- Email: [email protected]
Coronary Ischemic Events after First Atrial Fibrillation: Risk and Survival
The American Journal of Medicine (2007) 120, 357-363
CLINICAL RESEARCH STUDY
Coronary Ischemic Events after First Atrial Fibrillation: Risk and Survival Yoko Miyasaka, MD, PhD,a,b Marion E. Barnes, MSc,a Bernard J. Gersh, MB, ChB, Dphil,a Stephen S. Cha, MS,c Kent R. Bailey, PhD,c James B. Seward, MD,a Toshiji Iwasaka, MD, PhD,b Teresa S. M. Tsang, MDa a c
Division of Cardiovascular Diseases and Internal Medicine, and bSection of Biostatistics, Mayo Clinic, Rochester, Minn. Cardiovascular Division, Department of Medicine II, Kansai Medical University, Osaka, Japan. ABSTRACT PURPOSE: We aimed to determine the long-term, gender-specific incidence and mortality risk of coronary ischemic events after first atrial fibrillation (AF). METHODS: In this longitudinal cohort study, adult residents of Olmsted County, Minnesota, with an electrocardiogram-confirmed AF first documented in 1980 to 2000 and without prior coronary heart disease, were followed to 2004. The primary outcome was first coronary events (angina with angiographic confirmation, unstable angina, nonfatal myocardial infarction, or coronary death). Sex-specific incidence of coronary ischemic events and survival after development of such events were assessed using Cox proportional hazards modeling. Kaplan-Meier estimates of risks for coronary ischemic events were compared with those predicted by the Framingham equation. RESULTS: Of the 2768 subjects (mean age 71 years, 48% were men), 463 (17%) had a first coronary event during a follow-up of 6.0 ⫾ 5.2 years. The unadjusted incidence was 31 per 1000 person-years, and there was no difference between men and women. The incidence was higher in men (hazard ratio 1.32, P ⫽ .004) after adjusting for age. The 10-year event estimates were 22% and 19% in men and women, respectively, by our Kaplan-Meier analyses, and 21% and 11%, respectively, by Framingham risk equation. The mortality risk after coronary events was higher in women (hazard ratio 2.99 vs 2.33; P ⫽ .044), even after multiple adjustment. CONCLUSIONS: First AF marks a high risk for new coronary ischemic events in both men and women. AF conferred additional risk for coronary events beyond conventional risk prediction in women only. The excess mortality risk associated with the development of coronary events was significantly greater in women. © 2007 Elsevier Inc. All rights reserved. KEYWORDS: Atrial fibrillation; Coronary heart disease; Incidence; Prognosis
Atrial fibrillation (AF) is a growing public health problem1 that has reached epidemic proportions.2,3 It is a potent risk factor for stroke,4,5 congestive heart failure,6,7 and possibly cognitive dysfunction.8,9 Although studies have shown that the risk of acute coronary ischemic events at the time of first AF is low,10,11 there are no This study was supported by the American Heart Association National Scientist Development Grant. There is no financial conflict on the part of any author. Requests for reprints should be addressed to Teresa S. M. Tsang, MD, Mayo Clinic, Division of Cardiovascular Diseases, 200 First Street SW, Rochester, MN 55905. E-mail address: [email protected]
0002-9343/$ -see front matter © 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.amjmed.2006.06.042
published longer-term data. There is a paucity of data with regard to the incidence and timing of coronary events after first AF, as well as the predictors for such events. In addition, it is unknown whether gender differences exist in the risk of post-AF coronary ischemic events and survival after these events. We conducted a community-based longitudinal cohort study that spanned 2 decades and assessed the gender-specific incidence, timing, and predictors of coronary ischemic events after first diagnosis of AF in patients without prior evidence of coronary heart disease. We determined the gender-specific mortality risk conferred by the development of coronary ischemic events.
358
METHODS Study Setting
The American Journal of Medicine, Vol 120, No 4, April 2007 no history or evidence of coronary heart disease before or at the time of AF diagnosis.
The community-based cohort study was approved by the Definition and Ascertainment of Outcome Mayo Foundation Institutional Review Board. Olmsted Coronary Ischemic Events. The primary outcome of inCounty, Minnesota, is well suited terest was a first documented corfor the conduct of studies with onary ischemic event, defined as long-term follow-up because of a angina pectoris with angiographic CLINICAL SIGNIFICANCE number of unique features.12 Geoconfirmation, unstable angina, graphically, the community is relnonfatal acute myocardial infarc● Patients diagnosed with first atrial fibrilatively isolated from other urban tion, or coronary death occurring lation constitute a high-risk group for centers, and medical care is delivafter the date of first AF. We subsequent new coronary ischemic ered by only a few health care chose well-defined end points events. providers, principally the Mayo with high specificity and opted to Clinic and its associated hospitals. ● Atrial fibrillation conferred additional exclude the patients who had preThe majority of Olmsted County risk for the development of coronary sumed angina but without angioresidents return to the Mayo graphic confirmation of obstrucischemic events, beyond conventional Clinic on a regular basis, allowing tive coronary artery disease. risk prediction, in women only. capture of health-related events. A Coronary artery disease was deprevious study showed that 96% ● The excess mortality risk associated with fined as an angiographic finding of of Olmsted County women resithe development of new coronary events atherosclerotic narrowing of least dents aged 65 to 74 years returned post-atrial fibrillation was greater in 50% in any of the 3 main arterial to the Mayo Clinic within a 3-year women. distributions. Unstable angina was period.12 For each patient at the defined as the presence of new or Mayo Clinic, a unified medical accelerated ischemic symptoms record containing details of all inwith or without electrocardiopatient and outpatient encounters is maintained. Coded digraphic changes, but without elevation of cardiac enzymes. agnoses are electronically entered into a medical diagnostic Acute myocardial infarction was considered present if at index, allowing easy identification of all patients with a least 2 of the 3 diagnostic criteria were fulfilled: compatible diagnosis of interest. An electronic electrocardiogram dataclinical presentation, diagnostic cardiac enzyme/biomarker base has been in use at Mayo since 1976, allowing rapid levels, and electrocardiographic changes consistent with identification of the coded interpretation and diagnoses for myocardial infarction. Coronary death was defined as death each electrocardiogram. All electrocardiograms can also be attributable to fatal acute myocardial infarction and death retrieved for direct review. with the primary cause being coronary heart disease (based on documentation per death certificates or physician adjudication by chart review). The ascertainment of outcomes Incident Atrial Fibrillation Cohort and was accomplished through comprehensive review of the Study Population medical records in addition to cross-referencing the multiple The medical records of adult residents of Olmsted County, administrative databases to identify any inconsistencies. Minnesota, who had a first AF documented between January 1, 1980, and December 31, 2000, in any of the Mayo administrative databases (medical index, surgical index, electrocardiographic, and echocardiographic databases) were reviewed and followed forward in the medical records to March 2004 by 1 cardiologist (Y. M.). For the purpose of this study, we excluded subjects who had any history or evidence of coronary heart disease and subjects who sustained a first coronary ischemic event on the same day as the first AF. Because the unit record system at Mayo dated back to the early 1900s, any electrocardiogram performed in any patient was contained within the unit record. Any AF that occurred before the establishment of administrative databases would have been identified from the chart review process. Final inclusion in the study population required the following: The AF was sustained, allowing confirmation by 12-lead electrocardiogram; the AF episode was verified as the first recognized AF event for the person; and there was
Death. The secondary outcome of interest was death. Ascertainment of death was accomplished through comprehensive review of the medical records, death certificates, Vital Status Information from Mayo Registration, Minnesota State Death Tapes, and Social Security Death Index. Statistical Analysis. Baseline characteristics were summarized by means and standard deviations or frequency percentages and compared between subjects with or without subsequent development of coronary ischemic events by using age-adjusted Cox proportional hazards analyses for both sexes. Average annual incidence rates were reported by sex and 10-year age interval, and summaries were tabulated as cases per 1000 person-years. The overall and sex-specific cumulative incidence of first coronary ischemic events after first AF was estimated using the Kaplan-Meier method. Incidence was assessed as a function of age and sex, using
Miyasaka et al
Risk of Coronary Ischemic Events in AF
a Cox proportional hazards model for time to first coronary ischemic event. Cox models based on clinical variables at the time of first AF were also developed using backward stepwise selection. The clinical covariates considered included type of AF (paroxysmal vs chronic) at initial diagnosis, body mass index, systolic or diastolic blood pressure, history of systemic hypertension, dyslipidemia, diabetes mellitus, smoking, peripheral vascular disease, stroke, congestive heart failure, valvular heart disease, chronic obstructive pulmonary disease, obstructive sleep apnea, regular alcohol use, hyperthyroidism, chronic renal disease, and malignancy. We also analyzed whether AF provides additional predictive information, beyond conventional risk factor consideration, for the development of subsequent coronary ischemic events. We compared our sex-specific coronary incidence rates, estimated by Kaplan-Meier analysis, with those predicted by the Framingham risk equation13 at 10 years. The association of first coronary ischemic event with subsequent survival was estimated using time-dependent proportional hazards models, controlling for age and sex, and with and without clinical variables. The same clinical
Table 1
359 covariates were considered, and the same selection method was used. The effect of interim coronary revascularization (percutaneous coronary intervention or coronary artery bypass grafting) on mortality was accounted for by including it as a time-dependent covariate in the model. Mortality ratios relative to the general Minnesota white population were calculated by dividing the observed number of mortality events by an “expected number” of deaths. This expected number was calculated as the sum of each individual’s cumulative hazard, based on the standard age, sex, and calendar-year specific life-table survival probability up to the individual’s follow-up time. Confidence intervals (CIs) were based on the Poisson distribution, and gender comparisons were based on a likelihood ratio chi-square test. All tests of significance were 2-tailed, and a P value less than .05 was considered statistically significant.
