OBJECTIVE—With the current obesity epidemic, one would expect a prevalence increase in the metabolic syndrome. Therefore, in the San Antonio Heart Study, a population-based study with worsening obesity, we examined the metabolic syndrome and its effect on incident cardiovascular disease (CVD).
RESEARCH DESIGN AND METHODS—We enrolled 5,158 subjects in two cohorts: 1979–1982 and 1984–1988. We reexamined 3,682 (71.4%) subjects in 1987–1990 (cohort 1) and 1991–1996 (cohort 2) and assessed a 7.5-year incidence of CVD in 4,635 (90.0%) participants. We used the metabolic syndrome definition of the National Cholesterol Education Program–Adult Treatment Panel III.
RESULTS—At baseline, the metabolic syndrome was less prevalent in cohort 1 than in cohort 2: in men, 20.4 vs. 29.3% (P < 0.001); in women, 16.3 vs. 26.3% (P < 0.001). The prevalence increased in men and women of both Mexican-American and non-Hispanic white ethnic groups between 1979–1982 and 1991–1996 (P for trend <0.001 for each of the groups). There was an excess of incident CVD in cohort 2 relative to cohort 1 (odds ratio 1.37 [95% CI 1.02–1.84]) after adjustment for age, sex, ethnic origin, socioeconomic status, history of CVD, diabetes, total cholesterol, smoking, and family history of heart attack. Further adjustment for the metabolic syndrome reduced this difference (1.26 [0.93–1.71]) because the metabolic syndrome predicted incident CVD (1.58 [1.14–2.18]).
CONCLUSIONS—In San Antonio, Texas, an increase in the prevalence of the metabolic syndrome between 1979–1982 and 1984–1988 contributes to explain a higher CVD incidence.
In the U.S. population, obesity has reached epidemic proportions. In the National Health and Nutrition Examination Survey (NHANES) 1999–2000, 64.5% of subjects aged 20–74 years are overweight or obese (1). Similar trends have been demonstrated for abdominal obesity (2) and type 2 diabetes (3). Abdominal obesity is a central element of the metabolic syndrome (4,5) and a strong predictor of incident metabolic syndrome (4). Thus, it has been hypothesized that the prevalence of the metabolic syndrome has increased in the U.S. population (6,7)
Despite alarming trends in obesity and diabetes, trends in other individual metabolic disorders have been more favorable. Hypertension had been declining before 1988–1991 (8) but may be increasing in recent years (9). HDL cholesterol levels have been reported without significant changes (10). Nevertheless, there is little information regarding the effect of those trends on the prevalence of the metabolic syndrome as well as on cardiovascular disease (CVD) risk, particularly in view of the continuing decline in CVD mortality (11).
In this report, we examined the prevalence of the metabolic syndrome in the San Antonio Heart Study (SAHS), a population-based study with already-characterized adverse trends in obesity (12) and type 2 diabetes (13). We also assessed the effect of the metabolic syndrome on incident CVD.
RESEARCH DESIGN AND METHODS
The SAHS targeted Mexican-American and non-Hispanic white inhabitants of San Antonio. Study protocols were approved by the institutional review board of the University of Texas Health Science Center at San Antonio. All subjects gave written informed consent. Detailed descriptions of this study have been published (13,14). Briefly, all men and nonpregnant women aged 25–64 years residing in randomly selected households from low-, middle-, and high-income census tracts were considered eligible for participation. A total of 5,158 individuals (response rate was 65.3%) were enrolled in two phases: cohort 1, from January 1979 to December 1982 and cohort 2, from January 1984 to December 1988 (14). Of those, 3,682 (71.4%) participants returned to a follow-up examination: cohort 1, between October 1987 and November 1990 (median 8.0 years) and cohort 2, between October 1991 and October 1996 (median 7.4 years) (13). Incident CVD was assessed in 4,635 (90.0%) participants. In 3,581 subjects, assessment was carried out by interview at the follow-up examination or by death certificate. In 1,054 subjects who did not return to the follow-up examination, assessment was accomplished by telephone or home interview.
