atrial fibrillation at baseline on study endpointswere estimated using proportional hazards models atfirst, with adjustment only for the age of subjects to gen-erate age-adjusted rate ratios (RR) and 95% confidenceintervals (CI). Plots of the cumulative risk of a majorcardiovascular event (hospitalization or death) and all-cause mortality, according to the mean age of subjects atbaseline, were also generated in this way, with the meanage of subjects at baseline held constant. Graphical com-parisons of the hazard functions for these two endpointsconfirmed that the proportional hazard assumption wasmet. To estimate the independent contribution of atrialfibrillation, additional models were constructed, includ-ing all baseline variables associated with an outcome at P�0.1, with stepwise rejection of those variables when P

was �0.05. When appropriate, continuous data were cat-egorized into quartiles, and rate ratios were calculated forthe lowest versus highest quartiles. Subjects without ra-diographic data (4%) were excluded from these models.Blood glucose concentrations were missing in approxi-mately one third of patients. As the hazard models did notdiffer substantially in the presence or absence of thesedata, rate ratios were generated in the 6693 men and 7990women with complete data records (ignoring blood glu-cose concentrations). Separate models using the same co-variates but only including those subjects with blood glu-cose measurements were used to generate rate ratios forblood glucose concentrations �7.0 mmol/L. We testedfor an interaction (among atrial fibrillation, sex, and out-come) using a multiplicative term.

RESULTS

The 100 subjects with atrial fibrillation at baseline (1972–1976) were older and had more cardiovascular disease,ECG abnormalities, and a lower mean adjusted FEV1

than did subjects without atrial fibrillation (Table 1). Atotal of 15 subjects (10 men) had atrial fibrillation in theabsence of any concurrent signs and symptoms of cardio-vascular disease (“lone” atrial fibrillation). There wereseveral significant differences between subjects with loneatrial fibrillation, the remainder of those with atrial fibril-lation, and men and women without atrial fibrillation.Mean (�SD) blood pressure (systolic/diastolic) was124/78 (�11/10) mm Hg in those with lone atrial fibril-lation, compared with 153/88 (�24/13) mm Hg in thosewith other forms of atrial fibrillation (P �0.001 systolic,P � 0.01 diastolic) and 149/86 (�24/13) mm Hg in theremaining subjects (P �0.001 systolic, P � 0.04 diastol-ic). Adjusted FEV1% was 93% � 18% in those with loneatrial fibrillation, compared with 71% � 23% in thosewith other forms of atrial fibrillation (P � 0.002) and91% � 23% in the remaining patients (P � 0.72).

Cardiovascular Events and All-Cause Mortalityduring 20-Year Follow-upOverall, women with atrial fibrillation had an approxi-mately fivefold increase in the risk of any cardiovascularevent (hospitalization or death) compared with an ap-proximately twofold increase in risk in men (Table 2, Fig-ure 1). In both men and women, the major excess incardiovascular events was due to stroke and heart failure.Of the 47 women with atrial fibrillation at baseline, 42(89%; 95% CI: 77% to 97%) had a cardiovascular eventduring the 20-year follow-up, compared with 2276 ofthe 8307 (27%; 95% CI: 26% to 28%) women withoutatrial fibrillation (P �0.001). Of the 53 men with atrialfibrillation, 35 (66%; 95% CI: 52% to 79%) had a car-

Long-term Risks Associated with Atrial Fibrillation/Stewart et al

360 October 1, 2002 THE AMERICAN JOURNAL OF MEDICINE� Volume 113

diovascular event, compared with 3151 of the 6999(45%; 95% CI: 44% to 46%) men without atrial fibril-lation (P �0.01).

As expected, proportionately more women (71%) thanmen (56%) were alive after 20 years of follow-up. How-

ever, among subjects with atrial fibrillation, a similar pro-portion of women (86% [37/43]; 95% CI: 74% to 93%)and men (72% [38/53]; 95% CI: 60% to 85%) died within20 years (both P �0.001 compared with men and womenwithout atrial fibrillation).

