Age-adjusted cumulative probability of death due to coronary heart disease (CHD), stratified by myocardial infarction (MI) and duration of diabetes at baseline in men (A) and women (B). LDM indicates long-standing diabetes mellitus; neither, no diabetes or MI; and RDM, recent diabetes mellitus.
Natarajan S, Liao Y, Sinha D, Cao G, McGee DL, Lipsitz SR. Sex Differences in the Effect of Diabetes Duration on Coronary Heart Disease Mortality. Arch Intern Med. 2005;165(4):430-435. doi:10.1001/archinte.165.4.430
Copyright 2005 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2005
It is not known whether the coronary heart disease (CHD) mortality risk associated with recent (RDM; <10 years) or long-standing diabetes mellitus (LDM; ≥10 years) varies by sex.
The relationship between diabetes duration and CHD mortality was evaluated among 10 871 adults (aged 35-74 years at baseline) using the 1971-1992 National Health and Nutrition Examination Survey Epidemiologic Follow-up Study.
The CHD mortality rates per 1000 person-years in men with no myocardial infarction (MI) or diabetes, MI only, RDM only, LDM only, MI and RDM, and MI and LDM were 5.5 (95% confidence interval, 4.8-6.2), 15.2 (11.6-20.0), 13.2 (7.9-22.1), 11.4 (6.4-20.3), 36.0 (16.7-77.7), and 35.4 (14.0-89.7), respectively. The corresponding rates in women were 2.9 (2.5-3.3), 7.3 (5.0-10.8), 5.2 (3.5-7.7), 10.7 (7.5-15.5), 9.3 (4.3-19.9), and 21.6 (6.1-76.0), respectively. Compared with MI, the multivariate hazard ratios and their 95% confidence intervals (adjusted for age, race, smoking, hypertension, total cholesterol level, and body mass index) for fatal CHD in men with RDM, LDM, MI and RDM, and MI and LDM were 0.7 (0.3-1.3), 0.8 (0.4-1.4), 3.2 (1.4-7.4), and 2.4 (0.8-6.7), respectively. The corresponding ratios in women were 0.9 (0.6-1.3), 1.8 (1.1-3.2), 1.3 (0.5-3.5), and 1.6 (0.2-10.9), respectively.
In men, RDM and LDM were associated with as high a risk for CHD death as MI. In women, although RDM had a CHD mortality risk similar to MI, LDM had an even greater risk. Because women with LDM are at very high risk for CHD mortality, current guidelines may need to be further refined to match intensity of treatment to risk in these women.
Coronary heart disease (CHD) is the leading cause of mortality and morbidity in developed countries, with a high case-fatality rate, emphasizing the need for aggressive preventive strategies. Recent data indicate that individuals with diabetes mellitus have as high a risk for fatal CHD as do individuals with established CHD.1 Previous studies have shown that diabetes has a greater impact on the risk for CHD in women than in men. In contrast to the substantial decrease in CHD mortality in individuals without diabetes, national trends demonstrate a modest decline in heart disease mortality in men with diabetes and an increase in heart disease mortality in women with diabetes.2 This emphasizes the need to further understand sex differences in the risk associated with diabetes.
Recent guidelines recommend aggressive management of other CHD risk factors in individuals with diabetes. Although diabetes has a greater effect on CHD mortality in women compared with men,3,4 it is not clear whether the risk associated with diabetes is constant or varies over time, and it is not known whether there is a differential risk over time by sex. To determine this for the US population as a whole, evaluation of a national population sample that includes men and women with diabetes is necessary. The specific aims of this investigation were: (1) to evaluate the independent effect of recent (RDM) and long-standing diabetes (LDM) on subsequent CHD mortality in men and women; (2) to quantify the sex-specific effects of diabetes duration on CHD mortality compared with prevalent myocardial infarction (MI), an accepted marker of increased CHD mortality risk; and (3) to determine the effect of each additional year of diabetes on CHD mortality.
This analysis used cohort data from the First National Health and Nutrition ExaminationSurvey (NHANES I) Epidemiologic Follow-up Study. The baseline evaluation was conducted from 1971 to 1975 in the NHANES I, a multistage, stratified, probability sample of the US civilian noninstitutionalized population. The NHANES I oversampled the poor, women of childbearing age (25-44 years), and those 65 years and older. The NHANES I Epidemiologic Follow-up Study is the prospective follow-up study of NHANES I participants aged 25 to 74 years in 1971 to 1975 with follow-up surveys in 1982 to 1984, 1986, 1987, and 1992. This analysis was restricted to the 10 871 individuals aged 35 to 74 years, representing 77 592 596 US adults, during the baseline evaluation.
