The association between metabolic control and prevalent macrovascular disease in Type 2 diabetes: The VA Cooperative Study in diabetes

Article Outline

• Abstract
• 1. Introduction
• 2. Methods
  • 2.1. The VADT
  • 2.2. Assessment of glycemic control and lipids
  • 2.3. Definition of prevalent MVD
  • 2.4. Statistical analysis
• 3. Results
  • 3.1. Study population
  • 3.2. Relationship between HbA1c and MVD
  • 3.3. Assessment of the association between HbA1c and other macrovascular risk factors
  • 3.4. Analysis of the independent association of HbA1C with prevalent MVD, adjusting for potentially confounding variables
• 4. Discussion
• Acknowledgments
• References
• Copyright

Abstract 

The problem

Macrovascular disease (MVD), especially coronary heart disease, is the most common cause of mortality in Type 2 diabetes. We assessed the association between demographic and clinical variables (particularly HbA1c) and prevalent MVD at time of enrollment into the VA Diabetes Trial (VADT), a 7-year randomized trial to determine whether intensive glycemic control will reduce risk of MVD events in older participants with established Type 2 diabetes.

Research design and methods

We compared the demographic, treatment, and clinical characteristics of participants with and without known MVD, then assessed the interaction of multiple variables with HbA1c. Logistic regression models evaluated the association between HbA1c quartiles and prevalence of MVD, adjusting for potentially confounding variables.

Results

Several variables were associated with prevalent MVD (age, duration of diabetes, insulin use but not daily dosage, smoking history, hypertension, BMI, Caucasian race, non-Hispanic ethnicity, lower HDL cholesterol, higher triglycerides, lower LDL cholesterol, and statin use). In univariate analysis, there was no association of HbA1c with MVD (mean: 9.4±1.46% in those with MVD, 9.5±1.58% in those without). Multivariate analyses found little confounding of the lack of association of HbA1c with MVD. Only adjustment for age produced a slight increase in the odds ratio, but only for the highest quartile of HbA1c.

Conclusions

In this cross-sectional analysis, MVD was associated with a number of clinical and demographic variables but not with HbA1c. Determining whether intensive lowering of HbA1c will reduce the prospective rate of MVD events in this population of older participants with established Type 2 diabetes is the primary objective of our trial.

Keywords: Macrovascular disease, Glycemic control

Back to Article Outline

1. Introduction 

Persons with Type 2 diabetes have a markedly increased risk of macrovascular disease (MVD) compared with age- and gender-matched persons without the disease (Geiss et al., 1995, Haffner et al., 1998, Kannel & McGee, 1979, Wingard & Barrett-Connor, 1995). Diabetes is associated with a number of macrovascular risk factors, including obesity, hypertension, and dyslipidemia. However, the risk for MVD remains independently associated with diabetes, even when controlling for other known risk factors (Haffner et al., 1998, Stamler et al., 1993), suggesting that hyperglycemia itself may be a macrovascular risk factor. Metabolic control of diabetes is strongly associated with prevalent (Knuiman et al., 1986, Krolewski et al., 1995) and incident (Chase et al., 1989, Klein et al., 1994, Stratton et al., 2000, The Diabetes Control and Complications Research Group, 1993) microvascular complications such as retinopathy, nephropathy, and neuropathy. However, studies investigating the association of metabolic control with prevalent or incident MVD have shown no association (Meigs et al., 1997, Smith et al., 2004), relatively weak associations compared with microvascular outcomes (Klein, 1995, Stratton et al., 2000), or a possible paradoxical association of incident MVD with improved metabolic control (Abraira et al., 1997, Levin et al., 2000) when high doses of insulin were the primary mode of therapy.

The VA Diabetes Trial (VADT) is a 7-year trial to investigate whether intensive glycemic control affects primary or secondary risk of MVD in persons with established Type 2 diabetes. Participants with and without known MVD were enrolled in the trial, allowing the opportunity to investigate the association of metabolic control (baseline HbA1c) with prevalent MVD in this large population.

