From the Department of Family and Preventive Medicine, University of California, San Diego
The metabolic syndrome is a cluster of physiologic and metabolic risk factors linked to cardiovascular disease. Over the years, a number of different sets of criteria defining the metabolic syndrome have been proposed; the definition contained in the National Cholesterol Education Program Adult Treatment Panel III (ATP III) recommendations is the most recent and widely publicized of these.1 In the present study, we examine the association of the metabolic syndrome, as identified based on the ATP III criteria, with myocardial infarction (MI) and stroke in a representative sample of American adults.
Patients and methods
The data source for our analysis was the National Health and Nutrition Examination Survey III (NHANES III), a survey conducted by the National Center for Health Statistics of the Centers for Disease Control and Prevention from 1988 to 1994. NHANES III is notable for its large size, its extensive anthropometric, physiologic, and laboratory testing (including physician examinations), and its complex, multistage sampling design, which allows results to be extrapolated to the total US population. Details of the data collection protocol have been described elsewhere.2 Our study was based on a subset of 15,922 NHANES subjects aged 20 to 89 years for whom the relevant data were available. NHANES III is a cross-sectional study. The events of interest, MI and stroke, were identified based on subjects’ self-reports of history of physician-diagnosed disease.
Using ATP III criteria, the metabolic syndrome was identified based on the presence of at least three of the following five component conditions: insulin resistance, hypertension, hypertriglyceridemia, low high-density lipoprotein (HDL) cholesterol, and abdominal obesity, defined by high waist circumference.1 Insulin resistance was identified as fasting glucose of 110 mg/dL or above; for the purposes of this analysis, insulin resistance could also be identified by a self-report of current use of insulin or oral hypoglycemic agents. Hypertension was defined as systolic blood pressure of 130 mm Hg or above or diastolic blood pressure of 85 mm Hg or above; for the purposes of this analysis, the hypertension could also be identified by a self-report of current use of antihypertension medication. Hypertriglyceridemia was identified based on triglyceride levels of 150 mg/dL or above. Low HDL cholesterol was identified as below 40 mg/dL in men or below 50 mg/dL in women. High waist circumference was defined as greater than 102 cm in men or greater than 88 cm in women.
Logistic regression analysis was used to determine the association
of the metabolic syndrome and its component conditions with MI, stroke, and combined MI and/or stroke (MI/stroke). We controlled for age, sex, race, and smoking status as possible confounders. Analyses were performed using the SUDAAN software, which takes into account the complex sampling design of NHANES III and allows weighting of results and computation of confidence intervals (CIs) and statistical tests appropriate to the total US population.3
Of 15,922 subjects, 752 had a history of MI, 464 had a history of stroke, and 1,098 had a history of MI/stroke, with unadjusted prevalences of 4.7%, 2.9%, and 6.9%, respectively. When weighted to the general US population, the prevalences were 3.7%, 2.0%, and 5.2% for MI, stroke, and MI/stroke, respectively.
Table 1 shows the results of logistic regression models estimating the association of MI and stroke with the full metabolic syndrome and its component conditions, adjusted for age, sex, race, and smoking status. Two models—one for the full metabolic syndrome and another for the five component conditions—are shown for each disease history. Table 2 shows sex-specific results limited to MI/stroke for power reasons. Odds ratios (ORs) and CIs for the covariates of age, sex, race, and smoking are shown for the full syndrome model; those from the individual component conditions models (not shown) were very similar in direction and magnitude. Results are presented graphically in the figure.
Based on ATP III criteria, the metabolic syndrome was significantly associated with MI/stroke (OR, 2.05; 95% CI, 1.64—2.57), MI (OR, 2.01; 95% CI, 1.53–2.64), and stroke (OR, 2.16; 95% CI, 1.48–3.16; table 1). It was significantly associated with MI/stroke in both men (OR, 1.93; 95% CI, 1.34–2.78) and women (OR, 2.20; 95% CI, 1.56–3.11; table 2). This difference in ORs between men and women was not statistically significant based on interaction models.