RESULTS A total of 4618 subjects (51% were men; mean age 73 years) were confirmed to have developed a first AF during 1980 to 2000. We excluded 1776 subjects (38%) because they had a history of coronary heart disease and 74 subjects
Baseline Characteristics of the Study Population Stratified by Subsequent Development of Coronary Ischemic Events Men
Women
Variable
No Events (n ⫽ 1091)
Had Events (n ⫽ 227)
No Events (n ⫽ 1214)
Had Events (n ⫽ 236)
Age (y) Paroxysmal AF, N (%) BMI (kg/m2) Systemic hypertension, N (%) Hypertensive therapy, N (%) Systolic BP (mm Hg) Diastolic BP (mm Hg) Dyslipidemia, N (%) Lipid-lowering therapy, N (%) Total cholesterol (mg/dL)§ HDL cholesterol (mg/dL)§ Diabetes mellitus, N (%) Smoking, N (%) Peripheral artery disease, N (%) Stroke, N (%) Prior CHF, N (%) VHD, N (%) COPD, N (%) Obstructive sleep apnea, N (%) Regular alcohol use, N (%) Hyperthyroidism, N (%) Chronic renal disease, N (%) History of malignancy, N (%)
66 ⫾ 18 797 (73) 27.3 ⫾ 5.4 691 (63) 373 (34) 133 ⫾ 18 78 ⫾ 11 344 (32) 19 (1.7) 191 ⫾ 47 45 ⫾ 14 125 (11) 762 (70) 72 (6.6) 79 (7.2) 54 (4.9) 142 (13) 245 (22) 29 (2.7) 225 (21) 9 (0.8) 137 (13) 314 (29)
69 ⫾ 13‡ 151 (67) 28.4 ⫾ 5.2† 172 (76)‡ 91 (40)† 138 ⫾ 20† 79 ⫾ 11 87 (38)† 4 (1.8) 198 ⫾ 45 44 ⫾ 12 37 (16)† 174 (77)* 19 (8.4)* 18 (7.9) 9 (4.0) 39 (17)* 60 (26) 9 (4.0)* 47 (21) 2 (0.9) 36 (16)† 46 (20)
76 ⫾ 14 825 (68) 26.5 ⫾ 6.9 961 (79) 613 (50) 139 ⫾ 21 78 ⫾ 11 329 (27) 20 (1.6) 204 ⫾ 45 55 ⫾ 17 143 (12) 424 (35) 80 (6.6) 91 (7.5) 89 (7.3) 264 (22) 187 (15) 6 (0.5) 72 (5.9) 21 (1.7) 99 (8.2) 364 (30)
77 ⫾ 11‡ 147 (62) 27.4 ⫾ 7.3* 215 (91)‡ 136 (58)‡ 147 ⫾ 19‡ 80 ⫾ 10 64 (27) 3 (1.3) 211 ⫾ 46 52 ⫾ 14 49 (21)‡ 79 (33) 30 (13)‡ 17 (7.2) 18 (7.6)* 58 (25) 36 (15) 2 (0.8) 9 (3.8) 5 (2.1) 26 (11)‡ 48 (20)
AF ⫽ atrial fibrillation; BMI ⫽ body mass index; BP ⫽ blood pressure; CHF ⫽ congestive heart failure; COPD ⫽ chronic obstructive pulmonary disease; HDL ⫽ high-density lipoprotein; VHD ⫽ valvular heart disease. Values are given as mean ⫾ standard deviation or number (percentage). P value for events status differences in each characteristics by Cox proportional hazards analysis after adjusting for age. *.01 ⬍ P ⬍.05 vs no events. †.001 ⬍ P ⬍.01 vs no events. ‡P ⬍.001 vs no events. §The available number of data were 1999 for total cholesterol and 1143 for HDL cholesterol.
360
The American Journal of Medicine, Vol 120, No 4, April 2007 confirmation (72 men [32%], 45 women [19%]), unstable angina (45 men [20%], 59 women [25%]), nonfatal acute myocardial infarction (66 men [29%], 79 women [34%]), and coronary death (44 men [19%], 53 women [22%]).
Incidence, Timing, and Predictors of Coronary Ischemic Events
Figure 1 The unadjusted incidence of first coronary ischemic events by time from first AF diagnosis among men (left) and women (right). AF ⫽ atrial fibrillation.
(2%) because they had a coronary ischemic event that occurred on the same day as the first AF. The remaining 2768 subjects (48% were men, mean age 71 years, range 18-107 years) constituted the study population. Of these, 463 patients (17%) had a total of 841 coronary ischemic events during a mean follow-up period of 6.0 ⫾ 5.2 years, and 1700 died. The proportion of our cohort followed to death or to within 2 years of the last follow-up of this study (March 2004) was 92.4%. The baseline characteristics of the study population, stratified by sex and subsequent coronary ischemic events, are shown in Table 1.
Distribution of Coronary Ischemic Events The type of first coronary ischemic events (227 men, 236 women) were as follows: angina pectoris with angiographic
The unadjusted incidence of coronary ischemic events after AF was 31 per 1000 person-years and was not significantly different between men and women (30.2 vs 31.4 per 1000 person-years; P ⫽ .68). The risk of new coronary ischemic events was greatest during the first year of AF (cumulative incidence 4.7%, 95% CI, 3.9-5.6) and plateaued thereafter (2.5% per year after first year, cumulative event rate 14.4% at 5 years, 95% CI, 12.8-15.9) (Figure 1) (Table 2). When adjusted for age, male sex was associated with increased risk of coronary ischemic events (hazard ratio [HR] 1.32, 95% CI, 1.09-1.60, P ⫽ .004). When age stratified, the incidence of coronary ischemic events increased throughout the age range in women but did not increase in men after age 85 years or more (Table 2). When a sex-specific correction for “age 85 years or more” was added to the model with a linear term for age and sex, there was a significant (P ⫽ .028) downward correction for men but no apparent departure from linearity for women (P ⫽ .06 for the difference between the corrections). The overall incidence of coronary ischemic events did not differ between men and women, likely because the mean age of women was 10 years older (76 vs 66 years). In a multivariable Cox model for the prediction of first coronary events (Table 3), advancing age, male sex, higher systolic blood pressure, and history of systemic hyperten-
Table 2 Age-and Sex-Specific Incidence Rates and Kaplan-Meier Estimated Cumulative Incidence of Coronary Ischemic Events for Patients Diagnosed with First Atrial Fibrillation Kaplan-Meier Estimated Cumulative Incidence
Overall Men ⬍55 y 55-64 y 65-74 y 75-84 y ⱖ85 y Overall Women ⬍55 y 55-64 y 65-74 y 75-84 y ⱖ85 y Overall
N
Person-y
Incidence Rate/1000 Person-y*
1 y, % (95% CI)
5 y, % (95% CI)
2768
15046.53
30.77 (463)
4.7 (3.9-5.6)
14.4 (12.8-15.9)
299 207 333 323 156 1318
2573.86 1428.25 1829.45 1321.64 371.87 7525.07
12.04 27.31 38.81 52.96 43.03 30.17
(31) (39) (71) (70) (16) (227)
0.7 2.5 8.4 6.3 5.0 4.8
(0.0-1.7) (0.3-4.7) (5.3-11.5) (3.3-9.2) (0.9-9.0) (3.5-6.0)
5.4 11.8 19.7 19.5 17.2 14.5
(2.5-8.2) (6.6-16.7) (14.7-24.4) (14.0-24.7) (7.7-25.7) (12.2-16.7)
101 139 288 533 389 1450
965.59 1058.37 1826.05 2589.81 1081.63 7521.45
10.36 21.73 29.02 35.52 53.62 31.38
(10) (23) (53) (92) (58) (236)
1.0 5.4 4.2 4.6 6.1 4.7
(0.0-2.9) (1.4-9.2) (1.7-6.6) (2.7-6.4) (3.5-8.7) (3.5-5.8)
3.2 10.2 10.9 16.1 22.7 14.3
(0.0-6.6) (4.5-15.6) (6.9-14.9) (12.3-19.7) (16.3-28.6) (12.1-16.4)
CI ⫽ confidence interval. *Data are presented as incidence rate followed in parentheses by the actual number of cases observed.
