FAMILY HISTORY OF DIABETES AND CARDIOVASCULAR DISEASE RISK FACTORS AND MORTALITY AMONG EUGLYCEMIC, BORDERLINE HYPERGLYCEMIC, AND DIABETIC ADULTS 1

Transcription

1 AMERICAN JOURNAL OF EPIDEMIOLOGY Vol. 125, No. 6 Copyright 1987 by The John* Hopkins University School of Hygiene and Public Health Printed in U.SA. All rights reserved FAMILY HISTORY OF DIABETES AND CARDIOVASCULAR DISEASE RISK FACTORS AND MORTALITY AMONG EUGLYCEMIC, BORDERLINE HYPERGLYCEMIC, AND DIABETIC ADULTS 1 DEBORAH L. WINGARD AND ELIZABETH BARRETT-CONNOR Wingard, 0. L (Dept of Community and Medicine, U. of California, San Diego, La Jolla, CA 92093) and E. Barrett-Connor. of diabetes and cardiovascular disease risk factors and mortality among euglycemlc, borderline hyperglycemic, and diabetic adults. Am J Epidemiol 1987; 125: In a prospective population-based study begun in 1972 in Rancho Bernardo, California, the association of cardiovascular disease risk factors, at baseline, with a family of diabetes and subsequent death from cardiovascular disease was investigated among 3,081 euglycemic, 1,290 borderline hyperglycemic, and 347 diabetic adults between 20 and 79 years of age. The main difference in risk factor distribution was in diabetic men and women years of age, of whom those with a family of diabetes were older and had higher levels of fasting plasma cholesterol and triglyceride. In addition, in all age-sexdiabetes status groups, those with a family of diabetes Included a greater proportion with a family of heart attack. In five out of 12 groups, the association was statistically significant. Significant independent predictors of cardiovascular disease mortality in euglycemic and borderline hyperglycemic adults were age, sex, cholesterol, and systolic blood pressure, while fasting plasma glucose was a significant predictor in eugtycemic adults only. of heart attack was significantly associated with cardiovascular disease mortality In eugtycemic and borderline hyperglycemic adults. of diabetes was not significantly associated with mortality risk in any of the three groups, but the risk followed an increasing trend from euglycemic (0.89) to borderline hyperglycemic (1.17) to diabetec (1.31) adults. cardiovascular diseases; diabetes mellltus A number of studies have reported that Of interest is whether persons with a family non-insulin-dependent diabetes has a of diabetes are more Likely to have strong familial (1-4) and probably genetic cardiovascular disease risk factors or an component (2-5), diabetic adults have an increased risk of cardiovascular disease increased risk of cardiovascular disease (6, mortality, independent of any other risk 7), and cardiovascular disease risk factors factor differences. are more common in diabetic adults (6, 7). Only a few studies have addressed these questions, usually indirectly. For example, r> ~J t KI- * AM u o,, Q QC J Wykeham Balme and Cole (8) reported in Received for publication March 31, 1986 and in J.,,.,,. f finalformaugust 18, tnat a family of diabetes oc- 1 Department of Community and Medicine, CUTred with about equal frequency in dia- School of Medicine, University of California, San ^tics with and without hypertension, SUg- Diego, La Jolla, CA (Reprint requests to Dr..,.,,.,., Deborah L. Wingard.) gesting that a family of diabetes Thisresearchwas supported by the National Insti- does not increase one's risk of hypertension tute of Arthritis>. Diabetesand Digestive and Kidney th t ag^ated ^th developing dia- Diseases Grant RO1-AM , The authors thank Janice McPhillips for her as- betes itself. In 1975, Brunzell et al. (9) sistance with the statistical analyses. reported that diabetes occurred with equal 948