Back to Article Outline

2. Methods 

2.1. The VADT 

The trial began in December 2000, enrolling 1792 men and women with established Type 2 DM. Participants were over age 40 and had inadequate glycemic control, defined as a baseline HbA1c of ≥7.5% on maximal dose of at least one oral anti-hyperglycemic agent or on insulin. Participants with a history of MVD were enrolled in the trial, as long as they did not meet the exclusion criteria related to MVD (current Canadian Heart Association Class III or IV angina; New York Heart Association Class III or IV congestive heart failure; ongoing lower extremity gangrene) and had not had stroke, myocardial infarction, or revascularization procedure within 6 months prior to enrollment. Study participants were randomized to intensive (goal HbA1c <6.0%) or standard glycemic control (mean HbA1c 1.5% higher than the intensive group mean), as previously described (Abraira et al., 2003), with intensive management of other macrovascular risk factors in both arms of the study. Randomization was stratified by the presence or absence of MVD and by use of insulin at baseline. The current analysis is based on the baseline cohort for the trial. More than 96% of values for each variable, other than calculated LDL cholesterol, were available for these analyses.

2.2. Assessment of glycemic control and lipids 

Hemoglobin A1c and lipids were measured during the screening period in all participants. Blood samples were obtained locally and sent to Tufts University, Boston, MA, for central assessment. HbA1c was measured by turbidometric immunoinhibition assay (Roche Diagnostics, Indianapolis, IN). Reference range was 4.8–6.0%. Total cholesterol, HDL cholesterol, and triglycerides were measured by enzymatic end-point assays (Roche Diagnostics). LDL cholesterol was calculated by the Friedewald calculation (Friedewald, Levy, & Fredrickson, 1972) if the value for triglycerides was <400 mg/dl.

2.3. Definition of prevalent MVD 

Prevalent MVD was defined as patient or chart (computerized medical record) history of myocardial infarction, coronary revascularization, CHA Class I or II angina, NYHA Class I or II CHF, claudication, critical limb ischemia, extremity revascularization, amputation for ischemia, stroke, TIA, or carotid revascularization.

2.4. Statistical analysis 

Differences between means were assessed with t tests, and those between proportions with chi-squared tests. Associations among continuous variables were assessed by Pearson correlation coefficient, while those among categorical variables were assessed by the Mantel-Haenszel chi-squared test for trend.

Logistic regression models were used to evaluate the association between HbA1c quartiles and prevalence of MVD, adjusting for potentially confounding variables in a multivariate analysis. The lowest quartile of HbA1c (7.5–8.2%) was used as a reference. The measures considered as covariates included the variables in Table 2.

Table 2. Correlation of demographic and clinical variables with quartiles of HbA1c

Back to Article Outline

3. Results 

3.1. Study population 

Participants had a mean age of 60.0 years (S.D.=8.7 years). The majority (97%) were male. Eighteen percent of participants self-identified their ethnicity as Hispanic or Latino. Eighty percent of participants identified their race as Caucasian and 18% as African American, with few participants identifying themselves as Asian, Native American, or Native Hawaiian/Pacific Islander. Most participants (89.1%) had at least a high school education, with 35% having some college and 12% being college graduates. One third of the participants were employed full time, 9% part time, and 19% were unemployed or on disability. Thirty-nine percent of the participants described themselves as retired.

Forty-six percent of the participants had a history of MVD, with most (39% of the total cohort) having known coronary artery disease. Mean duration of diagnosed diabetes was 11.5 years, with a standard deviation of 7.5 years. Mean HbA1c was 9.44±1.53%. Fifty-two percent of the participants were taking insulin, either alone or in combination with oral agents, and the remainder were on oral agents alone. On average, participants were obese, with a mean BMI of 31.3±4.5 kg/m². Hypertension was a known diagnosis in 72% of participants, with a mean blood pressure (BP) overall of 131/77 (±17/10) mm Hg. Seventeen percent of the participants were current smokers, 56% were former smokers, and 27% had never smoked. Fasting lipid profile results (median, interquartile range) were the following: total cholesterol, 177 (155–203); HDL-cholesterol, 34 mg/dl (29–41); LDL-cholesterol, 104 mg/dl (86–126); and triglycerides, 162 mg/dl (112–238).

As shown in Table 1, participants with MVD were older, more likely to be current or former smokers, more likely to have hypertension, had a longer duration of diabetes, and a higher prevalence of insulin use. Total daily insulin dosage did not differ between groups, nor did the use of metformin. Fewer participants with MVD used sulfonylurea drugs at baseline than those without. Race and ethnicity were significantly related to MVD, with fewer MVD participants being Hispanic or Latino and more being Caucasian.