During the baseline examination, sociodemographic information, a detailed medical history, results of a standardized physical examination, and laboratory data were collected.5 Myocardial infarction at baseline was defined on the basis of participant self-report of physician-diagnosed heart attack. Diabetes was defined on the basis of participant report of physician-diagnosed diabetes. Individuals with diabetes were also asked when it was first diagnosed. Based on whether diabetes duration at baseline was less than 10 years or 10 years or greater, it was categorized as RDM or LDM.
Participants or their proxies were contacted in 1982 to 1984, 1986, 1987, and 1992.6- 9 For individuals who had died, the cause of death was ascertained from death certificates. Persons whose cause of death was coded as 410 through 414 in the International Classification of Diseases, Ninth Revision, were considered to have had fatal CHD.10 Among participants who had died, death certificates were obtained for 97.3% of individuals with diabetes and 96.5% without diabetes at baseline.
Smoking status was classified as current (regular smoking in the year before the visit) and nonsmoking on the basis of responses to the initial baseline questionnaire or responses to lifetime smoking questions in the follow-up questionnaires.11 Hypertension was defined as systolic blood pressure of 140 mm Hg or greater, diastolic blood pressure of 90 mm Hg or greater, or use of antihypertensive medications. Total cholesterol level was measured from frozen serum samples at the Centers for Disease Control and Prevention, Atlanta, Ga. Body mass index (BMI) was calculated as weight in kilograms divided by the square of height in meters. Ethnicity was categorized as white or nonwhite.
We initially performed analyses separately by sex. We compared baseline characteristics for a 6-level composite diabetes duration/MI variable that included no MI or diabetes, MI only, RDM only, LDM only, MI and RDM, and MI and LDM. Age-adjusted CHD mortality rates were obtained by a life-table product-limit method. Age-adjusted CHD mortality curves were determined for the 6-level diabetes duration–MI variable, and we tested the equality of curves by means of a Wald statistic.
We determined the independent effect of RDM or LDM on CHD mortality using Cox proportional hazards models.12 The proportionality assumption was tested and met. All multivariate analyses adjusted for baseline age, race, hypertension, smoking, serum total cholesterol level, and BMI, and the relative risks were reported as hazard ratios (HR). Because results of adjustment for menopause were not significant, adjustment for menopause was not included in the final models. Persons without diabetes or MI were initially used as the reference group. Then, to quantify the effect of RDM or LDM on CHD mortality, persons with prevalent MI were used as the comparison group. To formally test for sex differences in the effect of diabetes duration and MI on CHD mortality, interactions between sex and the 6-level composite diabetes duration–MI variable were tested in a hierarchical Cox model combining men and women. The effect of each additional year of diabetes was determined using diabetes duration as a linear variable in a Cox model that included baseline age, race, MI, hypertension, smoking, serum total cholesterol level, and BMI. All analyses were performed using SAS13 and SUDAAN14 and used the appropriate weighting and clustering variables to obtain population estimates.
Table 1 presents the baseline characteristics. In both men and women, there were fewer current smokers and more subjects with hypertension in the LDM groups. There were no significant differences in BMI or total cholesterol level between the RDM and LDM groups in men or in women. Women with LDM were older than women with RDM.
The age-adjusted CHD mortality rates per 1000 person-years (Table 2) in men with RDM only (13.2; 95% confidence interval [CI], 7.9-22.1) and LDM only (11.4; 95% CI, 6.4-20.3) were similar to the rate seen in women with LDM only (10.7; 95% CI, 7.5-15.5). The age-adjusted CHD mortality curves for the different categories in men and women are shown in the Figure. There was no significant difference (P>.05) in age-adjusted 20-year CHD mortality rate among men with RDM alone (43%), LDM alone (56%), or MI alone (47%). In contrast, the 17-year mortality rate in women with LDM alone was significantly (P<.05) greater (51%) than the 20-year rates in women with RDM alone (22%) or MI alone (42%).
The independent risk for CHD mortality associated with RDM, LDM, or prevalent MI was determined from Cox models (Table 2). After multivariate adjustment, the relative risk for CHD death was similar in men with MI alone (HR, 3.4; 95% CI, 2.5-4.6) RDM (HR, 2.0; 95% CI, 1.0-4.0) or LDM (HR, 2.6; 95% CI, 1.5-4.7). In women, however, LDM was associated with a higher relative risk (HR, 4.8; 95% CI, 3.0-7.8) than RDM (HR, 1.5; 95% CI, 0.9-2.5) or MI (HR, 2.6; 95% CI, 1.6-4.3). To further quantify the risk associated with RDM or LDM, MI was used as the comparison group in the Cox model (Table 2). In men, RDM (HR, 0.7; 95% CI, 0.3-1.3) and LDM (HR, 0.8; 95% CI, 0.4-1.4) had a risk for CHD death equivalent to that of prevalent MI. In women, however, although RDM (HR, 0.9; 95% CI, 0.6-1.3) was associated with a risk for fatal CHD as high as that for MI, LDM (HR, 1.8; 95% CI, 1.1-3.2) had a greater risk for CHD mortality than did prevalent MI.