2 FAMILY HISTORY OF DIABETES AND CARDIOVASCULAR DISEASE 949 frequency in nonnolipidemic and hyperlipidemic relatives of hypertriglyceridemic index patients, suggesting that diabetes and hypertriglyceridemia are inherited separately. Schneider and Pastow (10) reported in 1982 that a sample of diabetics in Germany with a family of diabetes showed a cardiovascular risk profile identical to that of diabetics without a family. Ganda et al. (11) reported in 1985 that mean concentrations of total, low density, and high density lipoprotein cholesterol and of triglycerides did not differ between persons with normal glucose tolerance tests who had no family of diabetes and those whose mother and father had diabetes. However, women with mild glucose intolerance, whose parents had diabetes, had significantly elevated lipid levels. In this report, we examine the distribution of cardiovascular disease risk factors among euglycemic, borderline hyperglycemic, and diabetic adults in Rancho Bernardo, California, according to the presence or absence of a family of diabetes. The independent association of a family of diabetes with cardiovascular disease mortality over a 12-year period is also assessed. MATERIALS AND METHODS Between 1972 and 1974, 82 per cent of the adult residents of a white, upper-middle class community of Rancho Bernardo, California, participated in a heart disease risk factor survey as part of the baseline visit of the Lipid Research Clinics Prevalence Study. Participants were asked to come to the clinic between seven A.M. and 11 A.M. after a 12-hour fast. At the visit, participants completed standardized questionnaires which included one question about a family of diabetes in any firstdegree relatives (parents, siblings, and/or children). Also included were questions regarding a personal of diabetes and cardiovascular disease, cigarette smoking, and current use of medications for diabetes, hyperlipidemia, and hypertension. Blood pressure was measured with a standard sphygmomanometer after the subject had been seated at least five minutes and was repeated in those whose initial reading exceeded 160/90 mmhg, the lower of the two readings being recorded. Height in inches and weight in pounds were measured in light clothing without shoes; obesity was estimated with the body mass index (weight/height 2 ). Fasting plasma cholesterol and triglyceride levels were determined in a standardized Lipid Research Clinics laboratory with an Auto-analyzer (Technicon Instruments Corp., Tarrytown, NY). Fasting plasma glucose was determined by the hexokinase method for true glucose in a routine hospital diagnostic laboratory. Approximately 99.5 per cent of the original study participants have been followed for vital status for 12 years (12). Death certificates have been obtained for all decedents and coded by a certified nosologist according to the Eighth Revision of the International Classification of Diseases, Adapted. In one third of the cohort, underlying cause of death attributed to cardiovascular disease (ICDA-8 codes ) was validated by interview and medical record review, with 85 per cent agreement with the death certificate diagnosis. For this analysis, subjects were classified as diabetic, borderline hyperglycemic, or euglycemic, since the effect of family of diabetes might vary by diabetic status. Diabetes was defined by self-reported personal of diabetes, and/or by use of medication for hyperglycemia, and/or by fasting plasma glucose levels greater than or equal to 140 mg/dl. Physicians validated a diagnosis of self-report of diabetes in 79 per cent of a subset of 96 participants who reported a personal of diabetes. Subjects with fasting plasma glucose levels of 110 mg/dl or greater and less than 140 mg/dl and no personal of diabetes or current use of drugs for hyperglycemia were classified as borderline hyperglycemic. Those subjects with fasting plasma glucose levels less than 110 mg/dl and no self-

3 950 WINGARD AND BARRETT-CONNOR reported personal of diabetes or current use of drugs for hyperglycemia were classified as euglycemic. The present analysis excludes subjects with missing fasting plasma glucose levels, subjects who did not fast for at least 12 hours prior to their visit, pregnant women, nonwhites, and persons under 20 or over 79 years of age at the time of their visit. Male and female subjects aged and those aged years were analyzed separately. Within each of the four age-sex categories, two-tailed statistical tests were done to co"mpare mean levels of age, cholesterol, systolic blood pressure, fasting plasma glucose, and body mass index for subjects with and without a family of diabetes. The skewed triglyceride values were transformed to logarithm (log 10) of measured values for statistical comparisons. In