Acquisition of data and definition of variables and outcomes
Socioeconomic status was assessed at baseline by Duncan’s socioeconomic index, a global measure based on occupational prestige (15). Waist circumference was measured at the level of the umbilicus. Systolic (1st phase) and diastolic (5th phase) blood pressure were recorded with a random-zero sphygmomanometer (Hawksley-Gelman, Sussex, U.K.), with the participant in the sitting position. They were reported as the mean of the second and third blood pressure readings. Blood specimens were obtained after a 12- to 14-h fast. A 75-g oral glucose load (Orangedex; Custom Laboratories, Baltimore, MD) was administered to determine the 2-h glucose value. Plasma glucose and serum lipids were measured with an Abbott Bichromatic Analyzer (South Pasadena, CA) in the laboratory of the Division of Clinical Epidemiology in San Antonio (13,14). LDL cholesterol level was calculated using the Friedewald formula in samples with triglyceride levels <4.5 mmol/l.
We examined diabetes status using the 1999 criteria of the World Health Organization (16). We defined history of CVD as self-reported heart attack, stroke, coronary revascularization procedure, or angina (by Rose Angina questionnaire) at baseline (17) and incident CVD as self-reported heart attack, stroke, or coronary revascularization procedure at follow-up or any mention of cardiovascular death on the death certificate (ICD-9 codes 390–459) (18). We used the Framingham risk equations to estimate the 10-year coronary heart disease risk (19) and the National Cholesterol Education Program–Adult Treatment Panel III (ATPIII) definition of the metabolic syndrome (20). This definition required three or more of the following criteria: elevated waist circumference (>102 cm in men and >88 cm in women), hypertriglyceridemia (≥1.7 mmol/l), low HDL cholesterol (<1.0 mmol/l in men and <1.3 mmol/l in women), high blood pressure (≥130/85 mmHg or pharmacological treatment for hypertension), and elevated fasting glucose (≥5.6 mmol/l or pharmacological treatment for diabetes) (20).
Since waist circumference was not measured in cohort 1 at baseline (1979–1982), we used cohort 2 baseline data (1984–1988) to develop sex-specific logistic regression equations to predict waist circumference from log-transformed weight and height and ethnic origin. We selected the cut point that maximized sensitivity plus specificity. We used imputed elevated waist circumference to generate the prevalence of the metabolic syndrome in 1979–1982. To validate our approach, we imputed elevated waist circumference using the selected cut points in cohort 1 follow-up data (1987–1990). Agreement between ATPIII and imputed cut points for waist circumference was high (κ statistic = 0.69 [95% CI 0.66–0.73]). Agreement between ATPIII-defined (prevalence, 34.3%) and imputed (prevalence, 36.2%) metabolic syndrome was very high (κ statistic = 0.91 [0.89–0.93]).
Data analysis
We performed statistical analyses with the SAS statistical software system version 8.0 (SAS Institute, Cary, NC). We evaluated incident CVD and baseline differences in dichotomous variables by multiple logistic regression analysis. We assessed baseline differences in continuous variables or the logarithm of the odds for Framingham risk estimates of coronary heart disease events over 10 years by one-way ANCOVA. We examined prevalence trends and compared the prevalence of the metabolic syndrome (or individual disorders) between baseline and follow-up by GENMOD, a procedure that fits generalized linear models. GENMOD with the logit link was able to account for the fact that follow-up data represented repeated measurements of the same individuals. Age squared was included as one of the covariates to allow for a nonlinear association between age and the dependent variable. We used the κ statistic to ascertain agreement between ATPIII-defined and imputed waist circumference (or metabolic syndrome). All probability values were two sided.
RESULTS
Compared with cohort 1, many baseline characteristics were different in cohort 2 (Table 1). In cohort 2, low HDL cholesterol and high blood pressure were more frequent, and total cholesterol levels and smoking rates were lower. Elevated fasting glucose was less frequent, but type 2 diabetes was equally prevalent. Additionally, elevated waist circumference was more common in cohort 2 women than in cohort 1 women. From baseline to follow-up, elevated waist circumference and low HDL cholesterol became more prevalent and smoking less frequent. Elevated fasting glucose and type 2 diabetes and total cholesterol levels also became more common in cohort 2 but less in cohort 1.
At baseline, the metabolic syndrome was more frequent in cohort 2 than in cohort 1 (Table 2). This difference was statistically significant in men and women of both ethnic groups (Table 3) and in most of the considered age-groups in both men (30–39 years, P = 0.474; 40–49 years, P = 0.028; 50–59 years, P < 0.001) and women (30–39 years, P = 0.032; 40–49 years, P < 0.001; 50–59 years, P < 0.001) (Fig. 1). During the enrollment years (between 1979 and 1988), the odds for enrolling a subject with the metabolic syndrome went up 7.6% per year in men (95% CI 3.7–11.6) and 9.9% per year in women (6.1–13.8). Similarly, the prevalence of the metabolic syndrome was higher at follow-up than at baseline in both cohorts (Tables 2 and 3; Fig. 1). When we examined both cohorts together, there was a rapid rise from 1979–1982 to 1991–1996 (Table 3).
