PROGRESS IN CARDIOLOGY
Diabetes, fibrinogen, and risk of cardiovascular disease: The Framingham experience William B. Kannel, MD, MPH,” Ralph B. D’Agostino, PhD,b Peter W. F. Wilson, MD,” Albert J. Belanger, MS,b and David R. Gagnon, MDd Boston, Muss.
Three decades of epidemiologic research from the Framingham Study have shown diabetes to be a powerful predisposing factor for cardiovascular disease in general, and of coronary disease in particular.lM6 Although coronary disease risk factors occur in excess in diabetics, impaired glucose tolerance exerts an influence independent of the major identified coronary risk factors. This unique effect could be related to enhanced thrombogenic factors, such as fibrinogen, in the diabetic. Epidemiologic studies, including those at Framingham, have implicated fibrinogen in the occurrence of cardiovascular disease.7s The purpose of this study is to examine the joint role of impaired glucose tolerance and fibrinogen in the occurrence of cardiovascular disease in general, and coronary disease in particular. The influence of diabetes will be examined and fibrinogen will be taken into account on the basis of the 16 years of follow-up since exam 10, when fibrinogen was measured. METHODOLOGY
At the tenth biennial examination, there were 3595 surviving subjects remaining in the Framingham cohort. The method of fibrinogen determination was perfected in the Framingham Study laboratory for these subjects only after the midpoint of the examination cycle. Thus only 1499 subjects had satisfactory fibrinogen values determined: 662 men and 837 women who ranged in age from 47 to 79 years. Of From %Section of Preventive Medicine and Epidemiology, Boston University Medical Center; bDepartment of Mathematics, Boston University; CNational Heart, Lung, and Blood Institute, Framingham Heart Study; and dthe Department of Neurology, Boston University Medical Center. Supported by the National Heart, Lung, and NIH-NOl-HV-92922 and NIH-NOl-HV-52971, Foundation, and by Merck, Sharp & Dohme. Received
for publication
Reprint requests: William Medicine and Epidemiology, tar’s Office Building, Suite
4/l/21978
672
Feb.
22, 1990;
accepted
Blood Institute Grant nos. by The Charles A. Dana Apr.
12, 1990.
B. Kannel, MD, MPH, Section of Preventive Boston University School of Medicine, DOC1105, 720 Harrison Ave., Boston, MA 02218.
these, 108 men and 79 women had already experienced a cardiovascular event, which left 554 men and 760 women at risk of a first cardiovascular occurrence in relation to fibrinogen and diabetic status. Fibrinogen was measured with the Swain modification of the method of Ratnoff and Menzie.‘O Diabetes was considered to be present when the patient was so diagnosed by a physician, had an abnormal glucose tolerance test result, or was on insulin or hypoglycemic agents. The cardiovascular disease end points have been determined by biennial clinical examinations, hospital protocol summaries, and medical examiner reports; all have been subjected to a review by a panel of investigators with criteria previously reported.2 Other variables that were examined in this study were major cardiovascular risk factors and included: (1) age (years), (2) serum cholesterol (mg/lOO ml), (3) systolic blood pressure (mm Hg), (4) cigarette consumption (cigarettes per day), (5) left ventricular hypertrophy (LVH) by ECG (absent, doubtful, or definitely present), (6) glucose intolerance (present or absent), (7) glycosuria (definitely present vs absent or doubtful), (8) HDL cholesterol (mg/lOO ml). All variables were measured at exam 10, and if the measurement was unavailable for exam 10, values that were recorded at exam 9 were used. Cardiovascular events that were examined in this study include coronary heart disease, stroke (including transient ischemic attacks), intermittent claudication, and congestive heart failure. Coronary heart disease events that were examined included myocardial infarction, angina pectoris, sudden death, and all coronary mortality. Statistics: The prevalence rate of a risk factor was computed as the ratio of the number of subjects with the particular risk factor to the total number of subjects free of cardiovascular disease and who had fibrinogen measured. These are presented as percentages. Incidence rates for cardiovascular diseases were computed as the ratio of new cases over the 16
Volume
120
Number
3
years of follow-up to the total number of subjects at risk. For comparison purposes these were adjusted by the direct age adjustment method and are presented as rates per thousand. The Chi square trends test for proportions was used to test for significant trends in incidence rates. Cox proportional hazard regressions were used to examine the effect of diabetes on the occurrence of cardiovascular events. Separate analyses were performed for each sex, and the effect of diabetes on cardiovascular events was examined in age-adjusted, standard risk-factor-adjusted (age, serum cholesterol, systolic blood pressure, cigarette consumption, and LVH by ECG) and fibrinogen plus standard risk-factor-adjusted models. Statistical significance was determined at the 0.05 level of significance with a two-tailed test.