addition, the percentage of subjects with self-reported family of heart attack (at any age) and the percentage of current smokers were compared among those with and without a family of diabetes by chi-square analysis within each of the age-sex categories. The Cox proportional hazards model (13) was used to determine the independent contribution of a family of diabetes to death due to cardiovascular disease after adjusting for differences in age, sex, cholesterol, systolic blood pressure, fasting plasma glucose, body mass index, smoking, and family of heart attack. This was done for diabetic, borderline hyperglycemic, and euglycemic adults separately. Because of the small numbers of deaths among subjects aged 20-49, survival analyses were limited to those aged RESULTS There were 6,110 people who participated in the Rancho Bernardo Heart Disease Risk Factor Survey from Of these, 4,718 subjects were white, nonpregnant, had fasted for at least 12 hours prior to their visit, were years of age, and did not have missing fasting plasma glucose levels. Among these, there were 1,224 euglycemic, 695 borderline hyperglycemic, and 211 diabetic men, and 1,837 euglycemic, 595 borderline hyperglycemic, and 133 diabetic women. The majority of diabetics were not insulin-dependent, based on a subsample of 105 diabetics, only five of whom reported current use of insulin. of diabetes was significantly (p < 0.005) associated with diabetic status, being most common among diabetic men and women (29 per cent each), less common among borderline hyperglycemic men and women and euglycemic women (17 per cent each), and least common among euglycemic men (13 per cent). The association of family of diabetes with diabetic status cannot be entirely based on recall bias. Diabetics identified only by an elevated fasting plasma glucose (i.e., those without a of diabetes or use of drugs for hyperglycemia) still reported more family of diabetes (22 per cent) than euglycemic (15 per cent) or borderline hyperglycemic (17 per cent) adults. The distribution of risk factors by family of diabetes among men and women is presented in tables 1 and 2, respectively, for those years of age, and in tables 3 and 4, respectively, for those aged In younger diabetic men and women, family of diabetes was associated with higher mean age, higher mean total plasma cholesterol, and higher mean plasma triglyceride levels as well as a positive family of heart attack. The differences in mean cholesterol and triglyceride levels in younger diabetic men with a family of diabetes compared with those without persisted after covariance analysis adjusted for age. (Given the small number of diabetic women aged 20-49, the associations were not statistically significant.) Most of these associations were not consistent for older diabetic men or women or for euglycemic or borderline hyperglycemic men or women. These results are comparable to the higher cholesterol and triglyceride values found in small groups of men aged years with a strong family of type II diabetes compared with men of the

4 TABLE 1 Distribution of selectedriskfactors, by family of diabetes, in euglycemic, borderline hyperglycemic, and diabetic^ men aged years, Rancho Bernardo, CA, Risk factor^ Age (years) Cholesterol (mg/100 ml) Triglycerides (mg/ml) Systolic blood pressure (mmhg) Fasting plasma glucose (FPG) (mg/dl) Quetelet index (weight/height 1 X 100) Smoking (% current) of heart attack (%with) (n-345) 36.6 ± ± ± ± ± ± * Euglycemic Borderline hyperglycemic Diabetic (n-36) 37.5 ± ± ± ± ± ± (n - 137) 37.1 ± ± ± ± ± ± (n - 32) 38.6 ± ± ± ± ± ± (n - 16) 40.3" ± * ± * ± ± ± ± ** (n - 10) 46.3 ± ± ± ± ± ± * p < 0.05: no family of diabetes vs. positive family of diabetes. ** p < 0.01: no family of diabetes vs. positive family of diabetes. t Diabetic FPG S: 140 mg/dl and/or personal of diabetes and/or current use of drugs for hyperglycemia; borderline hyperglycemic; 110 s FPG < 140 and no personal of diabetes or current use of drugs for hyperglycemia; euglycemic: FPG < 110 mg/dl and no personal of diabetes or current use of drugs for hyperglycemia. t For continuous variables, mean ± standard deviation. 70.0