Table 1. Characteristics of participants with and without known MVD

Participants with MVD had lower HDL cholesterol levels, higher levels of triglycerides, and lower LDL cholesterol levels than did those without. Participants with MVD were significantly more likely to be on statins than those without MVD, which partially explains the paradoxical LDL association.

3.2. Relationship between HbA1c and MVD 

As shown in Table 1, there was no significant difference in mean HbA1c between the group with known MVD and that without (9.4±1.46% vs. 9.5±1.58%, P=.221); the median values were identical (9.0%). The prevalence of MVD was also equivalent across quartiles of HbA1c, with the odds ratio for the highest quartile of HbA1c identical to that of the lowest quartile (1.0; 95% confidence interval 0.75–1.31). There was also no association between quartiles of fasting plasma glucose and MVD (data not shown).

3.3. Assessment of the association between HbA1c and other macrovascular risk factors 

The relationship of metabolic control to MVD is potentially confounded by an association between HbA1c and other variables associated with MVD risk. As shown in Table 2, total cholesterol, LDL cholesterol, and prevalence of insulin use and mean daily insulin dosage all increased across increasing quartiles of HbA1c, while mean age decreased. The percentage of participants who were Hispanic or African American increased with increasing quartiles of HbA1c. The use of metformin and sulfonylurea drugs did not significantly differ across the quartiles of HbA1c.

3.4. Analysis of the independent association of HbA1C with prevalent MVD, adjusting for potentially confounding variables 

Correlations between HbA1c, known MVD risk factors, duration of diabetes, and diabetes therapy could potentially mask an association between metabolic control and prevalent MVD in unadjusted analyses. While many of the variables in Table 1, Table 2 contributed significantly to the prediction of prevalent MVD, little or no confounding of the association between HbA1c and MVD was seen. In multivariate analyses adjusting for the variables shown in Table 2, only adjustment for age produced a statistically significant increase in the odds ratio of MVD with higher HbA1c, and only for the highest quartile (>10.2%) of HbA1c [OR 1.3 (1.05–1.77)] compared with the lowest quartile of <8.3%.

Back to Article Outline

4. Discussion 

Glycemic control is strongly linked to the presence (Knuiman et al., 1986, Krolewski et al., 1995) and risk of development (Chase et al., 1989, Klein et al., 1994, Stratton et al., 2000, The Diabetes Control and Complications Research Group, 1993) of microvascular and neuropathic complications of diabetes. Interventions to improve glycemic control greatly reduce the risk of development and progression of these complications (Ohkubo et al., 1995, The Diabetes Control and Complications Research Group, 1993, UK Prospective Diabetes Study (UKPDS) Group, 1998). The microvascular complications retinopathy and nephropathy are highly specific to diabetes mellitus, thus, a strong link to chronic hyperglycemia is not surprising.

The main cause of morbidity and mortality in Type 2 diabetes is macrovascular or cardiovascular disease. Coronary artery, cerebrovascular, and peripheral vascular diseases are not specific to diabetes, and these complications have known associations with nonglycemic variables both in diabetic and nondiabetic individuals. Some analyses have suggested an association between diabetic glycemic control and the presence (Singer, Nathan, Anderson, Wilson, & Evans, 1992) or risk of the development of MVD (Khaw et al., 2004, Klein, 1995, Kuusisto et al., 1994, Laakso, 1996, Selvin et al., 2004, Stratton et al., 2000). However, others have found no association of HbA1c and prevalent MVD in Type 2 diabetes (Meigs et al., 1997, Smith et al., 2004), similar to our findings.

In addition, in contrast to the striking reduction in risk of development or progression of microvascular complications found in several landmark trials of intensive versus standard glycemic control, reductions in MVD risk with intensive glycemic control have been difficult to prove. Neither the DCCT nor the Kumamoto study showed a significant effect on macrovascular outcome, although this conclusion is limited by the low rates of these outcomes in those studies. Intensive glycemic control in the UKPDS had no significant effect on total MVD or MVD mortality, but did show a reduction in nonfatal MI that approached statistical significance (P=.052; UK Prospective Diabetes Study, 1998). In contrast, the early pilot trial for our VA Cooperative Study in insulin-treated Type 2 patients found a tendency toward worsening of MV outcomes with intensive glycemic control achieved by using high doses of insulin (Abraira et al., 1997, Levin et al., 2000).