To formally test whether the relationship between diabetes duration and MI with CHD mortality depends on sex, interactions between sex and the composite diabetes duration–MI variable were tested with men and women combined. Significant interactions (P<.02) were found, implying that the effect of diabetes duration and MI on CHD mortality differs for men and women. The HRs from the model with interactions are presented in Table 3. When compared with women without MI or diabetes, men with RDM had an HR for CHD death of 4.6 and men with LDM had an HR of 4.9. In contrast, women with RDM had an HR of 1.6, whereas women with LDM had an HR of 4.9. Women with LDM had a risk for CHD death similar to that of men with RDM or LDM; ie, LDM seems to remove the female protection against CHD.
To elucidate the effect of each year of diabetes on risk for fatal CHD, Cox models examined duration of diabetes (in years) as a continuous variable while adjusting for baseline age, smoking, total cholesterol level, BMI, hypertension, race, and prevalent MI. Each additional year of diabetes was associated with an HR of 1.07 (95% CI, 1.05-1.09) in men and 1.08 (95% CI, 1.06-1.10) in women.
This population-based analysis reemphasizes the magnitude of diabetes as a major risk factor for CHD mortality in men and women, documents sex differences in the effect of diabetes duration on CHD mortality risk, and quantifies it by comparing the risk with that in individuals with prevalent MI. In men, those with RDM and those with LDM had risks for CHD death equivalent to that for those with prevalent MI. In women, although those with RDM had a risk for CHD death as high as that for those with prevalent MI, LDM was associated with an even greater risk. Because women in the LDM group had a rate of CHD death as high as that of men with diabetes, LDM seems to remove the female protection against CHD death. Finally, each additional year of diabetes was associated with an increased risk for CHD death in both men and women.
Women, but not men, with LDM were older and had a larger proportion of persons with hypertension than did women with RDM. Even after adjusting for these and other risk factors in multivariate Cox models, LDM was associated with an increased risk for CHD mortality. Although sex differences among individuals with diabetes have been noted regarding triglyceride and high-density lipoprotein cholesterol levels,15 endothelial dysfunction,16 levels of lipid peroxidation,17 and nitric oxide production,18 little is known regarding the differential sex effects of diabetes duration. Long-standing diabetes may have greater deleterious effects on all these mechanisms. One previous study postulated that diabetes duration increases incident CHD by effects associated with longer duration of hyperglycemia19 such as longer exposure to a procoagulant state (eg, greater thrombosis risk, accelerated atherosclerosis)20 and increased arterial wall protein glycation leading to luminal occlusion.21
Findings from the Rancho Bernardo Study3 indicate that women with diabetes have a poor survival rate after CHD and that this may be worse in women with diabetes of longer duration. Results of the Nurses’ Health Study,22 which included only women, indicated that the risk for fatal CHD increased monotonically with the duration of diabetes. They found that the relative risks (adjusted for age, BMI, smoking, menopausal status, and family history of premature MI) for fatal CHD in those with diabetes for 11 to 15, 16 to 25, and more than 25 years were 5.5, 6.4, and 11.9, respectively; the relative risk for fatal CHD in those with prevalent CHD in the same model adjusting for the same covariates was 5.5. Data from the Pittsburgh Epidemiology of Diabetes Complications Study,23 which included only patients with type 1 diabetes, as well as a study of elderly Finns with type 2 diabetes,19 showed an association between duration of diabetes and fatal cardiovascular events but did not evaluate sex differences. Our analysis evaluated US population-based data with maximal follow-up of more than 20 years, included men and women belonging to different races, and adjusted for important confounders to provide insights regarding the differential effect of diabetes duration in men and women. Because diabetes and MI have marked sex differences in subsequent CHD rates, it is crucial to analyze them by sex.24 Although previous studies have shown a greater impact of diabetes in women compared with men, they did not evaluate the effect of diabetes duration and did not determine the relative strength of the association by comparison with a marker of increased risk.