5 TABLE 2 Distribution of selected risk factors, by family of diabetes, in euglycemic, borderline hypergtycemic, and diabetic^ women aged years, Rancho Bernardo, CA, * Risk factor}: Age (years) Cholesterol (mg/100 ml) Trigiycerides (mg/ml) Systolic blood pressure (mmhg) Fasting plasma glucose (FPG) (mg/dl) Quetelet index (weight/height* xloo) Smoking (% current) of heart attack (% with) (n - 494) 36.7 ± ± ± ± ± ± Euglycemic Borderline hyperglycemk Diabetic (n - 91) 38.2 ± ± ± ± ± ± (n - 87) 36.8 ± ± ± ± ^2 ± ± (n - 15) 36.8 ± ± ± ± ± ± (n-11) 35.9 ± ± ± ± ± ± (n-6) 40.8 ± ± ± ± ± ± * No statistically significant (p < 0.06) differences were found between those with no family of diabetes and those with a positive family of diabetes. t Diabetic; FPG mg/dl and/or personal of diabetes and/or current use of drugs for hyperglycemia; borderline hyperglycemic: FPG < 140 and no personal of diabetes or current use of drugs for hyperglycemia; euglycemic: FPG < 110 mg/dl and no personal of diabetes or current use of drugs for hyperglycemia. t For continuous variables, mean ± standard deviation to

6 CO 2 TABLE 3 Distribution of selected risk factors, by family of diabetes, in euglycemic, borderline hypergtycemic, and diabetic^ men aged years, Rancho Bernardo, CA, * Risk factor J Age (years) Cholesterol (mg/100 ml) Triglycerides (mg/ml) Systolic blood pressure (mmhg) Fasting plasma glucose (FPG) (mm/dl) Quetelet index (weight/height* X 100) Smoking (% current) of heart attack (%with) (n - 745) 66.9 ± ± ± ± ± ± Euglycemic Borderline hyperglycemic Diabetic (n - 118) 66.4 ± ± ± ± ± ± (n-441) 66.9 ± ± ± ± ± ± (n-85) 65.5 ± ± ± ± * ± ± (n - 134) 66.8 ± ± ± ± ± ± (n-61) 64.4 ± ± ± ± ± ± * No statistically significant (p < 0.06) differences were found between those with no family of diabetes and thoee with a positive family of diabetes. t Diabetic FPG a 140 mg/dl and/or personal of diabetes and/or current use of drugs for hyperglycemia; borderline hyperglycemic: 110 S FPG < 140 and no personal of diabetes or current use of drugs for hyperglycemia; euglycemic: FPG < 110 mg/dl and no personal of diabetes or current use of drugs for hyperglycemia. t For continuous variables, mean ± standard deviation. 51.0

7 s z a 03 % 5 3H o 2; O TABLE 4 Distribution of selected risk factors, by family of diabetes, in euglycemic, borderline hyperglycemic, and diabetic^ women aged years, Rancho Bernardo, CA, Risk factor): Age (years) Cholesterol (mg/100 ml) Tnglycerides (mg/ml) Systolic blood pressure (mmhg) Fasting plasma glucose (FPG) (mg/dl) Quetelet indei (weight/height* x 100) Smoking (% current) of heart attack {% with) (n - 1,036) 64.2 ± ± ± * ± ± ± " Euglycemic (n-217) 63.6 ± ± ± ± ± ± Borderline hyperglycemic (n-409) 64.6 ± ± ± ± ± ± " (n-84) 63.0 ± ± ± ± ± ± (n = 87) 64.7 ± ± ± ± ± ± Diabetic (n = 32) 63.9 ± ± ± ± ± ± p < 0.05: no family of diabetes vs. positive family of diabetes. " p < 0.01: no family of diabetes vs. positive family of diabetes. t Diabetic: FPG S 140 mg/dl and/or personal of diabetes and/or current use of drugs for hyperglycemia; borderline hyperglycemic: 110 S FPG < 140 and no personal of diabetes or current use of drugs for hyperglycemia; euglycemic: FPG < 110 mg/dl and no personal of diabetes or current use of drugs for hyperglycemia. t For continuous variables, mean ± standard deviation. 56.3

8 FAMILY HISTORY OF DIABETES AND CARDIOVASCULAR DISEASE 955 same age without a positive family (14). Only the percentage of subjects with a positive family of heart attack was consistently higher among all subjects with a family of diabetes compared with those without a family of diabetes. The association was significant in five of the 12 age-sex groups: euglycemic men and women aged 20-49, diabetic men aged 20-49, and euglycemic and borderline hyperglycemic women aged The consistency of the trend in three of the four agesex groups suggests that family of heart attack was associated with a positive family of diabetes. It should be noted, however, that none of the comparisons presented in tables 1-4 met the stricter significance criteria for multiple comparison testing suggested by Dixon and Massey (15). A total of 389 subjects aged died of cardiovascular disease: 9.5 per cent of euglycemic, 13.3 per cent of borderline hyperglycemic, and 17.5 per cent of diabetic adults. Table 