Diabetes/fibrinogen
673
Table I. Percentage prevalence of cardiovascular risk factors Framingham Heart Study, exam 10 Risk factors
Men
Women
30.5 41.1
29.6 44.5
20.0 21.8
20.3 38.3
14.9 12.9
8.4 13.9
55.4 39.5
41.4 54.6
44.2 25.0
43.7 16.3
0.9 2.4
0.5 2.4
24.0 21.8
28.0 36.8
Definite hypertension 47-64 65-79
Hypercholesterolemia 47-64 65-79
Glucose intolerance 47-64 65-79
Obesity 47-64 65-79
Cigarette smoking 47-64 65-79
Definite ECG LVH 47-64 65-79
OBSERVATIONS
Over the 16 years of follow-up beyond the tenth biennial examination of the 1314 subjects who had satisfactory fibrinogen measurements and were free of cardiovascular disease at exam 10, 408 men and women developed some manifestations of cardiovascular disease, and 141 men and 118 women developed coronary heart disease in particular. Furthermore, of these 1314 subjects at the tenth exam, 508 men and 717 women aged 47 to ‘79 years had satisfactory fibrinogen determinations and were free of diabetes, and 46 men and 43 women had developed diabetes. There were also 56 men and 53 women who had blood sugar levels that exceeded 120 mg/dl. Impaired glucose tolerance was one of the less common cardiovascular risk factors and occurred with similar prevalence in both sexes. By exam 10, it had affected 14.9% and 12.9% of men aged 47 to 64 and 65 to 79 years, respectively (Table I). In women, it affected 8.4% and 13.9%) respectively (Table I). Analysis of secular trends in the Framingham cohort suggests that at any age, diabetes is more common now than it was before. The most common lethal sequela of diabetes was atherosclerotic large-vessel disease. Diabetes predisposed subjects to all of the major atherosclerotic diseases, and coronary heart disease was the most common and most lethal (Table II). On the basis of the 16 years of follow-up after exam 10 of the Framingham Cohort Study, it is evident that diabetes is a powerful predisposing factor for coronary heart disease (Table II). For men, the age-adjusted risk ratio is 1.73, and for women it is 2.5, where the risk ratio compares those with diabetes to those without. The accepted criteria for diabetes have changed
and cardiovascular risk
Tachychardia 47-64 65-79
Definitions for prevalences:
Definite hypertension: BP >160/95 mm Hg; Hypercholesterolemia: Serum cholesterol >250 mg/dl; Glucose intolerance: blood glucose >120 mg/dl, glycosuria or diabetes mellitus; Obesity: metropolitan relative weight >120; Cigarette smoking: any at exam 10; Tachychardia: heart rate >84/min.
Table II. Age-adjusted relative risk of clinical manifestations of cardiovascular disease for diabetics versus nondiabetics 16-year follow-up, Framingham Heart Study Age-adjusted Manifestation
of CVD
Any CVD event CVA Intermittent claudication Cardiac failure Coronary heart disease Myocardial infarction Angina pectoris Sudden death Coronary mortality
risk
ratio
Men
Women
2.31* 1.51 5.27* 2.55t 1.73t 2.16% 1.23 2.51s 2.3q
2.47* 1.82 2.60.t 4.92* 2.50% 4.37* 1.59 3.60t
Based on 554 men (46 diabetic) and 760 women (43 diabetic) who were free of cardiovascular disease at exam 10 whose fibrinogen was measured. Age adjusted with the Cox proportional hazards model. *p < 0.001. tp < 0.05. tp < 0.01.
$p < 0.1.
over the four decades of the Framingham Study. However, it is clear that impaired glucose intolerance, no matter how defined, predicts coronary heart disease and is associated with an increased risk (Ta-
September
674
Kannel
et al.
American
III. Age-adjusted risk of coronary heart diseaseby indicator ofglucoseintolerance 30-yearfollow-up, Framingham Heart Study Table
Age-adjusted
risk
ratio
Manifestation of glucose intolerance
Men
Women
Diabetes Glucose intolerance Glycosuria
1.73* 1.56* 3.47t
2.50t 2.38$ 2.36
Based on 554 men (46 diabetic) and 760 women (43 diabetic) who were free of cardiovascular disease at exam 10 and whose fibrinogen was measured. Age adjusted with the Cox proportional hazards model. *p
Diabetes, fibrinogen, and risk of cardiovascular disease: The Framingham experience William B. Kannel, MD, MPH,” Ralph B. D’Agostino, PhD,b Peter W. F. Wilson, MD,” Albert J. Belanger, MS,b and David R. Gagnon, MDd Boston, Muss.