Table 1. Characteristics of the Renfrew/Paisley Cohort, by the Presence or Absence of Atrial Fibrillation at Baseline

Characteristic

Men(n � 7052)

Women(n � 8354)

AtrialFibrillation

(n � 53)

No AtrialFibrillation(n � 6999)

PValue

AtrialFibrillation

(n � 47)


PValue

Mean � SD or Number (%) Mean � SD or Number (%)

Age (years) 57 � 6 54 � 4 �0.001 56 � 6 54 � 6 0.03Cardiovascular risk factors

Past history of stroke 3 (6) 90 (1) 0.03 6 (13) 98 (0.1) �0.001Any form of chest pain 7 (13) 655 (9) 0.33 13 (28) 786 (10) �0.001Current or ex-smoker 40 (75) 5629 (80) 0.86 27 (57) 4494 (54) 0.66Serum cholesterol (mmol/L) 6.8 � 0.9 5.8 � 1.0 0.01 6.4 � 0.9 5.9 � 1.0 0.01

Clinical profileBody mass index (kg/m2) 25.9 � 3.4 26.4 � 3.9 0.20 25.8 � 4.5 24.6 � 4.4 0.07Systolic blood pressure (mm Hg) 149 � 23 146 � 23 0.37 154 � 23 150 � 13 0.34Diastolic blood pressure (mm Hg) 86 � 13 84 � 10 0.34 90 � 15 85 � 14 0.02Cardiothoracic ratio 0.53 � 0.04 0.47 � 0.05 �0.001 0.58 � 0.04 0.48 � 0.05 �0.001Blood glucose (mmol/L) 5.2 � 1.5 5.2 � 1.3 0.78 6.5 � 3.4 5.1 � 1.4 �0.001Adjusted FEV1(%) 74 � 24 87 � 23 �0.001 71 � 23 93 � 24 �0.001Chronic bronchitis 10 (19) 399 (6) �0.001 3 (7) 341 (4) 0.31Pathological Q waves on ECG 5 (9) 233 (3) 0.03 0 163 (2) 1.0ST-segment changes on ECG 11 (21) 344 (5) �0.001 20 (43) 495 (6) �0.001Left bundle branch block on ECG 10 (21) 533 (8) �0.001 16 (34) 565 (7) �0.001

ECG � electrocardiogram; FEV1 � forced expiratory volume in 1 second.

Table 2. Cardiovascular Morbidity and Mortality during 20-Year Follow-up, by Sex and by the Presence or Absence of AtrialFibrillation at Baseline

Outcome

Men (n � 7052) Women (n � 8354)

AtrialFibrillation

(n � 53)


Rate Ratio(95% Confidence

Interval)*

AtrialFibrillation

(n � 47)


Rate Ratio(95% Confidence

Interval)*

Number of Events (%) Number of Events (%)

Cardiovascular hospitalization 26 (49) 2356 (34) 1.9 (1.0–3.4) 32 (68) 2132 (26) 3.1 (1.6–6.0)Acute myocardial infarction 5 (9) 902 (13) 0.4 (0.1–1.4) 1 (2) 601 (7) 0.6 (0.1–4.0)Stroke 6 (11) 452 (6) 2.2 (0.8–5.9) 9 (19) 531 (6) 6.5 (3.1–13.8)Heart failure 5 (9) 328 (5) 3.2 (1.4–7.1) 11 (23) 289 (4) 11.2 (6.1–20.5)

Cardiovascular death 26 (49) 1752 (25) 2.09 (1.1–4.1) 33 (70) 1281 (15) 5.8 (4.1–8.1)Myocardial infarction 11 (21) 1154 (16) 1.5 (0.8–6.1) 5 (11) 742 (9) 1.8 (0.7–4.3)Stroke 6 (11) 298 (4) 2.7 (1.2–6.1) 9 (19) 379 (5) 5.4 (2.9–12.3)Heart failure 2 (4) 52 (1) 5.4 (1.3–22) 1 (2) 46 (1) 4.7 (0.7–34.4)

Cardiovascular death or hospitalization 35 (66) 3151 (45) 1.8 (1.2–2.5) 42 (89) 2276 (27) 4.9 (3.6–6.6)Myocardial infarction 11 (21) 1161 (17) 1.4 (0.8–2.4) 7 (15) 825 (10) 1.6 (0.7–3.7)Stroke 9 (17) 596 (9) 2.1 (1.1–4.3) 14 (30) 697 (8) 5.5 (3.3–9.2)Heart failure 6 (11) 356 (5) 3.0 (1.3–6.7) 12 (26) 326 (4) 10.9 (6.2–19.3)

* Adjusted for age.