Despite the adjustment for age, age squared, ethnic origin, and socioeconomic status, cohort 1 had lower baseline Framingham risk scores than cohort 2: in men, 5.76% (95% CI 5.57–5.96) vs. 6.68% (6.48–6.89) (P < 0.001); in women, 1.46% (1.42–1.50) vs. 1.97% (1.91–2.03). A total of 269 (5.8%) individuals developed incident CVD. In a multiple logistic regression analysis, there were more incident CVD events in cohort 2 relative to cohort 1 (odds ratio 1.37 [95% CI 1.02–1.84]), after adjustment for age, sex, ethnic origin, socioeconomic status, history of CVD, diabetes, total cholesterol, smoking, and family history of heart attack. This difference was no longer statistically significant after the addition of the metabolic syndrome as a covariate (1.26 [0.93–1.71]). The metabolic syndrome was an independent predictor of incident CVD (1.58 [1.14–2.18]).
Because diabetes could inflate the prevalence of individual metabolic disorders and drive much of the CVD risk, we performed similar analyses among nondiabetic subjects. At baseline, the metabolic syndrome was less prevalent in cohort 1 than in cohort 2 (Table 2), and Framingham risk scores were also lower in cohort 1: in men, 5.17% (95% CI 5.00–5.35) vs. 5.96% (5.78–6.21) (P < 0.001); in women, 1.18% (1.15–1.21) vs. 1.58% (1.54–1.63). Incident CVD was assessed in 4,105 (90.7%) of 4,524 nondiabetic participants. A total of 170 (4.1%) subjects developed incident CVD events. There was an excess of events in cohort 2 relative to cohort 1 (odds ratio 1.49 [95% CI 1.04–2.12]), independently of age, sex, ethnic origin, socioeconomic status, history of CVD, total cholesterol, smoking, and family history of heart attack. Further adjustment for the metabolic syndrome reduced this difference (1.35 [0.94–1.95]) and demonstrated that the metabolic syndrome predicted incident CVD (1.75 [1.21–2.54]).
Because of the SAHS design (two independent cohorts each examined at two points), a potential source of bias in the assessment of trends was the possibility of baseline differences between participants who returned to follow-up and those who did not. Return rates were lower in cohort 1 diabetic men (P = 0.001) and women (P < 0.001) and in cohort 1 women with elevated fasting glucose (P = 0.030) or the metabolic syndrome (P = 0.018) after the adjustment for age, ethnic origin, and socioeconomic status. Additionally, return rates were also lower in cohort 1 men (P = 0.033) and cohort 2 women (P = 0.027), with low HDL cholesterol levels and in cohort 1 (P = 0.040) and cohort 2 women (P = 0.010) with high blood pressure.
CONCLUSIONS
In San Antonio, Texas, worsening obesity is associated with a rapid rise in the prevalence of the metabolic syndrome. This increase is present in most of the considered age-groups, in men and women, in Mexican Americans and non-Hispanic whites, and in diabetic and nondiabetic individuals. Moreover, the metabolic syndrome has had a negative effect on CVD incidence.
Ford et al. (6) have already described an increase in the prevalence of the metabolic syndrome in the U.S. population aged ≥20 years between 1988–1994 and 1999–2000. The increase is statistically significant in women (from 27.0 to 32.9%, P = 0.014) but not in men (from 31.4 to 31.8%, P = 0.866), even though a prevalence increase of elevated waist circumference is demonstrated in both sexes (2,6). Further studies are needed because sample size is considerably smaller in the NHANES 1999–2000 than the other survey. Besides, a simultaneous increase in obesity and the metabolic syndrome is not always present, as we have recently reported in the Mexico City Diabetes Study (21). In San Antonio, there is a higher prevalence of the metabolic syndrome in cohort 2 than in cohort 1. It is likely that this represents an increase with time, from 1978–1982 to 1984–1988, because both cohorts are cross-sectional surveys of the population in San Antonio. The increase is observed across all age-groups, in men and women, in Mexican Americans and non-Hispanic whites, and in diabetic and nondiabetic subjects. Based on follow-up results, we can also speculate that the prevalence has increased from 1979–1982 to 1991–1996.