Three decades of epidemiologic research from the Framingham Study have shown diabetes to be a powerful predisposing factor for cardiovascular disease in general, and of coronary disease in particular.lM6 Although coronary disease risk factors occur in excess in diabetics, impaired glucose tolerance exerts an influence independent of the major identified coronary risk factors. This unique effect could be related to enhanced thrombogenic factors, such as fibrinogen, in the diabetic. Epidemiologic studies, including those at Framingham, have implicated fibrinogen in the occurrence of cardiovascular disease.7s The purpose of this study is to examine the joint role of impaired glucose tolerance and fibrinogen in the occurrence of cardiovascular disease in general, and coronary disease in particular. The influence of diabetes will be examined and fibrinogen will be taken into account on the basis of the 16 years of follow-up since exam 10, when fibrinogen was measured. METHODOLOGY
At the tenth biennial examination, there were 3595 surviving subjects remaining in the Framingham cohort. The method of fibrinogen determination was perfected in the Framingham Study laboratory for these subjects only after the midpoint of the examination cycle. Thus only 1499 subjects had satisfactory fibrinogen values determined: 662 men and 837 women who ranged in age from 47 to 79 years. Of From %Section of Preventive Medicine and Epidemiology, Boston University Medical Center; bDepartment of Mathematics, Boston University; CNational Heart, Lung, and Blood Institute, Framingham Heart Study; and dthe Department of Neurology, Boston University Medical Center. Supported by the National Heart, Lung, and NIH-NOl-HV-92922 and NIH-NOl-HV-52971, Foundation, and by Merck, Sharp & Dohme. Received
for publication
Reprint requests: William Medicine and Epidemiology, tar’s Office Building, Suite
4/l/21978
672
Feb.
22, 1990;
accepted
Blood Institute Grant nos. by The Charles A. Dana Apr.
12, 1990.
B. Kannel, MD, MPH, Section of Preventive Boston University School of Medicine, DOC1105, 720 Harrison Ave., Boston, MA 02218.
these, 108 men and 79 women had already experienced a cardiovascular event, which left 554 men and 760 women at risk of a first cardiovascular occurrence in relation to fibrinogen and diabetic status. Fibrinogen was measured with the Swain modification of the method of Ratnoff and Menzie.‘O Diabetes was considered to be present when the patient was so diagnosed by a physician, had an abnormal glucose tolerance test result, or was on insulin or hypoglycemic agents. The cardiovascular disease end points have been determined by biennial clinical examinations, hospital protocol summaries, and medical examiner reports; all have been subjected to a review by a panel of investigators with criteria previously reported.2 Other variables that were examined in this study were major cardiovascular risk factors and included: (1) age (years), (2) serum cholesterol (mg/lOO ml), (3) systolic blood pressure (mm Hg), (4) cigarette consumption (cigarettes per day), (5) left ventricular hypertrophy (LVH) by ECG (absent, doubtful, or definitely present), (6) glucose intolerance (present or absent), (7) glycosuria (definitely present vs absent or doubtful), (8) HDL cholesterol (mg/lOO ml). All variables were measured at exam 10, and if the measurement was unavailable for exam 10, values that were recorded at exam 9 were used. Cardiovascular events that were examined in this study include coronary heart disease, stroke (including transient ischemic attacks), intermittent claudication, and congestive heart failure. Coronary heart disease events that were examined included myocardial infarction, angina pectoris, sudden death, and all coronary mortality. Statistics: The prevalence rate of a risk factor was computed as the ratio of the number of subjects with the particular risk factor to the total number of subjects free of cardiovascular disease and who had fibrinogen measured. These are presented as percentages. Incidence rates for cardiovascular diseases were computed as the ratio of new cases over the 16
Volume
120
Number
3
years of follow-up to the total number of subjects at risk. For comparison purposes these were adjusted by the direct age adjustment method and are presented as rates per thousand. The Chi square trends test for proportions was used to test for significant trends in incidence rates. Cox proportional hazard regressions were used to examine the effect of diabetes on the occurrence of cardiovascular events. Separate analyses were performed for each sex, and the effect of diabetes on cardiovascular events was examined in age-adjusted, standard risk-factor-adjusted (age, serum cholesterol, systolic blood pressure, cigarette consumption, and LVH by ECG) and fibrinogen plus standard risk-factor-adjusted models. Statistical significance was determined at the 0.05 level of significance with a two-tailed test.