5 presents sex-specific and age-adjusted risks of cardiovascular disease mortality for these three groups of adults. Note that the risks are elevated for both borderline hyperglycemic and diabetic men and women, although not statistically significant. No difference in risk between men and women is apparent. Table 6 presents the independent contribution of each risk factor to cardiovascular disease mortality for euglycemic, borderline hyperglycemic, and diabetic subjects aged years. Sex was considered as a covariate in the Cox model rather than separating the diabetic status groups into male and female and analyzing smaller numbers of deaths in subgroups, since separate analyses demonstrated no sex-family of diabetes interaction. Only age and blood pressure were independently predictive of cardiovascular disease mortality in all three diabetic status categories, while sex and cholesterol levels were each independently associated with cardiovascular disease mortality in euglycemic and borderline hyper- TABLE5 Sex-specific, age-adjusted cardiovascular disease mortality for family of diabetes among euglycemic, borderline hyperglycemic, and diabetic^ adults aged years, Rancho Bernardo, CA, of diabetes Men Yes No Risk* Women Yea No Risk* Cardiovascular disease mortality:): Euglycemic Borderline hyperglycemic Diabetic * No risk was statistically significant (p 0.05). t Diabetic: fasting plasma glucose (FPG) > 140 mg/dl and/or personal of diabetes and/or current use of drugs for hyperglycemia; borderline hyperglycemic: 110 s FPG < 140 and no personal of diabetes or current use of drugs for hyperglycemia; euglycemic: FPG < 110 mg/dl and no personal of diabetes or current use of drugs for hyperglycemia. J Rates per 100. glycemic subjects. Cigarette smoking was of borderline significance in euglycemic (risk = 1.42, p = 0.08) and diabetic (risk = 2.02, p = 0.06) adults. In this model, with both sexes combined, fasting plasma glucose was not significantly associated with cardiovascular death. Given that prior research (16) suggests that triglyceride is the lipid found in greatest excess in diabetics, the above analysis was repeated with log 10 triglyceride in the model instead of cholesterol. The risk associated with triglyceride was 1.31 (p = 0.001) for borderline hyperglycemic adults but was absent (1.00) for euglycemic and diabetic adults. In this analysis, risks associated with other covariates were virtually unchanged. A positive family of heart attack was independently associated with cardiovascular disease mortality in euglycemic and borderline hyperglycemic subjects, and the relative risk was similar although not statistically significant in diabetics. of diabetes was not significantly associated with mortality risk in any of the

9 956 WINGARD AND BARRETT-CONNOR TABLES Independent relative risk of selected risk factors for cardiovascular disease mortality among euglycemic, borderline hyperglycemic, and diabetic] adults aged years, based on Cox proportional hazards model, Rancho Bernardo, CA, Risk factor? Age (per 5 years) Sex (males vs. females) Cholesterol (per 36 mg/100 ml) Systolic blood pressure (per 25 mmhg) Fasting plasma glucose (FPG) (per 10 mg/dl) Quetelet index (per 0.5 lbs/in' X 100) Smoking (current vs. nonsmoker) of heart attack (yes vs. no) of diabetes (yes vs. no) Cardiovascular disease deaths (n) Euglycemic (n = 2,114) 1.87*** 2.32"* 1.17* 1.39*** * Cardiovascular disease mortality Borderline hyperglycemic (n 1,020) 1.78*** 2.99*** 1.26** 1.34" " Diabetk (n-303) 1.83"* * *p<0.06. "p<0.01. "*p< t Diabetic: FPG > 140 mg/dl and/or personal of diabetes and/or current use of drugs for hypergrycemia; borderline hyperglycemic: 110 < FPG < 140 and no personal of diabetes or current use of drugs for hyperglycemia; euglycemic: FPG < 110 mg/dl and no personal of diabetes or current use of drugs for hyperglycemia. t For continuous variables, apart from age, intervals represent approximately one standard deviation. three groups, but risk followed an increasing trend from euglycemic (0.89) to borderline hyperglycemic (1.17) to diabetic (1.31) adults. Given the correlation between family of diabetes and family of heart attack (which could reflect detection and/or recall bias