In a population of veterans with suboptimal diabetes control, we found no association of prevalent MVD and glycemic control in unadjusted analyses and only a weak association with very poor glycemic control (HbA1c >10.2%) when results were adjusted for age and ethnicity. There are several potential limitations of our analyses. The population was almost entirely male, which may have masked an association that is much stronger in women. However, no other studies of either microvascular or macrovascular outcomes in diabetes have shown large differences between the genders. Second, we excluded participants with relatively good glycemic control (HbA1c <7.5%) from our study. We had, however, a large range of HbA1c values in our population, and other studies have suggested a curvilinear or linear relationship of HbA1c to complications, with risk increasing sharply at higher ranges of HbA1c (Stratton et al., 2000, The Diabetes Control and Complications Research Group, 1993). Although some experts cite a “flattening” of risk for myocardial infarction with higher levels of HbA1c in the UKPDS post hoc analysis of glycemia and MVD, this observation is limited to a single figure that compares incidence rates for MI with the updated mean HbA1c. Similar findings were not seen when the results were expressed as hazard ratios, nor for any other macrovascular complication, and the authors themselves do not claim a “flattening” (Stratton et al., 2000).

Our results are also consistent with those of Meigs et al. (1997) and Smith et al. (2004), both of whom included the whole range of glycemic control in their diabetic populations. There could be “survivor bias” in our study, as those with MVD had survived their event for at least 6 months prior to enrollment. We think it unlikely, however, that survival bias would completely negate a large effect. Inasmuch as many diabetic participants with diabetes have asymptomatic MVD, and inasmuch as our participants have multiple MVD risk factors, some could argue that defining the presence of MVD by known clinical events is a limitation. However, all major epidemiologic analyses and clinical trials assess MVD events, not asymptomatic disease, as the appropriate clinically relevant outcomes of interest. Finally, it should be noted that glycemic control and the presence of known MVD were assessed at a single point in time at least 6 months after the MVD event and, in many cases, significantly later. It is possible that participants with known MVD had significantly worse glycemic control in the years prior to their MVD events, and that the diagnosis of MVD led to greater efforts by providers to improve glycemic control in these participants. The fact that participants with a history of MVD had lower LDL cholesterol levels and were more likely to be on statins could be seen as an example of this phenomenon. However, the mean HbA1c in the population was quite high, in contrast to the mean lipid levels and BP, which would suggest that glycemic therapy had not been intensified significantly.

Our participants were recruited to a study meant to address the effect of intensive glycemic control on primary or secondary prevention of macrovascular outcomes. The VADT has 86% power to detect a 21% relative reduction of macrovascular events. Even with the finding of no association of glycemic control and prevalent MVD, it remains important to definitively address the question of whether intensive glycemic control reduces the risk of incident macrovascular outcomes. There is more clear evidence that HbA1c predicts future macrovascular events (Klein, 1995, Kuusisto et al., 1994, Laakso, 1996) than that is associated with prevalence of MVD, as found in the Diabetes PORT (Meigs et al., 1997), the analysis by Smith et al. (2004) in VA patients, and our analysis. In previous studies of intensive glycemic control, MVD was often a secondary outcome that could not be adequately studied due to power issues. In addition, there was rarely intensive management of other macrovascular risk factors, such as lipids or BP, or broad use of preventive drugs such as aspirin in these trials.