Diabetes/fibrinogen
673
Table I. Percentage prevalence of cardiovascular risk factors Framingham Heart Study, exam 10 Risk factors
Men
Women
30.5 41.1
29.6 44.5
20.0 21.8
20.3 38.3
14.9 12.9
8.4 13.9
55.4 39.5
41.4 54.6
44.2 25.0
43.7 16.3
0.9 2.4
0.5 2.4
24.0 21.8
28.0 36.8
Definite hypertension 47-64 65-79
Hypercholesterolemia 47-64 65-79
Glucose intolerance 47-64 65-79
Obesity 47-64 65-79
Cigarette smoking 47-64 65-79
Definite ECG LVH 47-64 65-79
OBSERVATIONS
Over the 16 years of follow-up beyond the tenth biennial examination of the 1314 subjects who had satisfactory fibrinogen measurements and were free of cardiovascular disease at exam 10, 408 men and women developed some manifestations of cardiovascular disease, and 141 men and 118 women developed coronary heart disease in particular. Furthermore, of these 1314 subjects at the tenth exam, 508 men and 717 women aged 47 to ‘79 years had satisfactory fibrinogen determinations and were free of diabetes, and 46 men and 43 women had developed diabetes. There were also 56 men and 53 women who had blood sugar levels that exceeded 120 mg/dl. Impaired glucose tolerance was one of the less common cardiovascular risk factors and occurred with similar prevalence in both sexes. By exam 10, it had affected 14.9% and 12.9% of men aged 47 to 64 and 65 to 79 years, respectively (Table I). In women, it affected 8.4% and 13.9%) respectively (Table I). Analysis of secular trends in the Framingham cohort suggests that at any age, diabetes is more common now than it was before. The most common lethal sequela of diabetes was atherosclerotic large-vessel disease. Diabetes predisposed subjects to all of the major atherosclerotic diseases, and coronary heart disease was the most common and most lethal (Table II). On the basis of the 16 years of follow-up after exam 10 of the Framingham Cohort Study, it is evident that diabetes is a powerful predisposing factor for coronary heart disease (Table II). For men, the age-adjusted risk ratio is 1.73, and for women it is 2.5, where the risk ratio compares those with diabetes to those without. The accepted criteria for diabetes have changed
and cardiovascular risk
Tachychardia 47-64 65-79
Definitions for prevalences:
Definite hypertension: BP >160/95 mm Hg; Hypercholesterolemia: Serum cholesterol >250 mg/dl; Glucose intolerance: blood glucose >120 mg/dl, glycosuria or diabetes mellitus; Obesity: metropolitan relative weight >120; Cigarette smoking: any at exam 10; Tachychardia: heart rate >84/min.
Table II. Age-adjusted relative risk of clinical manifestations of cardiovascular disease for diabetics versus nondiabetics 16-year follow-up, Framingham Heart Study Age-adjusted Manifestation
of CVD
Any CVD event CVA Intermittent claudication Cardiac failure Coronary heart disease Myocardial infarction Angina pectoris Sudden death Coronary mortality
risk
ratio
Men
Women
2.31* 1.51 5.27* 2.55t 1.73t 2.16% 1.23 2.51s 2.3q
2.47* 1.82 2.60.t 4.92* 2.50% 4.37* 1.59 3.60t
Based on 554 men (46 diabetic) and 760 women (43 diabetic) who were free of cardiovascular disease at exam 10 whose fibrinogen was measured. Age adjusted with the Cox proportional hazards model. *p < 0.001. tp < 0.05. tp < 0.01.
$p < 0.1.
over the four decades of the Framingham Study. However, it is clear that impaired glucose intolerance, no matter how defined, predicts coronary heart disease and is associated with an increased risk (Ta-
September
674
Kannel
et al.
American
III. Age-adjusted risk of coronary heart diseaseby indicator ofglucoseintolerance 30-yearfollow-up, Framingham Heart Study Table
Age-adjusted
risk
ratio
Manifestation of glucose intolerance
Men
Women
Diabetes Glucose intolerance Glycosuria
1.73* 1.56* 3.47t
2.50t 2.38$ 2.36
Based on 554 men (46 diabetic) and 760 women (43 diabetic) who were free of cardiovascular disease at exam 10 and whose fibrinogen was measured. Age adjusted with the Cox proportional hazards model. *p