As expected, both in men and in women, persons with MI and diabetes had very high CHD mortality, and the patterns are consistent with that seen for the diabetes-only or MI-only groups. In women, the ranking of CHD mortality risk seems to be highest for LDM and MI, followed by LDM only, RDM and MI, MI, and then RDM. In men, the ranking of CHD mortality risk seems to be highest for LDM and MI or RDM and MI, followed by MI, LDM, and then RDM. However, because of the small sample with diabetes and MI, which provides less power to detect differences particularly in studies using a complex sampling design, caution is needed when interpreting and comparing the risk associated with the combined diabetes-MI groups.
The results of this investigation should be interpreted while taking into account certain limitations. First, information regarding family history of CHD, high-density lipoprotein cholesterol level, renal function, type of diabetes treatment, diabetes-related comorbidities, depression, and newer vascular risk factors (levels of fibrinogen, homocysteine, and C-reactive protein) was not available. Therefore, we were unable to adjust for these confounders. Second, information on prevalent MI and diabetes was obtained by self-report, which may have underestimated them because of lack of awareness and the use of less sensitive criteria for diagnosis of diabetes and MI during the time of the baseline examination. Previous studies have demonstrated the validity of self-report for these conditions and this did not differ by sex.25- 27 Third, death certificate information is not completely accurate to classify cause of death, which may have biased our findings toward the null. Fourth, participants were followed up for a 20-year period, and these analyses have not accounted for temporal changes in diagnostic criteria and treatment for diabetes and MI. Fifth, because we did not have information on hormone therapy, we did not adjust for it in the analyses. We adjusted for menopause status, but because it was not significant, it was not included in the final model. Finally, it is very difficult to estimate duration of diabetes accurately.
Despite these limitations, this analysis provides new information regarding the effect of diabetes duration on CHD mortality by quantifying the dramatic impact of LDM in women after accounting for other known CHD risk factors. These findings support the need for intensive approaches to prevent CHD in persons with diabetes. Although improved glycemic control has not been definitively proven to decrease CHD events, the benefits from aggressive treatment of hypertension, dyslipidemia, and platelet responsiveness are established.
Population-based analyses indicate that diabetes prevalence is likely to double in the early 21st century, with a corresponding increase in diabetes-related illness burden.28 Because treating dyslipidemia in diabetic persons without cardiovascular disease may be as cost-effective as treating nondiabetic persons with cardiovascular disease,29 and because women with LDM are at higher risk (based on our data), it is likely that treating them will be even more cost-effective. Because the intensity of cardiovascular preventive measures in diabetes is based on their cardiovascular risk, and because women with LDM may be at higher risk than women with established MI, current guidelines for treating women with LDM may need to be further refined.
This US population-based prospective study demonstrates that the effect of diabetes duration on CHD mortality varies by sex. In men, those with RDM or LDM have a risk for CHD mortality as high as that for men with MI. In women, although those with RDM have a risk for CHD death that is equivalent to that for women with prevalent MI, those with LDM have an even greater risk. Therefore, current cardiovascular prevention recommendations in women may need to be further refined to match intensity of treatment to CHD mortality risk. This analysis should provide the impetus to further improve current diabetes-associated CHD risk assessment to decrease the very high risk for CHD death associated with diabetes.
Correspondence: Sundar Natarajan, MD, MSc, 423 E 23rd St, Room 11101-S, New York, NY 10010 (firstname.lastname@example.org).
Accepted for Publication: August 25, 2004.
Financial Disclosure: None.
Funding/Support: The National Health and Nutrition Examination Survey I (NHANES I) Epidemiologic Follow-up Study was developed and funded by the National Center for Health Statistics, Hyattsville, Md; National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, Ga; the US Department of Agriculture, Washington, DC; and the National Institute on Aging, National Cancer Institute, National Institute of Child Health and Human Development, National Heart, Lung, and Blood Institute, National Institute on Alcohol Abuse and Alcoholism, National Institute of Mental Health, National Institute of Diabetes and Digestive and Kidney Diseases, National Institute of Arthritis and Musculoskeletal and Skin Diseases, National Institute of Allergy and Infectious Diseases, and National Institute of Neurological Disorders and Stroke, Bethesda, Md. This study was supported in part by Research Career Development Award RCD 000211 from the Department of Veterans Affairs Health Services Research and Development Service; and Public Health Service grants DK52329 from the National Institute of Diabetes and Digestive and Kidney Diseases, HL67460 from the National Heart, Lung, and Blood Institute, and AHRQ 10871 from the Agency for Health Research and Quality, Rockville, Md.
Disclaimer: The views expressed in this article are those of the authors and do not necessarily reflect those of the supporting agencies or of the Department of Veterans Affairs.
Additional Information: We obtained public-use NHANES I Epidemiologic Follow-up Study data from the National Center for Health Statistics.