Miyasaka et al
Risk of Coronary Ischemic Events in AF
361
Table 3 Overall and Sex-Specific Multivariable-Adjusted Hazard Ratios for the Coronary Ischemic Events in Patients Diagnosed with First Atrial Fibrillation Overall
Age/10 y Men Systolic BP/10 mm Hg Systemic hypertension Diabetes mellitus Peripheral artery disease Valvular heart disease Obstructive sleep apnea Chronic renal disease
Men
Women
HR
95% CI
HR
95% CI
HR
95% CI
1.30‡ 1.37† 1.06* 1.65‡ 1.76‡ 1.56† 1.39† 1.97* 1.47†
1.20-1.42 1.13-1.67 1.01-1.12 1.23-2.21 1.38-2.24 1.14-2.13 1.11-1.75 1.07-3.64 1.11-1.95
1.32‡ 1.04 1.47* 1.50* 1.39 1.39 1.97 1.34
1.18-1.46 0.96-1.13 1.02-2.12 1.04-2.17 0.85-2.28 0.98-1.97 0.99-3.89 0.91-1.96
1.29‡ 1.07 2.23† 2.04‡ 1.67* 1.44* 2.43 1.79†
1.13-1.47 0.99-1.14 1.33-3.73 1.47-2.82 1.12-2.50 1.06-1.95 0.59-9.97 1.17-2.72
BP ⫽ blood pressure, HR ⫽ hazard ratio, CI ⫽ confidence interval. *.01 ⬍ P ⬍.05. †.001 ⬍ P ⬍.01. ‡P ⬍.001.
sion, diabetes mellitus, peripheral artery disease, valvular heart disease, obstructive sleep apnea, and chronic renal disease were significant independent predictors. The type of AF, paroxysmal or chronic AF, was not significant. Separate models for men and women including these same variables appeared similar (Table 3), and none of the interactions between the risk factors and the gender were significant. A prediction formula for the probability of coronary events within 1, 5, or 10 years based on this model is given in the Appendix, and Table 4 shows an array of sample calculations.
Gender Differences in the Risk of Coronary Events after First Atrial Fibrillation Although men have higher age-adjusted incidence of coronary ischemic events after first AF, how our sex-specific coronary incidence rates relate to those predicted by the Framingham risk equation is of interest. Restricting the analysis to the subgroup aged 30 to 74 years (N ⫽ 1403), the same age range for which the Framingham coronary risk equation was developed,13 the estimations of coronary event rates at 10 years were 22% (95% CI, 18-26) for men and 19% (95% CI, 14-23) for women by Kaplan-Meier analyses, versus 21% for men and 11% for women by Framingham risk equation.13
Table 4
Impact of Coronary Ischemic Events on Survival In time-dependent Cox analysis adjusted for age, the hazard for death was highly significantly associated with the occurrence of post-AF coronary ischemic events in each sex (P ⬍.0001). When an overall model was fit, with age, sex, coronary events, and an interaction between coronary events and sex, the HR associated with events was significantly higher in women (HR 2.99, 95% CI, 2.53-3.53 vs HR 2.33, 95% CI, 1.94-2.81, P ⫽ .044). Before the occurrence of a coronary ischemic event, the age-adjusted HR for men versus women was 1.23 (95% CI, 1.10-1.37), whereas after the occurrence of a coronary ischemic event, there was no significant difference in age-adjusted mortality risk. After adjustment for age, clinical risk factors, and coronary revascularization as additional time-dependent covariates, the occurrence of a coronary ischemic event remained a highly significant independent predictor of death in both sexes (men, HR 2.71, 95% CI, 2.22-3.32; women, HR 3.55, 95% CI, 2.99-4.22; P ⫽ .033). Compared with the general Minnesota white population, the relative mortality HRs before coronary ischemic events were 1.56 (95% CI, 1.47-1.66) in men and 1.71 (95% CI, 1.62-1.82) in women (P ⫽ .10). After the development of a first coronary ischemic event, the corresponding relative
Probability of Coronary Ischemic Event within 5 Years Using a Prediction Formula* (Appendix) Men
Women
Risk Factors
None
HTN
HTN ⫹ DM
HTN ⫹ DM ⫹ VHD
None
HTN
HTN ⫹ DM
HTN ⫹ DM ⫹ VHD
Age 60 y Age 70 y Age 80 y
7% 9% 11%
11% 14% 17%
18% 23% 29%
24% 30% 37%
5% 6% 8%
8% 10% 13%
13% 17% 22%
18% 23% 29%
HTN ⫽ systemic hypertension; DM ⫽ diabetes mellitus; VHD ⫽ valvular heart disease. *Imputing the mean value of systolic blood pressure (137 mm Hg) for these estimations.
362 Table 5
The American Journal of Medicine, Vol 120, No 4, April 2007 Mortality Ratios Relative to the General Minnesota White Population after Atrial Fibrillation Diagnosis Men
Overall Pre-events Post-events
Women
Observed*
Expected†
Ratio‡ (95% CI)
Observed*
Expected†
Ratio‡ (95% CI)
710 563 147
411.67 361.11 50.56
1.72 (1.62-1.84) 1.56 (1.47-1.66) 2.91 (2.35-3.66)
990 806 184
516.40 470.99 45.41
1.92 (1.81-2.04)§ 1.71 (1.62-1.82) 4.05 (3.23-5.36)储
CI ⫽ confidence interval. *Observed is number of deaths in the absence of coronary ischemic events (pre-events) or occurring after coronary ischemic events (post-events). †Expected is the cumulative hazard (based on age and sex using Minnesota white population) for follow-up before or in the absence of coronary ischemic events (pre-events) or after the occurrence of coronary ischemic events (post-events). ‡Ratio of observed to expected represents hazard ratio relative to the general population. §.01 ⬍ P ⬍.05 vs men. 储P ⬍.001 vs men.
HRs were 2.91 (95% CI, 2.35-3.66) in men and 4.05 (95% CI, 3.23-5.36) in women (P ⬍.001) (Table 5).
DISCUSSION In our community-based cohort, the first AF marks a high risk for new coronary ischemic events in both men and women. In contrast with men, AF conferred an additional risk of ischemic events in women beyond conventional risk prediction. The excess mortality risk associated with the development of new coronary ischemic events was greater in women.
Incidence of Coronary Ischemic Events After First Atrial Fibrillation Coronary heart disease has been regarded as an independent risk factor of AF,14,15 although the data supporting this connection came from only a small handful of studies, and the event rate was modest.16,17 In the Framingham Heart Study, the 2-year age-adjusted incidence for chronic AF among patients with coronary heart disease, per 1000, was 8.9 for men and 2.3 for women.16 However, little is known regarding AF as a precursor of new coronary ischemic events in patients without prior evidence of coronary heart disease. Our study showed that the ischemic event rate was high after manifestation of first AF, and that the unadjusted incidence did not differ between men and women. Incidence, when adjusted for age, was higher in men. In fact, the incidence of post-AF coronary ischemic events was higher than that of post-AF stroke in the same cohort (10% at 5 years).18 The precise reasons for the high ischemic event rate after first AF among patients with no prior coronary heart disease could not be well delineated, but a likely explanation is that persons who develop AF frequently harbor the same cardiovascular risk factors for coronary ischemic events. In our cohort, more than 70% had hypertension, 30% had dyslipidemia, and 13% had diabetes. It is unclear why some develop ischemic events first and others develop AF first.