October 1, 2002 THE AMERICAN JOURNAL OF MEDICINE� Volume 113 361

Independent Predictors of Cardiovascular Eventsand All-Cause MortalityIn multivariate-adjusted models, atrial fibrillation wasindependently associated with cardiovascular events andall-cause mortality (Table 3). Atrial fibrillation was alsoassociated with cardiovascular mortality in women (RR� 2.8; 95% CI: 1.9 – 4.2) and men (RR � 1.8; 95% CI:1.3–2.8). The association between atrial fibrillation and

cardiovascular events was stronger in women than inmen (P for interaction � 0.001).

Lone Atrial Fibrillation and OutcomeLone atrial fibrillation conferred a nonsignificant 1.5-fold(95% CI: 0.6 –3.6) increased risk of a cardiovascular event(and a 1.8-fold [95% CI: 0.9 –3.8] risk of death), com-pared with the 2.0-fold (95% CI: 1.5–2.8) increased risk

Figure 1. Age-adjusted, sex-specific cardiovascular hospitalizations or deaths during 20-year follow-up, by the presence (n � 100)or absence (n � 15,306) of atrial fibrillation at baseline.

Table 3. Independent Predictors of Cardiovascular Events (Death or Hospitalization) and All-Cause Mortality during 20-YearFollow-up of 6693 Men and 7990 Women in the Renfrew/Paisley Cohort

Characteristic

Cardiovascular Event Stroke Heart Failure All-Cause Mortality

Men Women Men Women Men Women Men Women

Multivariate-Adjusted Rate Ratio (95% Confidence Interval)

Atrial fibrillation 1.8 (1.3–2.5) 3.0 (2.1–4.2) 2.5 (1.3–4.8) 3.2 (1.0–5.0) 3.4 (1.7–6.8) 3.4 (1.9–6.2) 1.5 (1.2–2.2) 2.2 (1.5–3.2)Age* 2.2 (2.0–2.4) 2.2 (2.0–2.5) 4.0 (3.1–5.1) 3.2 (2.3–4.3) 3.2 (2.2–4.5) 5.0 (3.9–6.4) 3.6 (3.2–4.0) 3.4 (3.0–4.0)History of chest pain 1.4 (1.3–1.5) 1.4 (1.3–1.6) — 1.4 (1.1–1.7) 1.6 (1.3–2.1) 1.2 (1.0–1.4) 1.3 (1.2–1.4) 1.2 (1.1–1.3)History of smoking 1.5 (1.3–1.6) 1.5 (1.4–1.6) 1.3 (1.0–1.6) 1.4 (1.1–1.9) 1.4 (1.1–1.8) 1.7 (1.5–2.0) 1.6 (1.4–1.8) 1.5 (1.4–1.7)History of stroke 1.6 (1.2–2.0) 2.0 (1.6–2.7) 2.2 (1.3–3.7) 2.3 (1.2–4.1) 2.6 (1.4–4.8) 3.1 (2.0–4.7) 1.5 (1.2–2.0) 1.8 (1.4–2.4)Q waves on ECG 1.4 (1.1–1.6) 1.3 (1.0–1.7) — — — — 1.2 (1.0–1.5) 1.4 (1.1–1.8)ST-segment changes (ECG) 1.3 (1.1–1.6) 1.2 (1.0–1.5) 2.0 (1.5–2.7) — 1.6 (1.1–2.5) 1.7 (1.3–2.1) — 1.3 (1.1–1.6)Left bundle branch block 1.5 (1.2–1.7) 1.4 (1.2–1.6) — 2.1 (1.6–2.9) 1.6 (1.0–2.4) — 1.5 (1.3–1.7) —Systolic blood pressure† 1.5 (1.4–1.8) 1.5 (1.3–1.7) 2.1 (1.5–2.9) 1.8 (1.3–2.5) 1.7 (1.2–2.6) 1.8 (1.4–2.4) 1.3 (1.1–1.5) 1.4 (1.1–1.6)Diastolic blood pressure‡ 1.2 (1.0–1.3) 1.2 (1.1–1.4) — — — 1.5 (1.2–2.0) 1.3 (1.1–1.4) 1.3 (1.1–1.6)Adjusted FEV1 (%)§ 2.2 (1.9–2.6) 2.3 (2.0–2.6) 2.6 (2.0–3.5) 1.9 (1.2–3.0) 1.1 (1.0–1.1) 2.1 (2.3–2.9) 1.2 (1.1–1.3) 1.2 (1.1–1.4)Cardiothoracic ratio �0.55 1.2 (1.0–1.4) 1.4 (1.2–1.5) — 1.9 (1.3–2.7) 1.8 (1.4–2.4) — 1.2 (1.1–1.3) —Blood glucose �7.0 mmol/L¶ — 1.6 (13–1.9) — — 2.6 (1.7–3.9) 1.8 (1.3–2.5) — 1.7 (1.4–2.1)

