Relationship Between Metabolic Risk Factor Clustering and Cardiovascular Mortality Stratified by High Blood Glucose and Obesity: NIPPON DATA90, 1990–2000

Cohort studies of the National Survey on Circulatory Disorders, Japan, are referred to as NIPPON DATA (National Integrated Project for Prospective Observation of Noncommunicable Disease and Its Trends in the Aged). NIPPON DATA includes two cohort studies. Baseline data were surveyed in 1980 and in 1990 (NIPPON DATA80 and NIPPON DATA90), and the details of these cohorts have been reported (10–15). Here, we analyzed data from NIPPON DATA90 because the baseline survey of NIPPON DATA80 does not include some important metabolic factors such as HDL cholesterol.

A total of 8,384 residents (3,504 men and 4,880 women, aged ≥30 years) from 300 randomly selected districts participated in the survey and were followed until 15 November 2000. The participation rate in this survey was 76.5%. Of the 8,384 participants, 1,165 were excluded because of a history of coronary heart disease or stroke (n = 371), information missing at the baseline survey (n = 636), and failure to access because of incomplete residential access information at the first survey (n = 158). The remaining 7,219 participants (2,999 men and 4,220 women) were included in the analysis.

The underlying causes of death in the National Vital Statistics were coded according to the ICD-9 until the end of 1994 and according to the ICD-10 from the start of 1995 until the end of 2000. Details of these classifications are described elsewhere (10–15). The Institutional Review Board of Shiga University of Medical Science (No. 12–18, 2000) approved this study.

Nonfasting blood samples were obtained at the baseline survey. The serum was separated and centrifuged soon after blood coagulation. Plasma samples were collected in siliconized tubes containing sodium fluoride and shipped to one laboratory (SRL, Tokyo, Japan) for blood measurements. Plasma glucose was measured enzymatically. Serum triglycerides and total cholesterol were also measured enzymatically, and HDL cholesterol was measured after heparin-calcium precipitation (16).

BMI was calculated as weight in kilograms divided by the square of height in meters. Baseline blood pressure was measured by trained observers using a standard mercury sphygmomanometer on the right arm of seated participants. Public health nurses obtained information on smoking, alcohol consumption, physical activity, and medical history. We divided participants into four categories of smokers (never-smoked, ex-smoker, and current smoker <20 or ≥20 cigarettes/day) and six categories of drinking (never-drinker; ex-drinker; and current drinker of 1, 2, 3, and 4 gou of sake/day; 1 gou [180 ml] is equivalent to 23 g of alcohol) (11). We divided participants into three categories of physical activity (yes or no for physical problems, and no for any other reason).

We defined metabolic factors as follows: obesity, BMI ≥25 kg/m²; high blood pressure, systolic blood pressure ≥130 mmHg, diastolic blood pressure ≥85 mmHg, administration of antihypertensive agents, or any combination of these; and high blood glucose, serum glucose ≥140 mg/dl, medication for diabetes, or both. Because our samples were nonfasting, the postload blood glucose level for diagnosis of impaired glucose tolerance was ≥140 mg/dl (17). We defined high triglycerides as nonfasting serum triglyceride ≥200 mg/dl and also as taking medication for dyslipidemia. Low HDL cholesterol was defined as serum HDL cholesterol ≤40 mg/dl for men and ≤50 mg/dl for women.

Continuous variables were compared using ANOVA, and the χ² test was used to compare the dichotomized variables to examine differences in baseline characteristics of participants according to the numbers of clustering metabolic factors.

The multivariate adjusted hazard ratio (HR) of all CVD mortality for each group was calculated using the Cox proportional hazards model adjusted for age, sex, total cholesterol, smoking, drinking, and physical activity category. When we calculated HR for an individual component of a metabolic factor, we further adjusted for other components of the metabolic factor. We used nonobese participants without any metabolic factor or participants with neither a metabolic factor nor high blood glucose as references in analyses stratified by obesity or high blood glucose (required component by the IDF and WHO, respectively). Because leaner participants also have a higher CVD mortality risk in Japan, we further analyzed a data subset excluding leaner participants (BMI <18.5 kg/m²) (18,19).

All CIs were estimated at the 95% level. P < 0.05 was considered significant. The Statistical Package for the Social Sciences (version 11.0J; SPSS Japan, Tokyo, Japan) was used to perform all analyses.

Table 1 shows the baseline characteristics of the study participants according to the numbers of metabolic factors. Total person-years were 69,170, and the mean follow-up period was 9.6 years. During follow-up, 625 participants died of all causes and 173 died of CVD. Table 2 shows the multiple adjusted HRs and 95% CIs according to individual components of metabolic risk factors.

Table 3 shows the number of deaths, multiple adjusted HRs, and 95% CIs according to various numbers of metabolic factors. The HRs for CVD mortality were higher in the group with more metabolic factors, but the trend was not statistically significant (P_trend = 0.074). The relationship between numbers of risk factors and CVD mortality did not differ according to sex (P_interaction = 0.70). We therefore combined men and women in the following analyses. The tendency for HR to be higher in those with more metabolic factors was similar for heart disease (three risk factors: HR 2.08 [95%CI 0.67–6.48]; four risk factors: 3.97[1.24–12.72]; five risk factors: not applicable) and stroke (three risk factors: 2.07 [0.67–6.37]; four risk factors: 1.23 [0.30–5.05]; five risk factors: 6.26 [CI, 1.13–34.60]) mortality. The HR tendency for all-cause mortality was similar, but the number of clustering metabolic factors was not significantly related to all-cause mortality (three risk factors: 1.16 [0.81–1.65]; four risk factors: 1.18 [0.77–1.80]; five risk factors: 1.44 [0.57–3.63]).