Similar to the increase in the prevalence or the metabolic syndrome, Framingham risk scores have also increased. Scores have worsened because changes in blood pressure and HDL cholesterol levels have outrun improvements in smoking rates and total cholesterol levels. Moreover, CVD incidence has also gone up partially because of the increase in the prevalence of the metabolic syndrome.
In San Antonio, prevalence changes in many of the CVD risk factors are similar to reported changes in the U.S. population. Obesity (1,2), diabetes (3,8), and hypertriglyceridemia (22) are increasing, while smoking rates are decreasing (8). Improvement in total cholesterol levels is observed in the SAHS before 1984–1988 but not between 1984–1988 and 1991–1996. Similar changes have been described in the NHANES (improvement before 1988–1994 and no changes between 1988–1994 and 1999–2000) (23) and in the Minnesota Heart Survey (24). In contrast, the SAHS differs from the NHANES regarding changes in HDL cholesterol levels and blood pressure. HDL cholesterol levels have worsened in the SAHS but have improved slightly in the NHANES (22). Two other studies also described opposite changes in HDL cholesterol levels: the Minnesota Heart Survey (24) reports no changes, and a study from New England (25) reports decreasing levels. Similarly, the prevalence of high blood pressure has increased in the SAHS before 1984–1988 but not between 1984–1988 and 1991–1996, whereas hypertension prevalence has decreased in the NHANES before 1988–1991 (8) and increased between 1988–1991 and 1999–2000 (9). HDL cholesterol levels and blood pressure have been partially responsible for the CVD risk increase in San Antonio. Therefore, we cannot infer from our results that a similar change in CVD risk has occurred in the U.S. population.
Prevalence differences in elevated fasting glucose are particularly surprising. The decline between 1979–1982 and 1984–1988 coincides with increases in BMI and the prevalence of the metabolic syndrome as well as with similar diabetes rates. We have no explanation for this result except for differences in the diabetes conversion rate (13). However, no changes in the prevalence of elevated fasting glucose along with worsening central obesity and prevalence of the metabolic syndrome are also present in the NHANES data (6). Additionally, the decline with time in the prevalence of diabetes and elevated fasting glucose in cohort 1 is also unexpected but may be related to lower return rates among subjects who had diabetes or elevated fasting glucose.
Among diabetic individuals, the metabolic syndrome is less prevalent in cohort 1 than in cohort 2. In most studies but not in all (7), the prevalence is >80% (It is 84.3% in the NHANES III [26].) A prevalence increase in the metabolic syndrome in diabetic subjects suggests that the increase in diabetes incidence is primarily driven not by genetic predisposition or sociodemographic factors but by metabolic disorders.
Our study has some limitations inherent to the SAHS design (two cohorts examined at two time points). One concern is that repeated measures of the same individuals are used for the analysis of trend. Cohorts at follow-up could have been enriched with participants with the metabolic syndrome. However, we have not noticed this potential source of bias. Furthermore, lower return rates are demonstrated in subjects with diabetes (cohort 1 men and women) or the metabolic syndrome (cohort 1 women). A second concern is the lack of waist circumference in cohort 1 baseline data. In this dataset, we have imputed waist circumference from weight and height using logistic regression equations. After testing this approach in cohort 1 follow-up data, we consider that the possibility of bias is small. Imputed waist circumference generates a slightly higher prevalence of the metabolic syndrome than measured waist circumference. Also, agreement between ATPIII-defined and imputed metabolic syndrome is very high.
In summary, the increase in the prevalence of the metabolic syndrome has had an adverse impact on CVD risk. Nevertheless, generalization of this finding may be problematic. Changes in HDL cholesterol levels and blood pressure partially explain the increase in CVD risk in San Antonio, but changes in HDL cholesterol levels and blood pressure have been different in the U.S. population. Therefore, further studies need to examine the impact of the metabolic syndrome on secular trends and regional differences in CVD risk in the U.S.
Prevalence of the metabolic syndrome in cohort 1 and cohort 2 participants after stratification by age-group, sex, and time point of examination. Baseline differences between cohort 1 and cohort 2 were assessed by multiple logistic regression analysis using age, ethnic origin, and socioeconomic status as covariates (differences presented in the results section). Baseline and follow-up differences were calculated by generalized linear models. Each generalized linear model had age, age squared, ethnic origin, and socioeconomic status as covariates and accounted for the fact that follow-up data represented repeated measurements of the same individuals (differences presented in the figure).