* Highest (60 to 64 years) versus lowest (45 to 49 years) 5-year age group.† Highest quartile (�162 mm Hg in men, �165 mm Hg in women) versus lowest quartile.‡ Highest quartile (�94 mm Hg in men, �93 mm Hg in women) versus lowest quartile.§ Lowest quartile (�76% in men and �79% in women) versus highest quartile.¶ In separate model among fewer subject (see Methods).FEV1 � forced expiratory volume in 1 second.



of a cardiovascular event (and 2.6-fold [95% CI: 1.9 –3.6]risk of death) conferred by atrial fibrillation associatedwith another cardiovascular problem.

DISCUSSIONA single ECG recording of atrial fibrillation in a middle-aged woman increases her risk of a cardiovascular eventfivefold in the next two decades; in men, the risk is in-creased twofold. Much of this excess risk is related tostroke and heart failure. Atrial fibrillation is also an inde-pendent predictor of cardiovascular and all-cause mor-tality in both sexes. However, in the absence of concom-itant cardiac disease (i.e., lone atrial fibrillation), the ad-verse effects of atrial fibrillation are less clear.

We had very long-term follow-up of a particularlycomplete population sample of a large number of initiallymiddle-aged men and women, with excellent follow-up(12,13). Although the relation between atrial fibrillationand stroke is clearly established (7,8), our data also showthat this arrhythmia is an independent risk factor forheart failure, particularly in women. The excess risk ofdeath related to heart failure is more difficult to quantify,as this diagnosis is not certified as the cause of death inScotland, as it is elsewhere (21).

Most previous studies of the relation between atrialfibrillation and heart failure were cross-sectional, makingit difficult to determine cause-effect. The Manitoba studyof male air-force recruits (mean age, 31 years) is the onlyother study that has reported a long-term, independentassociation (RR � 3.0) between atrial fibrillation andheart failure (22). Our data provide support for the con-cept of tachycardia-mediated cardiomyopathy (23–26),suggesting that atrial fibrillation could be causally relatedto, rather than just associated with, the development of

heart failure. Atrial fibrillation might simply be a markerof increased risk of another cardiovascular event, such asmyocardial infarction, which is the ultimate cause ofheart failure; however, we did not find that atrial fibrilla-tion was associated with an increased risk of coronaryevents. We also do not know whether some of the 100patients with atrial fibrillation at baseline had heart fail-ure at that time. A total of 13 subjects (10 men) did com-plain of dyspnea on exertion, but all fulfilled the MedicalResearch Council’s criteria for chronic bronchitis andonly 1 had subsequent heart failure. Moreover, only 3 ofthe 100 subjects with atrial fibrillation were hospitalizedwithin 3 years of the baseline screening because of heartfailure; 50% occurred more than 10 years later. Given thetypical short-term pattern of morbidity and mortality as-sociated with heart failure, there is good reason to suggestthat these subjects did not have latent heart failure atbaseline (22,27–29).