In a multivariate Cox model, higher systolic blood pressure, systemic hypertension, diabetes mellitus, and obstructive sleep apnea were significant independent predictors of new coronary ischemic events after first AF. Early diagnosis and control of these treatable conditions may have implications on cardiovascular outcomes in patients who have AF. Although the age-adjusted incidence of new coronary events post-AF was higher in men; AF actually conferred an additional risk (⬃70%) of new coronary events in women only, based on the comparison of our KaplanMeier 10-year event rates and those predicted by the Framingham risk equation. Our data set does not lend itself readily to analyses for potential explanatory factors. The underlying reasons for this gender difference in the impact of AF on subsequent coronary risk warrant further investigation.
Impact of Coronary Ischemic Events on Survival AF is known to be associated with excess mortality risk.14,19-22 Our study showed that the development of a first coronary ischemic event further increased the mortality risk, even after multivariable adjustment. In addition, the mortality risk relative to the general population was higher in women than in men. This was similar to our previous finding of greater relative mortality risk in women associated with post-AF stroke.18 The reasons for these differences could not be determined in this study but clearly require further studies.
LIMITATIONS There were inherent biases associated with the retrospective design. It is possible that the incidence of AF may have been underestimated, given that some patients may not have received care at the Mayo Clinic, although the underestimation would likely be small because the Mayo Clinic is the principal health care provider and referral center for Olmsted County.12 Also, we may have under-
Miyasaka et al
Risk of Coronary Ischemic Events in AF
estimated the coronary event rates because we did not include those who had presumed angina but did not have angiographic confirmation. The population of Olmsted County, Minnesota, is predominantly white and not racially ethnically diverse.
CONCLUSIONS Patients diagnosed with a first AF constitute a high-risk group for subsequent new coronary ischemic events. Although the age-adjusted incidence of such post-AF coronary ischemic events was higher in men, AF actually conferred an additional risk for the development of ischemic events, beyond conventional risk prediction, in women only. Further, the excess mortality risk associated with the development of post-AF new coronary events was greater in women. See appendix online at http://www.amjmed.com
References 1. 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. Braunwald E. Shattuck lecture— cardiovascular medicine at the turn of the millennium: triumphs, concerns, and opportunities. N Engl J Med. 1997;337:1360-1369. 3. Alpert JS. Atrial fibrillation: a growth industry in the 21st century. Eur Heart J. 2000;21:1207-1208. 4. Whisnant JP, Wiebers DO, O’Fallon WM, et al. A population-based model of risk factors for ischemic stroke: Rochester, Minnesota. Neurology. 1996;47:1420-1428. 5. Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent risk factor for stroke: the Framingham Study. Stroke. 1991;22:983988. 6. Scheinman MM. Atrial fibrillation and congestive heart failure: the intersection of two common diseases. Circulation. 1998;98:941-942. 7. Wang TJ, Larson MG, Levy D, et al. Temporal relations of atrial fibrillation and congestive heart failure and their joint influence on mortality: the Framingham Heart Study. Circulation. 2003;107:29202925. 8. Kilander L, Andren B, Nyman H, et al. Atrial fibrillation is an independent determinant of low cognitive function: a cross-sectional study in elderly men. Stroke. 1998;29:1816-1820.
363 9. Ott A, Breteler MM, de Bruyne MC, et al. Atrial fibrillation and dementia in a population-based study. The Rotterdam Study. Stroke. 1997;28:316-321. 10. Zimetbaum PJ, Josephson ME, McDonald MJ, et al. Incidence and predictors of myocardial infarction among patients with atrial fibrillation. J Am Coll Cardiol. 2000;36:1223-1227. 11. Friedman HZ, Weber-Bornstein N, Deboe SF, Mancini GB. Cardiac care unit admission criteria for suspected acute myocardial infarction in new-onset atrial fibrillation. Am J Cardiol. 1987;59:866-869. 12. Melton LJ 3rd. History of the Rochester Epidemiology Project. Mayo Clin Proc. 1996;71:266-274. 13. Wilson PW, D’Agostino RB, Levy D, et al. Prediction of coronary heart disease using risk factor categories. Circulation. 1998;97:18371847. 14. Krahn AD, Manfreda J, Tate RB, et al. The natural history of atrial fibrillation: incidence, risk factors, and prognosis in the Manitoba Follow-Up Study. Am J Med. 1995;98:476-484. 15. Tsang TS, Barnes ME, Bailey KR, et al. Left atrial volume: important risk marker of incident atrial fibrillation in 1655 older men and women. Mayo Clin Proc. 2001;76:467-475. 16. Kannel WB, Abbott RD, Savage DD, McNamara PM. Coronary heart disease and atrial fibrillation: the Framingham Study. Am Heart J. 1983;106:389-396. 17. Orndahl G, Thulesius O, Hood B. Incidence of persistent atrial fibrillation and conduction defects in coronary heart disease. Am Heart J. 1972;84:120-131. 18. Miyasaka Y, Barnes ME, Gersh BJ, et al. Time trends of ischemic stroke incidence and mortality in patients diagnosed with first atrial fibrillation in 1980 to 2000. Report of a Community-Based Study. Stroke. 2005;36:2362-2366. 19. Benjamin EJ, Wolf PA, D’Agostino RB, et al. Impact of atrial fibrillation on the risk of death: the Framingham Heart Study. Circulation. 1998;98:946-952. 20. Stewart S, Hart CL, Hole DJ, McMurray JJ. A population-based study of the long-term risks associated with atrial fibrillation: 20-year follow-up of the Renfrew/Paisley study. Am J Med. 2002;113:359-364. 21. Vidaillet H, Granada JF, Chyou PH, et al. A population-based study of mortality among patients with atrial fibrillation or flutter. Am J Med. 2002;113:365-370. 22. Gajewski J, Singer RB. Mortality in an insured population with atrial fibrillation. JAMA. 1981;245:1540-1544. 23. McKee PA, Castelli WP, McNamara PM, Kannel WB. The natural history of congestive heart failure: the Framingham study. N Engl J Med. 1971;285:1441-1446.
363.e1
APPENDIX Definition of Covariates Chronic versus paroxysmal AF was defined by whether there were recognizable intervening episodes of sinus rhythm. Systemic hypertension was defined by a physician’s diagnosis, need for antihypertensive therapy, or systolic blood pressure greater than 140 mm Hg or diastolic blood pressure greater than 90 mm Hg on at least 2 occasions that were not associated with acute illness or injury. Dyslipidemia was defined by a total cholesterol level of 200 mg/dL or greater, triglyceride level of 150 mg/dL or greater, low-density lipoprotein cholesterol level of 130 mg/dL or greater or high-density lipoprotein cholesterol less than 40 mg/dL on 2 or more occasions, or treatment with lipidlowering agents. Diabetes mellitus was defined by physician’s diagnosis and treatment with insulin or oral hypoglycemic agents. Smoking history was classified as past (⬎6 months prior) or current smoker. Stroke included development of any type of stroke, as defined by clinical documentation of the diagnosis with or without confirmatory findings on imaging studies. Congestive heart failure was defined by the presence of 2 major or 1 major and 2 minor Framingham
The American Journal of Medicine, Vol 120, No 4, April 2007 criteria.23 Valvular heart disease was defined by greater than mild stenosis or regurgitation by echocardiography or prior valve repair/replacement. Regular alcohol use was defined by self-reported consumption of more than 1 drink per day regularly. Chronic renal disease, chronic obstructive pulmonary disease, obstructive sleep apnea, and hyperthyroidism were defined by these clinical diagnoses in the medical records.
Prediction Formula for the Probability of Coronary Ischemic Event The proportional-hazards model-based probability of having had a coronary ischemic event within year 1, 5, or 10 of first AF diagnosis, as defined by the multivariate model in Table 3, is estimated as follows: PROB ⫽ 1 ⫺ (Pi)eXB, where XB ⫽ 0.265*(age/10) ⫹ 0.316*{1 if male} ⫹ 0.059*(systolic blood pressure/10) ⫹ 0.500*{1 if systemic hypertension} ⫹ 0.563*{1 if diabetes mellitus} ⫹ 0.446*{1 if peripheral artery disease} ⫹ 0.332*{1 if valvular heart disease} ⫹ 0.679*{1 if obstructive sleep apnea} ⫹ 0.385*{1 if chronic renal disease} ⫺ 3.450, and (P1, P5, P10) ⫽ (.9608, .8682, .7467).