Prevalence of the metabolic syndrome in cohort 1 and cohort 2 participants after stratification by age-group, sex, and time point of examination. Baseline differences between cohort 1 and cohort 2 were assessed by multiple logistic regression analysis using age, ethnic origin, and socioeconomic status as covariates (differences presented in the results section). Baseline and follow-up differences were calculated by generalized linear models. Each generalized linear model had age, age squared, ethnic origin, and socioeconomic status as covariates and accounted for the fact that follow-up data represented repeated measurements of the same individuals (differences presented in the figure).
Characteristics of participants stratified by sex, cohort, and follow-up status
| Cohort point estimate | Cohort 1 | Cohort 2 | P value† | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1979–1982 | 1987–1990 | P value* | 1984–1988 | 1991–1996 | P value* | ||||||
| Baseline | Follow-up | Baseline | Follow-up | 1979–1982 vs. 1984–1988 | |||||||
| Men | |||||||||||
| n | 957 | 716 | — | 1,264 | 841 | — | — | ||||
| Age (years)‡ | 44.4 ± 0.37 | 52.6 ± 0.41 | — | 43.1 ± 0.32 | 51.7 ± 0.39 | — | — | ||||
| Mexican Americans (%)‡ | 57.7 | 58.0 | — | 66.6 | 65.3 | — | — | ||||
| Socioeconomic index (0–100)‡ | 53 | 53 | — | 50 | 52 | — | — | ||||
| BMI (kg/m2) | 27.4 ± 0.15 | 28.0 ± 0.18 | <0.001 | 27.9 ± 0.13 | 28.9 ± 0.16 | <0.001 | 0.024 | ||||
| Elevated waist circumference (%) | 26.6 (23.8–29.6)§ | 28.6 (25.3–32.1) | 0.075 | 24.4 (22.0–27.0) | 35.2 (31.9–38.7) | <0.001 | 0.278 | ||||
| Hypertriglyceridemia (%) | 43.1 (39.9–46.4) | 42.3 (38.5–46.1) | 0.318 | 43.8 (41.0–46.7) | 49.3 (45.7–52.9) | 0.004 | 0.962 | ||||
| Low HDL cholesterol (%) | 12.4 (10.3–14.7) | 47.6 (43.8–51.4) | <0.001 | 50.8 (47.9–53.6) | 65.5 (62.0–68.8) | <0.001 | <0.001 | ||||
| High blood pressure (%) | 27.5 (24.6–30.7) | 37.5 (33.8–41.4) | <0.001 | 37.6 (34.7–40.6) | 33.8 (30.4–37.4) | 0.174 | <0.001 | ||||
| Elevated fasting glucose (%) | 30.1 (27.0–33.3) | 22.3 (19.4–25.6) | <0.001 | 19.8 (17.5–22.3) | 26.2 (23.1–29.5) | <0.001 | <0.001 | ||||
| Type 2 diabetes (%) | 9.9 (8.1–12.1) | 7.4 (5.8–9.4) | 0.029 | 8.5 (7.0–10.3) | 10.7 (8.7–13.0) | 0.004 | 0.314 | ||||
| Total cholesterol (mmol/l) | 5.58 ± 0.04 | 5.38 ± 0.04 | <0.001 | 5.21 ± 0.03 | 5.46 ± 0.04 | <0.001 | <0.001 | ||||
| LDL cholesterol (mmol/l) | 3.39 ± 0.03 | 3.47 ± 0.04 | 0.105 | 3.29 ± 0.03 | 3.59 ± 0.04 | <0.001 | 0.027 | ||||
| Current cigarette smoking (%) | 35.4 (32.3–38.7) | 24.8 (21.6–28.3) | <0.001 | 30.5 (27.9–33.2) | 23.5 (20.2–26.3) | <0.001 | 0.023 | ||||
| Women | |||||||||||
| n | 1,260 | 958 | — | 1,677 | 1,167 | — | — | ||||
| Age (years)‡ | 43.9 ± 0.31 | 52.3 ± 0.35 | — | 43.4 ± 0.27 | 51.8 ± 0.32 | — | — | ||||
| Mexican Americans (%)‡ | 58.4 | 58.0 | — | 69.8 | 70.1 | — | — | ||||