Atrial fibrillation was associated with an increased riskof cardiovascular and all-cause mortality, especially forwomen. Although this has been reported (6,30), there hasbeen some remaining uncertainty about the independenteffects of atrial fibrillation in the general population, be-cause the majority of prior studies described selected co-horts. Our analysis also confirmed the clear relation be-tween atrial fibrillation and stroke (7,8), which was oftenfatal.

Our study has several limitations, mainly in relation tothe collection of baseline data that may not have beenperceived to be important or that were not available at thetime of the initial survey. Thus, we do not have echocar-diographic findings for this cohort. Similarly, because theimportance of heart failure was under-recognized in the1970s and not sought specifically during screening, we donot have baseline data on this condition. The same is true

Figure 2. Age-adjusted, sex-specific survival during 20-year follow-up, by the presence (n � 100) or absence (n � 15,306) of atrialfibrillation at baseline.


October 1, 2002 THE AMERICAN JOURNAL OF MEDICINE� Volume 113 363

for valvular heart disease and thyroid disease. Our reportrelies on hospital discharge codings, with uncertain diag-nostic accuracy. Finally, because our cohort was not ex-amined serially during the 20 years of follow-up, we haveno information on the development of new atrial fibril-lation during this period, nor were we able to examine theeffects of changes in risk factors or treatments. About3.5% of our cohort was discharged from the hospital witha diagnosis of atrial fibrillation during the follow-up pe-riod (13), and these subjects were more likely to die pre-maturely. Thus, the development of new atrial fibrillationwould reduce the apparent effects of this arrhythmia atbaseline.

In conclusion, approximately 75% of subjects withatrial fibrillation in middle age will die or be hospitalizedfor a cardiovascular reason within 20 years, a two- tothreefold increase compared with those who do not haveatrial fibrillation. Much of this increased risk is associatedwith stroke and heart failure. Thus, therapeutic strategiesaimed at preventing heart failure, in addition to thoseused already to prevent stroke, may be important in pa-tients with atrial fibrillation.

ACKNOWLEDGMENTWe gratefully acknowledge the contribution of Pauline Mac-Kinnon, who is responsible for updating and maintaining theRenfrew/Paisley cohort database.

REFERENCES1. Chugh SS, Blackshear JL, Shen WK, et al. Epidemiology and natural

history of atrial fibrillation: clinical implications. J Am Coll Cardiol.2001;37:371–378.

2. Go AS, Hylek EM, Phillips KA, et al. Prevalence of diagnosed atrialfibrillation in adults: national implications for rhythm manage-ment and stroke prevention: the AnTicoagulation and Risk Factorsin Atrial Fibrillation (ATRIA) Study. JAMA. 2001;285:2370 –2375.

3. Stewart S, MacIntyre K, McCleod MC, et al. Trends in hospitalactivity, morbidity and case fatality related to atrial fibrillation inScotland, 1986 –1996. Eur Heart J. 2001;22:693–701.

4. Kannel WB, Abbot RD, Savage DD, et al. Epidemiological featuresof chronic atrial fibrillation: the Framingham study. N Engl J Med.1982;306:1018 –1024.

5. Wolf PA, Benjamin EJ, Belanger AJ, et al. Secular trends in theprevalence of atrial fibrillation: the Framingham Study. Am Heart J.1996;131:790 –795.

6. Kannel WB, Wolf PA, Benjamin EJ, et al. Prevalence, incidence,prognosis, and predisposing conditions for atrial fibrillation: pop-ulation-based estimates. Am J Cardiol. 1998;82:2N–9N.

7. Wolf P, Abbott R, Kannel W. Atrial fibrillation as an independentrisk factor for stroke: the Framingham Study. Stroke. 1991;22:983–998.

8. Wolf P, Mitchell J, Baker C, et al. Impact of atrial fibrillation onmortality, stroke and medical costs. Arch Intern Med. 1998;158:229 –234.

9. The Boston Area Anticoagulation Trial for Atrial Fibrillation Inves-tigators. The effect of low-dose warfarin on the risk of stroke inpatients with non-rheumatic atrial fibrillation. N Engl J Med. 1990;323:1505–1511.