CLINICAL RESEARCH STUDY
Coronary Ischemic Events after First Atrial Fibrillation: Risk and Survival Yoko Miyasaka, MD, PhD,a,b Marion E. Barnes, MSc,a Bernard J. Gersh, MB, ChB, Dphil,a Stephen S. Cha, MS,c Kent R. Bailey, PhD,c James B. Seward, MD,a Toshiji Iwasaka, MD, PhD,b Teresa S. M. Tsang, MDa a c
Division of Cardiovascular Diseases and Internal Medicine, and bSection of Biostatistics, Mayo Clinic, Rochester, Minn. Cardiovascular Division, Department of Medicine II, Kansai Medical University, Osaka, Japan. ABSTRACT PURPOSE: We aimed to determine the long-term, gender-specific incidence and mortality risk of coronary ischemic events after first atrial fibrillation (AF). METHODS: In this longitudinal cohort study, adult residents of Olmsted County, Minnesota, with an electrocardiogram-confirmed AF first documented in 1980 to 2000 and without prior coronary heart disease, were followed to 2004. The primary outcome was first coronary events (angina with angiographic confirmation, unstable angina, nonfatal myocardial infarction, or coronary death). Sex-specific incidence of coronary ischemic events and survival after development of such events were assessed using Cox proportional hazards modeling. Kaplan-Meier estimates of risks for coronary ischemic events were compared with those predicted by the Framingham equation. RESULTS: Of the 2768 subjects (mean age 71 years, 48% were men), 463 (17%) had a first coronary event during a follow-up of 6.0 ⫾ 5.2 years. The unadjusted incidence was 31 per 1000 person-years, and there was no difference between men and women. The incidence was higher in men (hazard ratio 1.32, P ⫽ .004) after adjusting for age. The 10-year event estimates were 22% and 19% in men and women, respectively, by our Kaplan-Meier analyses, and 21% and 11%, respectively, by Framingham risk equation. The mortality risk after coronary events was higher in women (hazard ratio 2.99 vs 2.33; P ⫽ .044), even after multiple adjustment. CONCLUSIONS: First AF marks a high risk for new coronary ischemic events in both men and women. AF conferred additional risk for coronary events beyond conventional risk prediction in women only. The excess mortality risk associated with the development of coronary events was significantly greater in women. © 2007 Elsevier Inc. All rights reserved. KEYWORDS: Atrial fibrillation; Coronary heart disease; Incidence; Prognosis
Atrial fibrillation (AF) is a growing public health problem1 that has reached epidemic proportions.2,3 It is a potent risk factor for stroke,4,5 congestive heart failure,6,7 and possibly cognitive dysfunction.8,9 Although studies have shown that the risk of acute coronary ischemic events at the time of first AF is low,10,11 there are no This study was supported by the American Heart Association National Scientist Development Grant. There is no financial conflict on the part of any author. Requests for reprints should be addressed to Teresa S. M. Tsang, MD, Mayo Clinic, Division of Cardiovascular Diseases, 200 First Street SW, Rochester, MN 55905. E-mail address: [email protected]
0002-9343/$ -see front matter © 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.amjmed.2006.06.042
published longer-term data. There is a paucity of data with regard to the incidence and timing of coronary events after first AF, as well as the predictors for such events. In addition, it is unknown whether gender differences exist in the risk of post-AF coronary ischemic events and survival after these events. We conducted a community-based longitudinal cohort study that spanned 2 decades and assessed the gender-specific incidence, timing, and predictors of coronary ischemic events after first diagnosis of AF in patients without prior evidence of coronary heart disease. We determined the gender-specific mortality risk conferred by the development of coronary ischemic events.
358
METHODS Study Setting
The American Journal of Medicine, Vol 120, No 4, April 2007 no history or evidence of coronary heart disease before or at the time of AF diagnosis.
The community-based cohort study was approved by the Definition and Ascertainment of Outcome Mayo Foundation Institutional Review Board. Olmsted Coronary Ischemic Events. The primary outcome of inCounty, Minnesota, is well suited terest was a first documented corfor the conduct of studies with onary ischemic event, defined as long-term follow-up because of a angina pectoris with angiographic CLINICAL SIGNIFICANCE number of unique features.12 Geoconfirmation, unstable angina, graphically, the community is relnonfatal acute myocardial infarc● Patients diagnosed with first atrial fibrilatively isolated from other urban tion, or coronary death occurring lation constitute a high-risk group for centers, and medical care is delivafter the date of first AF. We subsequent new coronary ischemic ered by only a few health care chose well-defined end points events. providers, principally the Mayo with high specificity and opted to Clinic and its associated hospitals. ● Atrial fibrillation conferred additional exclude the patients who had preThe majority of Olmsted County risk for the development of coronary sumed angina but without angioresidents return to the Mayo graphic confirmation of obstrucischemic events, beyond conventional Clinic on a regular basis, allowing tive coronary artery disease. risk prediction, in women only. capture of health-related events. A Coronary artery disease was deprevious study showed that 96% ● The excess mortality risk associated with fined as an angiographic finding of of Olmsted County women resithe development of new coronary events atherosclerotic narrowing of least dents aged 65 to 74 years returned post-atrial fibrillation was greater in 50% in any of the 3 main arterial to the Mayo Clinic within a 3-year women. distributions. Unstable angina was period.12 For each patient at the defined as the presence of new or Mayo Clinic, a unified medical accelerated ischemic symptoms record containing details of all inwith or without electrocardiopatient and outpatient encounters is maintained. Coded digraphic changes, but without elevation of cardiac enzymes. agnoses are electronically entered into a medical diagnostic Acute myocardial infarction was considered present if at index, allowing easy identification of all patients with a least 2 of the 3 diagnostic criteria were fulfilled: compatible diagnosis of interest. An electronic electrocardiogram dataclinical presentation, diagnostic cardiac enzyme/biomarker base has been in use at Mayo since 1976, allowing rapid levels, and electrocardiographic changes consistent with identification of the coded interpretation and diagnoses for myocardial infarction. Coronary death was defined as death each electrocardiogram. All electrocardiograms can also be attributable to fatal acute myocardial infarction and death retrieved for direct review. with the primary cause being coronary heart disease (based on documentation per death certificates or physician adjudication by chart review). The ascertainment of outcomes Incident Atrial Fibrillation Cohort and was accomplished through comprehensive review of the Study Population medical records in addition to cross-referencing the multiple The medical records of adult residents of Olmsted County, administrative databases to identify any inconsistencies. Minnesota, who had a first AF documented between January 1, 1980, and December 31, 2000, in any of the Mayo administrative databases (medical index, surgical index, electrocardiographic, and echocardiographic databases) were reviewed and followed forward in the medical records to March 2004 by 1 cardiologist (Y. M.). For the purpose of this study, we excluded subjects who had any history or evidence of coronary heart disease and subjects who sustained a first coronary ischemic event on the same day as the first AF. Because the unit record system at Mayo dated back to the early 1900s, any electrocardiogram performed in any patient was contained within the unit record. Any AF that occurred before the establishment of administrative databases would have been identified from the chart review process. Final inclusion in the study population required the following: The AF was sustained, allowing confirmation by 12-lead electrocardiogram; the AF episode was verified as the first recognized AF event for the person; and there was
Death. The secondary outcome of interest was death. Ascertainment of death was accomplished through comprehensive review of the medical records, death certificates, Vital Status Information from Mayo Registration, Minnesota State Death Tapes, and Social Security Death Index. Statistical Analysis. Baseline characteristics were summarized by means and standard deviations or frequency percentages and compared between subjects with or without subsequent development of coronary ischemic events by using age-adjusted Cox proportional hazards analyses for both sexes. Average annual incidence rates were reported by sex and 10-year age interval, and summaries were tabulated as cases per 1000 person-years. The overall and sex-specific cumulative incidence of first coronary ischemic events after first AF was estimated using the Kaplan-Meier method. Incidence was assessed as a function of age and sex, using
Miyasaka et al
Risk of Coronary Ischemic Events in AF
a Cox proportional hazards model for time to first coronary ischemic event. Cox models based on clinical variables at the time of first AF were also developed using backward stepwise selection. The clinical covariates considered included type of AF (paroxysmal vs chronic) at initial diagnosis, body mass index, systolic or diastolic blood pressure, history of systemic hypertension, dyslipidemia, diabetes mellitus, smoking, peripheral vascular disease, stroke, congestive heart failure, valvular heart disease, chronic obstructive pulmonary disease, obstructive sleep apnea, regular alcohol use, hyperthyroidism, chronic renal disease, and malignancy. We also analyzed whether AF provides additional predictive information, beyond conventional risk factor consideration, for the development of subsequent coronary ischemic events. We compared our sex-specific coronary incidence rates, estimated by Kaplan-Meier analysis, with those predicted by the Framingham risk equation13 at 10 years. The association of first coronary ischemic event with subsequent survival was estimated using time-dependent proportional hazards models, controlling for age and sex, and with and without clinical variables. The same clinical
Table 1
359 covariates were considered, and the same selection method was used. The effect of interim coronary revascularization (percutaneous coronary intervention or coronary artery bypass grafting) on mortality was accounted for by including it as a time-dependent covariate in the model. Mortality ratios relative to the general Minnesota white population were calculated by dividing the observed number of mortality events by an “expected number” of deaths. This expected number was calculated as the sum of each individual’s cumulative hazard, based on the standard age, sex, and calendar-year specific life-table survival probability up to the individual’s follow-up time. Confidence intervals (CIs) were based on the Poisson distribution, and gender comparisons were based on a likelihood ratio chi-square test. All tests of significance were 2-tailed, and a P value less than .05 was considered statistically significant.