Among the individual metabolic syndrome components, low HDL cholesterol was significantly associated with MI and MI/stroke, but not with stroke alone. Low HDL cholesterol was significantly associated with MI/stroke in men, but not in women. Insulin resistance was significantly associated with MI/stroke (OR, 1.30; 95% CI, 1.03—1.66), but not with MI or stroke separately. Hypertension was of borderline significance for MI/stroke (OR, 1.44; 95% CI, 1.00–2.08; P = .051); OR for MI and stroke separately were similar in magnitude and of borderline significance (P < .10). MI/stroke was significantly related to hyperten-sion in women (OR, 2.19; 95% CI, 1.15–4.14), but not in men. Triglycerides had the largest OR of any of the component conditions for MI, stroke, and MI/stroke (OR, 1.66; 95% CI, 1.20–2.30 for MI/stroke). This association was stronger in women; the association in men was only of borderline significance. Abdominal obesity (high waist circumference) was the only component that was not independently related to prevalent disease in any model.
Among the covariates, age was highly significantly related to prevalent disease, with ORs of 1.08 to 1.09 per additional year of age in all models. Female sex was significantly protective for combined MI/ stroke and MI, but not for stroke. Both current and past history of smoking were associated with significantly elevated odds of combined MI/stroke and MI; only current smoking was significantly associated with stroke.
The only statistically significant ethnic difference was seen in non-Hispanic blacks. In multivariate models, non-Hispanic blacks had significantly higher odds of MI/
stroke as compared with non-Hispanic whites (OR, 1.36; 95% CI, 1.06—1.76). Non-Hispanic blacks had increased odds of disease that were significant for stroke (OR, 1.49; 95% CI, 1.03–2.17) but not for MI. In sex-specific models, a significantly elevated risk among non-Hispanic blacks was seen in women (OR, 1.87; 95% CI, 1.31–2.67) but not men (OR, 1.05; 95% CI, 0.73–1.51).
The present study shows that the metabolic syndrome, as defined by ATP III criteria, is independently associated with a history of MI/
stroke, MI, and stroke. All of the component conditions of the syndrome, with the exception of abdominal obesity, are also independently associated with prevalent disease.
Previous studies have shown linear relationships between coronary heart disease and either body mass index (BMI) or waist-to-hip ratio independent of BMI after adjusting for other risk factors.4 The lack of an independent association of high waist circumference with prevalent disease in our study suggests that some of the effect of high waist
circumference is mediated by its role as a risk factor for other components of the syndrome. If this were so, we would expect that a model excluding the other component conditions would reveal a significant relationship between high waist circumference and MI/stroke, which, in fact, it does (OR, 1.43; 95% CI, 1.15—1.78).
One major limitation of our study is its cross-sectional design. Although NHANES III is a useful data source because of its size, scope, and level of detail, its cross-sectional nature makes it impossible to prove conclusively that the hypothesized causes (the metabolic syndrome) actually preceded the observed effects (MI and stroke). This type of analysis also excludes patients whose MI or stroke resulted in death.
Three prospective studies have examined the risks associated with the metabolic syndrome. Lakka and colleagues reported the association of the metabolic syndrome with cardiovascular disease and overall mortality in 1,209 Finnish men free of cardiovascular disease, diabetes, and cancer at baseline.5 They reported a lower prevalence of the syndrome (based on the ATP III definition) than in our study (8.8% versus 24.0%). Compared with the OR
for nonfatal MI of 2.01 in our study (95% CI, 1.53—2.64), they reported
a hazard ratio of 4.16 (95% CI, 1.60—10.8) for coronary mortality. In another analysis, McNeill and colleagues studied the association of the metabolic syndrome (based on ATP III criteria) with incident coronary artery disease in a population sample. They reported a hazard ratio of 2.58 for women and 1.74 for men, results that are also quite similar to our estimates.6 Finally, Malik and colleagues reported that the metabolic syndrome in NHANES II was associated with a significant increase in cardiovascular (OR, 2.29) and total (OR, 1.39) mortality.7 Thus, the limited available prospective data confirm our findings.
Our study further supports the clinical importance of the metabolic syndrome as a significant risk factor for cardiovascular disease. The results emphasize the need
for strategies to control this syndrome as well as its component conditions.