or a causal pathway), the analysis was repeated without family of heart attack in the model. Again, risks associated with covariates changed only slightly. The risk associated with family of diabetes still followed an increasing trend from euglycemic (0.92) to borderline hyperglycemic (1.25) to diabetic (1.33) adults. DISCUSSION In this population, a family of diabetes was significantly more common among those with diabetes. This association is not entirely a consequence of detection and/or recall bias, since diabetics newly identified by fasting plasma glucose levels also reported more family of diabetes. The most notable association between cardiovascular disease risk factor status and family of diabetes was the higher levels of fasting plasma cholesterol and triglyceride in diabetic men and women years of age with a family of diabetes. This was particularly impressive in men, for whom mean fasting plasma glucose levels did not differ by family ; thus, the differences were not due to a more severe illness or definitive diagnoses among diabetics with a family of diabetes. In addition, in all age-sex-diabetes status groups, those with a family of diabetes had a greater proportion with a family of heart attack. This latter correlation may be real or a consequence of detection bias: Some persons may be more inclined to investigate the of disease in their family and, if one is found, may be more likely to probe for other cases. Alternatively, relatives under medical care for a heart attack may more likely have received a diagnosis of diabetes, and vice versa.

10 FAMILY HISTORY OF DIABETES AND CARDIOVASCULAR DISEASE 957 Although a family of diabetes was associated with the presence of some cardiovascular risk factors and with an (nonsignificant) increase in cardiovascular disease mortality in borderline hyperglycemic and diabetic adults, it was not independently predictive of cardiovascular disease mortality when other risk factors were included in the Cox model. The risks followed an increasing trend from euglycemic (0.89) to borderline hyperglycemic (1.17) to diabetic (1.31) adults, but none were statistically significant. In this study, and in other studies (17-20), age, sex, cholesterol, and systolic blood pressure were significant predictors in euglycemic and borderline hyperglycemic adults. As in other studies (21-23), diabetic women lost their cardiovascular disease mortality advantage when compared with men: The male-to-female cardiovascular disease mortality ratio was 2.3 for euglycemic (p < 0.001) and 3.0 for borderline hyperglycemic (p < 0.001) adults; however, it was only 1.3 for diabetic adults (not significant). This study, as well as other studies (24, 25), found that cigarette smoking was an (borderline) independent predictor in two of the three groups. In a previous analysis of this population (26), an association was demonstrated between fasting plasma glucose and cardiovascular disease mortality in nondiabetic men, but not in women. In the present study, however, no significant association was found, probably because men and women were combined for analysis. As previously reported in this population (27), a family of heart attack was associated with increased cardiovascular disease mortality (significantly so in euglycemic and borderline hyperglycemic adults). That there was a greater proportion with a family of heart attack among those with a family of diabetes might be the pathway whereby family of diabetes had an impact on cardiovascular disease risk. Omitting family of heart attack from the analysis, however, had minimal effect on the risks associated with a family of diabetes. In summary, the present analysis indicates that diabetic men and women years of age with a family of diabetes were more likely to have cardiovascular disease risk factors (higher levels of fasting plasma cholesterol and triglyceride and a family of heart attack) than those without a family of diabetes. In nondiabetic and in older (50-79 years) diabetic adults, however, risk factors other than family of heart attack did not vary by family of diabetes. In addition, family of diabetes was not significantly associated with cardiovascular disease mortality for euglycemic, borderline