Intensive glycemic control is not entirely without risk, being known to increase weight gain and the risk of hypoglycemia. In the pilot study for the current trial, there was a suggestion that intensive glycemic control might even increase the risk of MVD events (Abraira et al., 1997, Levin et al., 2000). A study to address whether the risks of intensive glycemic control are outweighed by clear benefit in terms of the most common and clinically significant complications of Type 2 diabetes, macrovascular complications, is clearly warranted. Aggressive management of lipids and BP and the universal use of aspirin in both glycemic arms of the study should allow us to isolate the effect of glycemic control on MVD outcomes and, in addition, should attenuate or abolish any increase in risk of clinically significant microvascular outcomes in the standard glycemic control arm (UK Prospective Diabetes Study, 1998; Vijan, Hofer, & Hayward, 1997). Given that these nonglycemic interventions have been shown more definitively to date to reduce macrovascular risk in diabetes than has intensive glycemic control (Huang, Meigs, & Singer, 2001), and given our finding of no association of glycemic control with prevalent MVD in our baseline cohort whose BP and lipids were fairly well controlled, it may be that our trial will not demonstrate an advantage of intensive glycemic control in lessening macrovascular outcomes. Because intensive glycemic control has risks and costs, this would still be an important result, as it would suggest that resources be focused on nonglycemic risk factors in older established diabetic patients whose lifetime risk of clinically significant microvascular complications is low (Vijan et al., 1997) but whose risk of macrovascular morbidity and mortality is high.

Back to Article Outline

Acknowledgments 

This study is supported by the Cooperative Studies Program of the Department of Veterans Affairs Office of Research and Development, with additional support from the American Diabetes Association, National Eye Institute, Glaxo SmithKline Pharmaceuticals, Novo Nordisk Pharmaceuticals, Aventis Pharmaceuticals, and Kos Pharmaceuticals. Members of the VADT Study Group are listed in the appendix of Abraira et al. (2003). Preliminary analyses of this work were presented in abstract form at the 64th Annual Scientific Sessions of the American Diabetes Association in June 2004 and at the 12th International Congress of Endocrinology in September 2004.