RESULTS A total of 4618 subjects (51% were men; mean age 73 years) were confirmed to have developed a first AF during 1980 to 2000. We excluded 1776 subjects (38%) because they had a history of coronary heart disease and 74 subjects
Baseline Characteristics of the Study Population Stratified by Subsequent Development of Coronary Ischemic Events Men
Women
Variable
No Events (n ⫽ 1091)
Had Events (n ⫽ 227)
No Events (n ⫽ 1214)
Had Events (n ⫽ 236)
Age (y) Paroxysmal AF, N (%) BMI (kg/m2) Systemic hypertension, N (%) Hypertensive therapy, N (%) Systolic BP (mm Hg) Diastolic BP (mm Hg) Dyslipidemia, N (%) Lipid-lowering therapy, N (%) Total cholesterol (mg/dL)§ HDL cholesterol (mg/dL)§ Diabetes mellitus, N (%) Smoking, N (%) Peripheral artery disease, N (%) Stroke, N (%) Prior CHF, N (%) VHD, N (%) COPD, N (%) Obstructive sleep apnea, N (%) Regular alcohol use, N (%) Hyperthyroidism, N (%) Chronic renal disease, N (%) History of malignancy, N (%)
66 ⫾ 18 797 (73) 27.3 ⫾ 5.4 691 (63) 373 (34) 133 ⫾ 18 78 ⫾ 11 344 (32) 19 (1.7) 191 ⫾ 47 45 ⫾ 14 125 (11) 762 (70) 72 (6.6) 79 (7.2) 54 (4.9) 142 (13) 245 (22) 29 (2.7) 225 (21) 9 (0.8) 137 (13) 314 (29)
69 ⫾ 13‡ 151 (67) 28.4 ⫾ 5.2† 172 (76)‡ 91 (40)† 138 ⫾ 20† 79 ⫾ 11 87 (38)† 4 (1.8) 198 ⫾ 45 44 ⫾ 12 37 (16)† 174 (77)* 19 (8.4)* 18 (7.9) 9 (4.0) 39 (17)* 60 (26) 9 (4.0)* 47 (21) 2 (0.9) 36 (16)† 46 (20)
76 ⫾ 14 825 (68) 26.5 ⫾ 6.9 961 (79) 613 (50) 139 ⫾ 21 78 ⫾ 11 329 (27) 20 (1.6) 204 ⫾ 45 55 ⫾ 17 143 (12) 424 (35) 80 (6.6) 91 (7.5) 89 (7.3) 264 (22) 187 (15) 6 (0.5) 72 (5.9) 21 (1.7) 99 (8.2) 364 (30)
77 ⫾ 11‡ 147 (62) 27.4 ⫾ 7.3* 215 (91)‡ 136 (58)‡ 147 ⫾ 19‡ 80 ⫾ 10 64 (27) 3 (1.3) 211 ⫾ 46 52 ⫾ 14 49 (21)‡ 79 (33) 30 (13)‡ 17 (7.2) 18 (7.6)* 58 (25) 36 (15) 2 (0.8) 9 (3.8) 5 (2.1) 26 (11)‡ 48 (20)
AF ⫽ atrial fibrillation; BMI ⫽ body mass index; BP ⫽ blood pressure; CHF ⫽ congestive heart failure; COPD ⫽ chronic obstructive pulmonary disease; HDL ⫽ high-density lipoprotein; VHD ⫽ valvular heart disease. Values are given as mean ⫾ standard deviation or number (percentage). P value for events status differences in each characteristics by Cox proportional hazards analysis after adjusting for age. *.01 ⬍ P ⬍.05 vs no events. †.001 ⬍ P ⬍.01 vs no events. ‡P ⬍.001 vs no events. §The available number of data were 1999 for total cholesterol and 1143 for HDL cholesterol.
360
The American Journal of Medicine, Vol 120, No 4, April 2007 confirmation (72 men [32%], 45 women [19%]), unstable angina (45 men [20%], 59 women [25%]), nonfatal acute myocardial infarction (66 men [29%], 79 women [34%]), and coronary death (44 men [19%], 53 women [22%]).
Incidence, Timing, and Predictors of Coronary Ischemic Events
Figure 1 The unadjusted incidence of first coronary ischemic events by time from first AF diagnosis among men (left) and women (right). AF ⫽ atrial fibrillation.
(2%) because they had a coronary ischemic event that occurred on the same day as the first AF. The remaining 2768 subjects (48% were men, mean age 71 years, range 18-107 years) constituted the study population. Of these, 463 patients (17%) had a total of 841 coronary ischemic events during a mean follow-up period of 6.0 ⫾ 5.2 years, and 1700 died. The proportion of our cohort followed to death or to within 2 years of the last follow-up of this study (March 2004) was 92.4%. The baseline characteristics of the study population, stratified by sex and subsequent coronary ischemic events, are shown in Table 1.
Distribution of Coronary Ischemic Events The type of first coronary ischemic events (227 men, 236 women) were as follows: angina pectoris with angiographic
The unadjusted incidence of coronary ischemic events after AF was 31 per 1000 person-years and was not significantly different between men and women (30.2 vs 31.4 per 1000 person-years; P ⫽ .68). The risk of new coronary ischemic events was greatest during the first year of AF (cumulative incidence 4.7%, 95% CI, 3.9-5.6) and plateaued thereafter (2.5% per year after first year, cumulative event rate 14.4% at 5 years, 95% CI, 12.8-15.9) (Figure 1) (Table 2). When adjusted for age, male sex was associated with increased risk of coronary ischemic events (hazard ratio [HR] 1.32, 95% CI, 1.09-1.60, P ⫽ .004). When age stratified, the incidence of coronary ischemic events increased throughout the age range in women but did not increase in men after age 85 years or more (Table 2). When a sex-specific correction for “age 85 years or more” was added to the model with a linear term for age and sex, there was a significant (P ⫽ .028) downward correction for men but no apparent departure from linearity for women (P ⫽ .06 for the difference between the corrections). The overall incidence of coronary ischemic events did not differ between men and women, likely because the mean age of women was 10 years older (76 vs 66 years). In a multivariable Cox model for the prediction of first coronary events (Table 3), advancing age, male sex, higher systolic blood pressure, and history of systemic hyperten-
Table 2 Age-and Sex-Specific Incidence Rates and Kaplan-Meier Estimated Cumulative Incidence of Coronary Ischemic Events for Patients Diagnosed with First Atrial Fibrillation Kaplan-Meier Estimated Cumulative Incidence
Overall Men ⬍55 y 55-64 y 65-74 y 75-84 y ⱖ85 y Overall Women ⬍55 y 55-64 y 65-74 y 75-84 y ⱖ85 y Overall
N
Person-y
Incidence Rate/1000 Person-y*
1 y, % (95% CI)
5 y, % (95% CI)
2768
15046.53
30.77 (463)
4.7 (3.9-5.6)
14.4 (12.8-15.9)
299 207 333 323 156 1318
2573.86 1428.25 1829.45 1321.64 371.87 7525.07
12.04 27.31 38.81 52.96 43.03 30.17
(31) (39) (71) (70) (16) (227)
0.7 2.5 8.4 6.3 5.0 4.8
(0.0-1.7) (0.3-4.7) (5.3-11.5) (3.3-9.2) (0.9-9.0) (3.5-6.0)
5.4 11.8 19.7 19.5 17.2 14.5
(2.5-8.2) (6.6-16.7) (14.7-24.4) (14.0-24.7) (7.7-25.7) (12.2-16.7)
101 139 288 533 389 1450
965.59 1058.37 1826.05 2589.81 1081.63 7521.45
10.36 21.73 29.02 35.52 53.62 31.38
(10) (23) (53) (92) (58) (236)
1.0 5.4 4.2 4.6 6.1 4.7
(0.0-2.9) (1.4-9.2) (1.7-6.6) (2.7-6.4) (3.5-8.7) (3.5-5.8)
3.2 10.2 10.9 16.1 22.7 14.3
(0.0-6.6) (4.5-15.6) (6.9-14.9) (12.3-19.7) (16.3-28.6) (12.1-16.4)
CI ⫽ confidence interval. *Data are presented as incidence rate followed in parentheses by the actual number of cases observed.