hyperglycemic, or diabetic adults, either before or after adjusting for differences in cardiovascular disease risk factors. This analysis, in combination with the little indirect evidence available (8-11), suggests that cardiovascular disease risk is not inherited with diabetes risk but is a complication of the diabetic disease state. REFERENCES 1. Simpson NE. Diabetes in families of diabetics. Can Med Assoc J 1968^8: Keen H, Jarrett RJ. Environmental factors and genetic interactions. In: Creutzfeldt W, Kobbering J, Neel JV, eds. The genetics of diabetes mellitus. Berlin: Springer-Verlag, 1976: Mann JI, Houston AC. The aetiology of noninsulin dependent diabetes mellitus. In: Mann JI, Pyorala K, Teuscher A, eds. Diabetes in epidemiologic perspective. New York: Churchill Livingstone Inc, 1983: Everhardt J, Knowler WC, Bennett PH. Incidence and risk factors for non-insulin-dependent diabetes. In: National Diabetes Data Group. Diabetes in America: diabetes data compiled :(4)l-35 (DHHS NIH publication no. (PHS) ). 5. Leslie RDG, Pyke DA. Genetics of diabetes. In: Alberti KGMM, Krall LP, eds. The diabetes annual. Vol 1. New York: Elsevier Science Publishers BV, 1985: Barrett-Connor E, Orchard T. Diabetes and heart disease. In: National Diabetes Data Group. Diabetes in America: diabetes data compiled :(16)1-41 (DHHS NIH publication no. (PHS) ). 7. Pyorala K, Laakso M. Macrovascular disease in diabetes mellitus. In: Mann JL Pyorala K, Teuscher A, eds. Diabetes in epidemiologic perspective. New York: Churchill Livingstone Inc, 1983:

11 958 WINGARD AND BARRETT-CONNOR 8. Wykeham Balme H, Cole L. The heredity of hypertension in diabetes mellitus. Q J Med 1951; 20(80): BrunzeU JD, Hazzard WR, Motulsky AG, et al. Evidence for diabetes mellitus and genetic forms of hypertriglyceridemia as independent entities. Metabolism 1975;24(10): Schneider VH, Pastow A. Das kardiovaakulare risikoprofil des langzeitdiabetikers und seine beziehungen zur neirenfunktion. Z Gesamte Inn Med 1982;10: Ganda OP, Soeldner JS, Gleason RE. Alterations in plasma lipids in the presence of mild glucose intolerance in the offspring of two type II diabetic parents. Diabetes Care 1985;8(3): Austin MA, Berreyesa S, Elliott JL, et al. Methods for determining long-term survival in a population based study. Am J Epidemiol 1979;110: Lee ET. Statistical methods for survival data analysis. Belmont, CA: Lifetime Learning Publications, 1980: Bemtorp K, Lindgarde F. Impaired physical fitness and insulin secretion in normoglycaemic subjects with familial aggregation of type 2 diabetes mellitus. Diabetes Res 1985;2: Dixon WJ, Massey FJ. Introduction to statistical analysis. 4th ed. New York McGraw-Hill Co, 1983: Barrett-Connor E, Grundy SM, Holdbrook MJ. Plasma lipids and diabetes mellitus in an adult community. Am J Epidemiol 1982;115: Kannel WB, Cagtelli WP, Gordon T. Serum cholesterol, lipoproteins and the risk of coronary heart disease. Ann Intern Med 1971;74: Kannel WB. Role of blood pressure in cardiovascular mortality. Prog Cardiovasc Dis 1974;17: Kannel WB. Some lessons in cardiovascular epidemiology from Framingham. Am J Cardiol 1976;37: Tyroler HA, Heyden S, Bartel A, et al. Blood pressure and cholesterol as coronary heart disease risk factors. Arch Intern Med 1971;128: Barrett-Connor E, Wingard DL. Sex differential in ischemic heart disease mortality in diabetics: a prospective population-based study. Am J Epidemiol 1983; 118: Kannel WB, McGee DL. Diabetes and glucose tolerance as risk factors for cardiovascular disease: The Framingham Study. Diabetes Care 1979;2: Heyden S, Heiss G, Bartel AG, et al. Sex differences in coronary mortality among diabetics in Evans County, Georgia. J Chronic Dis 1980; 33: Hammond EC. Smoking in relation to death rates of one million men and women. Natl Cancer Inst Monogr 1966:19: Hammond EC, Garfinkel L. Coronary heart disease, stroke and aortic aneurysm: factors in the etiology. Arch Environ Health 1969;19: Barrett-Connor E, Wingard DL, Criqui MH, et al. Is borderline fasting hyperglycemia a risk factor for cardiovascular death? J Chronic Dis 1984; 37(9/10): Barrett-Connor E, Khaw KT. of heart attack as an independent predictor of death due to cardiovascular disease. Circulation 1984; 69(6):