Table 4 shows multiple adjusted HRs (95% CI) due to the number of metabolic factors except high blood glucose stratified by high blood glucose. The HRs trended to increase in both groups (with and without high blood glucose). The HR for CVD in participants with ≥3 metabolic factors but high blood glucose was modest and not statistically significant. Conversely, HRs were obviously higher for participants with high blood glucose and ≥2 other metabolic factors than those for participants with neither metabolic factors nor high blood glucose. The risk increases were statistically significant.

Table 4 also shows multiple adjusted HRs (95% CI) for CVD mortality according to the number of metabolic factors other than obesity stratified by obesity. The relationship between HRs and the numbers of metabolic factors was positive in both obese and nonobese groups. This relationship was unchanged when participants with lower BMI (≥18.5 kg/m²) were excluded.

We found that metabolic factor clustering was positively associated with CVD mortality in the general Japanese population. The risk increase in participants with both high blood glucose and ≥2 metabolic factors was significantly higher than in those with neither high blood glucose nor metabolic risk factors. The risk in nonobese participants with more metabolic factors was also increased.

Although investigating the relationship between metabolic factor clustering and CVD mortality is important, prospective studies on the topic are still scarce. On the basis of the NCEP and WHO definitions of metabolic syndrome, several investigators have reported that participants with metabolic syndrome or metabolic factor clustering have a high HR of CVD mortality (20–25). Ford (26) summarized prospective cohort studies and reported that the HRs of CVD mortality were 1.65 [5% CI 1.38–1.99] according to the NCEP definition and 1.93 [1.39–2.67] according to the WHO definition, respectively. This result is consistent with our findings that participants with more metabolic factors have a higher risk of CVD mortality. Our results were also comparable with those of a prospective study in Japan showing that the relative risk of cardiac diseases was 2.23 [1.14–4.34] in participants with ≥3 metabolic factors compared with that in participants with <3 metabolic factors (27).

The IDF definition requires obesity for diagnosis of metabolic syndrome. These guidelines explain that central (abdominal) obesity is a prerequisite for this diagnosis because it is easy to assess and independently associated with each of the other metabolic syndrome components (8). The IDF guidelines do not essentially require insulin resistance because it is difficult to measure in day-to-day clinical practice (7,8). However, although increased waist circumference is an important component of metabolic syndrome, some individuals with multiple risk factors and an increased risk of CVD mortality have normal waist circumference (28,29). For example, Katzmarzyk et al. (28) reported that waist circumference is a valuable component of metabolic syndrome, but they also raised the concern that the IDF requirement of an increased waist circumference warranted caution because a large proportion of individuals with normal waist circumference also have multiple risk factors and an increased risk of mortality.

We found here that nonobese participants with three or more metabolic factors had significantly higher HRs for CVD death and that their risk was similar to that of obese participants with the corresponding number of metabolic factors. Thus, a proportion of high-risk participants might be overlooked if obesity is a diagnostic requirement for metabolic syndrome. Waist circumference supposedly indicates visceral fat more accurately than BMI in terms of predicting diabetes (30). However, we did not have any information about waist circumference and used BMI as it closely correlates with waist circumference. Furthermore, BMI has been used to diagnose obesity in many epidemiological studies of metabolic syndrome (22,23), indicating that BMI was acceptable for our purposes. However, because of the use of BMI, we might have underestimated the impact of obesity on CVD mortality. A similar study using waist circumference should clarify the relation.

The WHO guidelines indicate that the presence of diabetes, impaired glucose tolerance, or insulin resistance is necessary for a diagnosis of metabolic syndrome because this condition is considered a special classification for those with the potential for diabetes (manifested as impaired glucose tolerance, impaired fasting glucose, or insulin resistance determined using the hyperinsulinemic-euglycemic clamp) (1,7). Here, we also stratified participants according to blood glucose level and found that the HR was higher among those with than among those without high blood glucose. These findings suggest that glucose tolerance plays an important role in CVD mortality. Some reports have shown higher HRs with use of the WHO rather than the NCEP definition of metabolic syndrome. This result means that the participants with impaired glucose tolerance have higher HRs, a finding that the present results support (26). However, several participants with clustering of metabolic factors other than impaired glucose tolerance also had an increased risk of CVD mortality.

Some limitations other than using BMI should be noted about the present study. First, we used nonfasting blood samples and thus we might have misclassified participants with high blood glucose or hypertriglyceridemia. Second, we did not adjust for socioeconomic status because relevant information was not available. However, all Japanese are covered by the national health insurance program and socioeconomic status does not affect access to treatment. Therefore, the impact of socioeconomic status on our findings should be minimal.

In summary, the CVD risk was obviously higher among individuals with than among those without high blood glucose and multiple metabolic risk factors, suggesting that high blood glucose plays an important role in CVD mortality. Conversely, the prevalence of nonobese participants with several metabolic factors was quite high and their CVD risk was high. Thus, metabolic factors should be carefully considered and appropriately managed even among individuals with a BMI <25.

This study was supported by a grant-in-aid from the Ministry of Health and Welfare under the auspices of the Japanese Association for Cerebro-cardiovascular Disease Control, a Research Grant for Cardiovascular Diseases (7A-2) from the Ministry of Health, Labor and Welfare, and a Health and Labor Sciences Research Grant, Japan (Comprehensive Research on Aging and Health: H11-Chouju-046, H14-Chouju-003, and H17-Chouju-012).