10. Peterson P, Boysen G, Godtfredsen J, et al. Placebo-controlled, ran-domised trial of warfarin and aspirin for prevention of thrombo-embolic complications in chronic atrial fibrillation. The Copenha-gen AFASAK study. Lancet. 1989;1:175–179.

11. Hart R, Benavente O, McBride R, Pearce L. Antithrombotic therapyto prevent stroke in patients with atrial fibrillation: a meta-analysis.Ann Intern Med. 1999;131:492–503.

12. Hawthorne VM, Watt GC, Hart CL, et al. Cardiorespiratory diseasein men and women in urban Scotland: baseline characteristics ofthe Renfrew/Paisley (Midspan) study population. Scott Med J. 1995;40:102–107.

13. Stewart S, Hart CL, Hole DA, McMurray JJV. Population preva-lence, incidence and predictors of atrial fibrillation in the Renfrew/Paisley Study. Heart. 2001;86:516 –521.

14. Rose GA. The diagnosis of ischaemic heart pain and intermittentclaudication in field surveys. Bull World Health Organ. 1962;27:645–658.

15. Medical Research Council. Definition and classification of chronicbronchitis for epidemiological purposes. Lancet. 1965;i:775–779.

16. Hole DJ, Watt GM, Davey-Smith, et al. Impaired lung function andmortality risk in men and women: findings from the Renfrew andPaisley prospective population study. BMJ. 1996;313:711–716.

17. Prineas RJ, Crow RS, Blackburn H. The Minnesota Code Manual ofElectrocardiographic Findings: Standards and Procedures for Mea-surement and Classification. Boston, Massachusetts: John Wright;1982.

18. Kendrick S, Clarke J. The Scottish record linkage system. HealthBull. 1993;51:72–79.

19. Manual of the International Statistical Classification of Diseases,Injuries and Causes of Death. 9th rev. Geneva, Switzerland: WorldHealth Organization; 1977.

20. Harley K, Jones C. Quality of Scottish Morbidity Record (SMR)data. Health Bull. 1996;54:410 –417.

21. Murdoch DR, Love MP, Robb SD, et al. Importance of heart failureas a cause of death: contribution to overall mortality and coronaryheart disease mortality in Scotland. Eur Heart J. 1998;19:1829 –1835.

22. Krahn AD, Manfreda J, Tate RB, et al. The natural history of atrialfibrillation: incidence, risk factors and prognosis in the ManitobaFollow-up Study. Am J Med. 1995;98:476 –484.

23. Redfield MM, Kay GN, Jenkins LS, et al. Tachycardia-relatedcardiomyopathy: a common cause of ventricular dysfunction inpatients with atrial fibrillation referred for atrioventricular abla-tion. Mayo Clin Proc. 2000;75:790 –795.

24. Edner M, Caidahl K, Bergfelt L, et al. Prospective study of left ven-tricular function JR. Evidence of inadequate investigation andtreatment of patients with heart failure. Br Heart J. 1994;71:584 –587.

25. Shinbane JS, Wood MA, Jensen DN, et al. Tachycardia-inducedcardiomyopathy: a review of animal models and clinical studies.J Am Coll Cardiol. 1997;29:709 –715.

26. Stevenson WG, Stevenson LW. Atrial fibrillation in heart failure.N Engl J Med. 1999;341:910 –911.

27. Clarke KW, Gray D, Hampton JR. Evidence of inadequate investi-gation and treatment of patients with heart failure. Br Heart J. 1994;71:584 –587.

28. Stewart S, MacIntyre K, Hole DA, et al. More malignant than can-cer? Five-year survival following a first admission for heart failure inScotland. Eur J Heart Fail. 2001;3:315–322.

29. Cowie MR, Wood DA, Coats AJS, et al. Survival of patients with anew diagnosis of heart failure: a population based study. Heart.2001;83:505–510.

30. Benjamin EJ, Wolf PA, D’Agostino RB, et al. Impact of atrial fibril-lation on the risk of death: the Framingham Heart Study. Circula-tion. 1999;98:946 –952.



a population-based study of the long-term risks associated with atrial fibrillation: 20-year...

Documents