Back to Article Outline

References 

1. Abraira C, Colwell J, Nuttall F, Sawin C, Henderson W, Comstock J, et al. Veterans Affairs Cooperative Study on Glycemic Control and Complications in Type II DM (VACSDM) Group: Cardiovascular events and correlates in the Veterans Affairs Diabetes Feasibility Trial. Archives of Internal Medicine. 1997;157:181–188
   • View In Article
   • MEDLINE
2. Abraira C, Duckworth W, McCaren M, Emmanuele N, Arca D, Reda D, et al. Design of the cooperative study on glycemic control and complications in diabetes mellitus type 2: Veterans Affairs Diabetes Trial. Journal of Diabetes and its Complications. 2003;17:314–322
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (175 KB)
   • CrossRef
3. Chase HP, Jackson WE, Hoops SL, Cockerham RS, Archer PG, O'Brien D. Glucose control and the renal and retinal complications of insulin-dependent diabetes mellitus. Journal of the American Medical Association. 1989;261:1155–1160
   • View In Article
4. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clinical Chemistry. 1972;18:499–502
   • View In Article
   • MEDLINE
5. Geiss S, Herman WH, Smith PJ. Mortality in non-insulin-dependent diabetes. In:  Harris M editors. Diabetes in America/National Diabetes Data Group. 2nd ed.. Bethesda, MD: National Institutes of Health; 1995;p. 133–155
   • View In Article
6. Haffner SM, Lehto S, Ronnemaa T, Pyorala K, Laakso M. Mortality from coronary heart disease in subjects with type 2 diabetes and in nondiabetic subjects with and without prior myocardial infarction. New England Journal of Medicine. 1998;339:229–234
   • View In Article
7. Huang ES, Meigs JB, Singer DE. The effect of interventions to prevent cardiovascular disease in patients with type 2 diabetes mellitus. American Journal of Medicine. 2001;111:633–642
   • View In Article
8. Kannel WB, McGee DL. Diabetes and cardiovascular disease: The Framingham Study. Journal of the American Medical Association. 1979;241:2035–2038
   • View In Article
9. Khaw K-T, Wareham N, Bingham S, Luben R, Welch A, Day N. Association of hemoglobin A1c with cardiovascular disease and mortality in adults: The European Prospective Investigation into Cancer in Norfolk. Annals of Internal Medicine. 2004;141:413–420
   • View In Article
10. Klein R. Hyperglycemia and microvascular and macrovascular disease in diabetes. Diabetes Care. 1995;18:258–268
    • View In Article
    • MEDLINE
11. Klein R, Klein BE, Moss SE, Cruikshanks KJ. Relationship of hyperglycemia to the long-term incidence and progression of diabetic retinopathy. Archives of Internal Medicine. 1994;154:2169–2178
    • View In Article
    • MEDLINE
12. Knuiman MW, Welborn TA, McCann VJ, Stanton KG, Constable IJ. Prevalence of diabetic complications in relation to risk factors. Diabetes. 1986;35:1332–1339
    • View In Article
    • MEDLINE
13. Krolewski AS, Laffel LMB, Krolewski M, Quinn M, Warram JH. Glycosylated hemoglobin and the risk of microalbuminuria in patients with insulin-dependent diabetes mellitus. New England Journal of Medicine. 1995;332:1251–1255
    • View In Article
14. Kuusisto J, Mykkanen L, Pyorala K, Laakso M. NIDDM and its metabolic control predict coronary heart disease in elderly subjects. Diabetes. 1994;43:960–967
    • View In Article
    • MEDLINE
15. Laakso M. Glycemic control and the risk of coronary heart disease in patients with non-insulin-dependent diabetes mellitus: The Finnish studies. Annals of Internal Medicine. 1996;124(1S-II)(Suppl.):127–130
    • View In Article
16. Levin SR, Coburn JW, Abraira C, Henderson WG, Colwell JA, Emanuele NV, et al. Effect of intensive glycemic control on microalbuminuria in type 2 diabetes. Diabetes Care. 2000;23:1478–1485
    • View In Article
    • MEDLINE
    • CrossRef
17. Meigs JB, Singer DE, Sullivan LM, Dukes KA, D'Agostino RB, Nathan DM, et al. Metabolic control and prevalent cardiovascular disease in non-insulin-dependent diabetes mellitus (NIDDM): The NIDDM Patient Outcomes Research Team. American Journal of Medicine. 1997;102:38–47
    • View In Article
18. Ohkubo Y, Kishikawa H, Araki E, Miyata T, Isami S, Motoyoshi S, et al. Intensive insulin therapy prevents the progression of diabetic microvascular complications in Japanese patients with non-insulin-dependent diabetes mellitus: A randomized prospective 6-year study. Diabetes Research and Clinical Practice. 1995;28:103–117
    • View In Article
    • Abstract
    • Full-Text PDF (1204 KB)
    • CrossRef
19. Selvin E, Marinopolos S, Berkenblit G, Rami T, Brancati F, Powe N, et al. Meta-analysis: Glycosylated hemoglobin and cardiovascular disease in diabetes mellitus. Annals of Internal Medicine. 2004;141:421–431
    • View In Article
20. Singer DE, Nathan DM, Anderson KM, Wilson PWF, Evans JC. Association of HbA1c with prevalent cardiovascular disease in the original cohort of the Framingham Heart Study. Diabetes. 1992;41:202–208
    • View In Article
    • MEDLINE
21. Smith NL, Chen L, Au DH, McDonnell M, Fihn SD. Cardiovascular risk factor control among veterans with diabetes. Diabetes Care. 2004;27(Suppl. 2):B33–B38
    • View In Article
    • CrossRef
22. Stamler J, Vaccaro O, Neaton JD, Wentworth D. Diabetes, other risk factors, and 12-year cardiovascular mortality for men screened in the Multiple Risk Factor Intervention Trial. Diabetes Care. 1993;2:434–444
    • View In Article
23. Stratton IM, Adler AI, Neil AW, Matthews DR, Manley SE, Cull CA, et al. Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35). British Medical Journal. 2000;321:405–412
    • View In Article
24. The Diabetes Control and Complications Research Group. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. New England Journal of Medicine. 1993;329:977–986
    • View In Article
25. UK Prospective Diabetes Study (UKPDS) Group. Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). The Lancet. 1998;352:837–853
    • View In Article
26. UK Prospective Diabetes Study (UKPDS) Group. Tight blood pressure control and risk of macrovascular and microvascular complications. British Medical Journal. 1998;317:703–713
    • View In Article
27. Vijan S, Hofer T, Hayward R. Estimated benefit from glycemic control in microvascular complications in type 2 diabetic patients. Annals of Internal Medicine. 1997;127:788–795
    • View In Article
    • MEDLINE
28. Wingard DL, Barrett-Connor E. Heart disease and diabetes. In:  Harris M editors. Diabetes in America/National Diabetes Data Group. 2nd ed.. Bethesda, MD: National Institutes of Health; 1995;p. 429–448
    • View In Article