Miyasaka et al
Risk of Coronary Ischemic Events in AF
361
Table 3 Overall and Sex-Specific Multivariable-Adjusted Hazard Ratios for the Coronary Ischemic Events in Patients Diagnosed with First Atrial Fibrillation Overall
Age/10 y Men Systolic BP/10 mm Hg Systemic hypertension Diabetes mellitus Peripheral artery disease Valvular heart disease Obstructive sleep apnea Chronic renal disease
Men
Women
HR
95% CI
HR
95% CI
HR
95% CI
1.30‡ 1.37† 1.06* 1.65‡ 1.76‡ 1.56† 1.39† 1.97* 1.47†
1.20-1.42 1.13-1.67 1.01-1.12 1.23-2.21 1.38-2.24 1.14-2.13 1.11-1.75 1.07-3.64 1.11-1.95
1.32‡ 1.04 1.47* 1.50* 1.39 1.39 1.97 1.34
1.18-1.46 0.96-1.13 1.02-2.12 1.04-2.17 0.85-2.28 0.98-1.97 0.99-3.89 0.91-1.96
1.29‡ 1.07 2.23† 2.04‡ 1.67* 1.44* 2.43 1.79†
1.13-1.47 0.99-1.14 1.33-3.73 1.47-2.82 1.12-2.50 1.06-1.95 0.59-9.97 1.17-2.72
BP ⫽ blood pressure, HR ⫽ hazard ratio, CI ⫽ confidence interval. *.01 ⬍ P ⬍.05. †.001 ⬍ P ⬍.01. ‡P ⬍.001.
sion, diabetes mellitus, peripheral artery disease, valvular heart disease, obstructive sleep apnea, and chronic renal disease were significant independent predictors. The type of AF, paroxysmal or chronic AF, was not significant. Separate models for men and women including these same variables appeared similar (Table 3), and none of the interactions between the risk factors and the gender were significant. A prediction formula for the probability of coronary events within 1, 5, or 10 years based on this model is given in the Appendix, and Table 4 shows an array of sample calculations.
Gender Differences in the Risk of Coronary Events after First Atrial Fibrillation Although men have higher age-adjusted incidence of coronary ischemic events after first AF, how our sex-specific coronary incidence rates relate to those predicted by the Framingham risk equation is of interest. Restricting the analysis to the subgroup aged 30 to 74 years (N ⫽ 1403), the same age range for which the Framingham coronary risk equation was developed,13 the estimations of coronary event rates at 10 years were 22% (95% CI, 18-26) for men and 19% (95% CI, 14-23) for women by Kaplan-Meier analyses, versus 21% for men and 11% for women by Framingham risk equation.13
Table 4
Impact of Coronary Ischemic Events on Survival In time-dependent Cox analysis adjusted for age, the hazard for death was highly significantly associated with the occurrence of post-AF coronary ischemic events in each sex (P ⬍.0001). When an overall model was fit, with age, sex, coronary events, and an interaction between coronary events and sex, the HR associated with events was significantly higher in women (HR 2.99, 95% CI, 2.53-3.53 vs HR 2.33, 95% CI, 1.94-2.81, P ⫽ .044). Before the occurrence of a coronary ischemic event, the age-adjusted HR for men versus women was 1.23 (95% CI, 1.10-1.37), whereas after the occurrence of a coronary ischemic event, there was no significant difference in age-adjusted mortality risk. After adjustment for age, clinical risk factors, and coronary revascularization as additional time-dependent covariates, the occurrence of a coronary ischemic event remained a highly significant independent predictor of death in both sexes (men, HR 2.71, 95% CI, 2.22-3.32; women, HR 3.55, 95% CI, 2.99-4.22; P ⫽ .033). Compared with the general Minnesota white population, the relative mortality HRs before coronary ischemic events were 1.56 (95% CI, 1.47-1.66) in men and 1.71 (95% CI, 1.62-1.82) in women (P ⫽ .10). After the development of a first coronary ischemic event, the corresponding relative
Probability of Coronary Ischemic Event within 5 Years Using a Prediction Formula* (Appendix) Men
Women
Risk Factors
None
HTN
HTN ⫹ DM
HTN ⫹ DM ⫹ VHD
None
HTN
HTN ⫹ DM
HTN ⫹ DM ⫹ VHD
Age 60 y Age 70 y Age 80 y
7% 9% 11%
11% 14% 17%
18% 23% 29%
24% 30% 37%
5% 6% 8%
8% 10% 13%
13% 17% 22%
18% 23% 29%
HTN ⫽ systemic hypertension; DM ⫽ diabetes mellitus; VHD ⫽ valvular heart disease. *Imputing the mean value of systolic blood pressure (137 mm Hg) for these estimations.
362 Table 5
The American Journal of Medicine, Vol 120, No 4, April 2007 Mortality Ratios Relative to the General Minnesota White Population after Atrial Fibrillation Diagnosis Men
Overall Pre-events Post-events
Women
Observed*
Expected†
Ratio‡ (95% CI)
Observed*
Expected†
Ratio‡ (95% CI)
710 563 147
411.67 361.11 50.56
1.72 (1.62-1.84) 1.56 (1.47-1.66) 2.91 (2.35-3.66)
990 806 184
516.40 470.99 45.41
1.92 (1.81-2.04)§ 1.71 (1.62-1.82) 4.05 (3.23-5.36)储
CI ⫽ confidence interval. *Observed is number of deaths in the absence of coronary ischemic events (pre-events) or occurring after coronary ischemic events (post-events). †Expected is the cumulative hazard (based on age and sex using Minnesota white population) for follow-up before or in the absence of coronary ischemic events (pre-events) or after the occurrence of coronary ischemic events (post-events). ‡Ratio of observed to expected represents hazard ratio relative to the general population. §.01 ⬍ P ⬍.05 vs men. 储P ⬍.001 vs men.
HRs were 2.91 (95% CI, 2.35-3.66) in men and 4.05 (95% CI, 3.23-5.36) in women (P ⬍.001) (Table 5).
DISCUSSION In our community-based cohort, the first AF marks a high risk for new coronary ischemic events in both men and women. In contrast with men, AF conferred an additional risk of ischemic events in women beyond conventional risk prediction. The excess mortality risk associated with the development of new coronary ischemic events was greater in women.
Incidence of Coronary Ischemic Events After First Atrial Fibrillation Coronary heart disease has been regarded as an independent risk factor of AF,14,15 although the data supporting this connection came from only a small handful of studies, and the event rate was modest.16,17 In the Framingham Heart Study, the 2-year age-adjusted incidence for chronic AF among patients with coronary heart disease, per 1000, was 8.9 for men and 2.3 for women.16 However, little is known regarding AF as a precursor of new coronary ischemic events in patients without prior evidence of coronary heart disease. Our study showed that the ischemic event rate was high after manifestation of first AF, and that the unadjusted incidence did not differ between men and women. Incidence, when adjusted for age, was higher in men. In fact, the incidence of post-AF coronary ischemic events was higher than that of post-AF stroke in the same cohort (10% at 5 years).18 The precise reasons for the high ischemic event rate after first AF among patients with no prior coronary heart disease could not be well delineated, but a likely explanation is that persons who develop AF frequently harbor the same cardiovascular risk factors for coronary ischemic events. In our cohort, more than 70% had hypertension, 30% had dyslipidemia, and 13% had diabetes. It is unclear why some develop ischemic events first and others develop AF first.
In a multivariate Cox model, higher systolic blood pressure, systemic hypertension, diabetes mellitus, and obstructive sleep apnea were significant independent predictors of new coronary ischemic events after first AF. Early diagnosis and control of these treatable conditions may have implications on cardiovascular outcomes in patients who have AF. Although the age-adjusted incidence of new coronary events post-AF was higher in men; AF actually conferred an additional risk (⬃70%) of new coronary events in women only, based on the comparison of our KaplanMeier 10-year event rates and those predicted by the Framingham risk equation. Our data set does not lend itself readily to analyses for potential explanatory factors. The underlying reasons for this gender difference in the impact of AF on subsequent coronary risk warrant further investigation.
Impact of Coronary Ischemic Events on Survival AF is known to be associated with excess mortality risk.14,19-22 Our study showed that the development of a first coronary ischemic event further increased the mortality risk, even after multivariable adjustment. In addition, the mortality risk relative to the general population was higher in women than in men. This was similar to our previous finding of greater relative mortality risk in women associated with post-AF stroke.18 The reasons for these differences could not be determined in this study but clearly require further studies.
LIMITATIONS There were inherent biases associated with the retrospective design. It is possible that the incidence of AF may have been underestimated, given that some patients may not have received care at the Mayo Clinic, although the underestimation would likely be small because the Mayo Clinic is the principal health care provider and referral center for Olmsted County.12 Also, we may have under-
Miyasaka et al
Risk of Coronary Ischemic Events in AF
estimated the coronary event rates because we did not include those who had presumed angina but did not have angiographic confirmation. The population of Olmsted County, Minnesota, is predominantly white and not racially ethnically diverse.
CONCLUSIONS Patients diagnosed with a first AF constitute a high-risk group for subsequent new coronary ischemic events. Although the age-adjusted incidence of such post-AF coronary ischemic events was higher in men, AF actually conferred an additional risk for the development of ischemic events, beyond conventional risk prediction, in women only. Further, the excess mortality risk associated with the development of post-AF new coronary events was greater in women. See appendix online at http://www.amjmed.com
References 1. 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. Braunwald E. Shattuck lecture— cardiovascular medicine at the turn of the millennium: triumphs, concerns, and opportunities. N Engl J Med. 1997;337:1360-1369. 3. Alpert JS. Atrial fibrillation: a growth industry in the 21st century. Eur Heart J. 2000;21:1207-1208. 4. Whisnant JP, Wiebers DO, O’Fallon WM, et al. A population-based model of risk factors for ischemic stroke: Rochester, Minnesota. Neurology. 1996;47:1420-1428. 5. Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent risk factor for stroke: the Framingham Study. Stroke. 1991;22:983988. 6. Scheinman MM. Atrial fibrillation and congestive heart failure: the intersection of two common diseases. Circulation. 1998;98:941-942. 7. Wang TJ, Larson MG, Levy D, et al. Temporal relations of atrial fibrillation and congestive heart failure and their joint influence on mortality: the Framingham Heart Study. Circulation. 2003;107:29202925. 8. Kilander L, Andren B, Nyman H, et al. Atrial fibrillation is an independent determinant of low cognitive function: a cross-sectional study in elderly men. Stroke. 1998;29:1816-1820.
363 9. Ott A, Breteler MM, de Bruyne MC, et al. Atrial fibrillation and dementia in a population-based study. The Rotterdam Study. Stroke. 1997;28:316-321. 10. Zimetbaum PJ, Josephson ME, McDonald MJ, et al. Incidence and predictors of myocardial infarction among patients with atrial fibrillation. J Am Coll Cardiol. 2000;36:1223-1227. 11. Friedman HZ, Weber-Bornstein N, Deboe SF, Mancini GB. Cardiac care unit admission criteria for suspected acute myocardial infarction in new-onset atrial fibrillation. Am J Cardiol. 1987;59:866-869. 12. Melton LJ 3rd. History of the Rochester Epidemiology Project. Mayo Clin Proc. 1996;71:266-274. 13. Wilson PW, D’Agostino RB, Levy D, et al. Prediction of coronary heart disease using risk factor categories. Circulation. 1998;97:18371847. 14. Krahn AD, Manfreda J, Tate RB, et al. The natural history of atrial fibrillation: incidence, risk factors, and prognosis in the Manitoba Follow-Up Study. Am J Med. 1995;98:476-484. 15. Tsang TS, Barnes ME, Bailey KR, et al. Left atrial volume: important risk marker of incident atrial fibrillation in 1655 older men and women. Mayo Clin Proc. 2001;76:467-475. 16. Kannel WB, Abbott RD, Savage DD, McNamara PM. Coronary heart disease and atrial fibrillation: the Framingham Study. Am Heart J. 1983;106:389-396. 17. Orndahl G, Thulesius O, Hood B. Incidence of persistent atrial fibrillation and conduction defects in coronary heart disease. Am Heart J. 1972;84:120-131. 18. Miyasaka Y, Barnes ME, Gersh BJ, et al. Time trends of ischemic stroke incidence and mortality in patients diagnosed with first atrial fibrillation in 1980 to 2000. Report of a Community-Based Study. Stroke. 2005;36:2362-2366. 19. Benjamin EJ, Wolf PA, D’Agostino RB, et al. Impact of atrial fibrillation on the risk of death: the Framingham Heart Study. Circulation. 1998;98:946-952. 20. Stewart S, Hart CL, Hole DJ, McMurray JJ. A population-based study of the long-term risks associated with atrial fibrillation: 20-year follow-up of the Renfrew/Paisley study. Am J Med. 2002;113:359-364. 21. Vidaillet H, Granada JF, Chyou PH, et al. A population-based study of mortality among patients with atrial fibrillation or flutter. Am J Med. 2002;113:365-370. 22. Gajewski J, Singer RB. Mortality in an insured population with atrial fibrillation. JAMA. 1981;245:1540-1544. 23. McKee PA, Castelli WP, McNamara PM, Kannel WB. The natural history of congestive heart failure: the Framingham study. N Engl J Med. 1971;285:1441-1446.
363.e1
APPENDIX Definition of Covariates Chronic versus paroxysmal AF was defined by whether there were recognizable intervening episodes of sinus rhythm. Systemic hypertension was defined by a physician’s diagnosis, need for antihypertensive therapy, or systolic blood pressure greater than 140 mm Hg or diastolic blood pressure greater than 90 mm Hg on at least 2 occasions that were not associated with acute illness or injury. Dyslipidemia was defined by a total cholesterol level of 200 mg/dL or greater, triglyceride level of 150 mg/dL or greater, low-density lipoprotein cholesterol level of 130 mg/dL or greater or high-density lipoprotein cholesterol less than 40 mg/dL on 2 or more occasions, or treatment with lipidlowering agents. Diabetes mellitus was defined by physician’s diagnosis and treatment with insulin or oral hypoglycemic agents. Smoking history was classified as past (⬎6 months prior) or current smoker. Stroke included development of any type of stroke, as defined by clinical documentation of the diagnosis with or without confirmatory findings on imaging studies. Congestive heart failure was defined by the presence of 2 major or 1 major and 2 minor Framingham
The American Journal of Medicine, Vol 120, No 4, April 2007 criteria.23 Valvular heart disease was defined by greater than mild stenosis or regurgitation by echocardiography or prior valve repair/replacement. Regular alcohol use was defined by self-reported consumption of more than 1 drink per day regularly. Chronic renal disease, chronic obstructive pulmonary disease, obstructive sleep apnea, and hyperthyroidism were defined by these clinical diagnoses in the medical records.
Prediction Formula for the Probability of Coronary Ischemic Event The proportional-hazards model-based probability of having had a coronary ischemic event within year 1, 5, or 10 of first AF diagnosis, as defined by the multivariate model in Table 3, is estimated as follows: PROB ⫽ 1 ⫺ (Pi)eXB, where XB ⫽ 0.265*(age/10) ⫹ 0.316*{1 if male} ⫹ 0.059*(systolic blood pressure/10) ⫹ 0.500*{1 if systemic hypertension} ⫹ 0.563*{1 if diabetes mellitus} ⫹ 0.446*{1 if peripheral artery disease} ⫹ 0.332*{1 if valvular heart disease} ⫹ 0.679*{1 if obstructive sleep apnea} ⫹ 0.385*{1 if chronic renal disease} ⫺ 3.450, and (P1, P5, P10) ⫽ (.9608, .8682, .7467).