| Socioeconomic index (0–100)‡ | 51 | 53 | — | 47 | 47 | — | — | ||||
| BMI (kg/m2) | 26.7 ± 0.17 | 27.6 ± 0.20 | <0.001 | 28.0 ± 0.15 | 29.4 ± 0.18 | <0.001 | <0.001 | ||||
| Elevated waist circumference (%) | 32.1 (29.3–35.0)§ | 47.4 (44.0–50.9) | <0.001 | 40.9 (38.4–43.5) | 60.9 (57.7–63.9) | <0.001 | <0.001 | ||||
| Hypertriglyceridemia (%) | 27.7 (25.1–30.5) | 29.8 (26.9–32.9) | 0.355 | 29.1 (26.8–31.5) | 36.3 (33.3–39.3) | <0.001 | 0.703 | ||||
| Low HDL cholesterol (%) | 19.8 (17.6–22.2) | 58.0 (54.6–61.3) | <0.001 | 54.5 (51.9–57.0) | 66.0 (63.0–68.8) | <0.001 | <0.001 | ||||
| High blood pressure (%) | 17.0 (14.8–19.5) | 23.4 (20.7–26.4) | <0.001 | 25.4 (23.1–27.9) | 27.5 (24.8–30.5) | 0.075 | <0.001 | ||||
| Elevated fasting glucose (%) | 20.7 (18.3–23.3) | 16.2 (14.0–18.7) | 0.002 | 15.4 (13.7–17.4) | 20.2 (17.8–22.7) | <0.001 | <0.001 | ||||
| Type 2 diabetes (%) | 8.9 (7.4–10.8) | 7.3 (5.9–9.0) | 0.082 | 9.8 (8.4–11.4) | 12.2 (10.4–14.3) | 0.344 | 0.685 | ||||
| Total cholesterol (mmol/l) | 5.48 ± 0.03 | 5.30 ± 0.03 | <0.001 | 5.17 ± 0.03 | 5.45 ± 0.03 | <0.001 | <0.001 | ||||
| LDL cholesterol (mmol/l) | 3.18 ± 0.03 | 3.31 ± 0.03 | <0.001 | 3.18 ± 0.02 | 3.49 ± 0.03 | <0.001 | 0.900 | ||||
| Current cigarette smoking (%) | 24.6 (22.1–27.2) | 16.2 (13.9–18.8) | <0.001 | 20.7 (18.8–22.8) | 14.0 (12.1–16.3) | <0.001 | 0.014 | ||||
| Cohort point estimate | Cohort 1 | Cohort 2 | P value† | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1979–1982 | 1987–1990 | P value* | 1984–1988 | 1991–1996 | P value* | ||||||
| Baseline | Follow-up | Baseline | Follow-up | 1979–1982 vs. 1984–1988 | |||||||
| Men | |||||||||||
| n | 957 | 716 | — | 1,264 | 841 | — | — | ||||
| Age (years)‡ | 44.4 ± 0.37 | 52.6 ± 0.41 | — | 43.1 ± 0.32 | 51.7 ± 0.39 | — | — | ||||
| Mexican Americans (%)‡ | 57.7 | 58.0 | — | 66.6 | 65.3 | — | — | ||||
| Socioeconomic index (0–100)‡ | 53 | 53 | — | 50 | 52 | — | — | ||||
| BMI (kg/m2) | 27.4 ± 0.15 | 28.0 ± 0.18 | <0.001 | 27.9 ± 0.13 | 28.9 ± 0.16 | <0.001 | 0.024 | ||||
| Elevated waist circumference (%) | 26.6 (23.8–29.6)§ | 28.6 (25.3–32.1) | 0.075 | 24.4 (22.0–27.0) | 35.2 (31.9–38.7) | <0.001 | 0.278 | ||||
| Hypertriglyceridemia (%) | 43.1 (39.9–46.4) | 42.3 (38.5–46.1) | 0.318 | 43.8 (41.0–46.7) | 49.3 (45.7–52.9) | 0.004 | 0.962 | ||||
| Low HDL cholesterol (%) | 12.4 (10.3–14.7) | 47.6 (43.8–51.4) | <0.001 | 50.8 (47.9–53.6) | 65.5 (62.0–68.8) | <0.001 | <0.001 | ||||
| High blood pressure (%) | 27.5 (24.6–30.7) | 37.5 (33.8–41.4) | <0.001 | 37.6 (34.7–40.6) | 33.8 (30.4–37.4) | 0.174 | <0.001 | ||||
| Elevated fasting glucose (%) | 30.1 (27.0–33.3) | 22.3 (19.4–25.6) | <0.001 | 19.8 (17.5–22.3) | 26.2 (23.1–29.5) | <0.001 | <0.001 | ||||
| Type 2 diabetes (%) | 9.9 (8.1–12.1) | 7.4 (5.8–9.4) | 0.029 | 8.5 (7.0–10.3) | 10.7 (8.7–13.0) | 0.004 | 0.314 | ||||
| Total cholesterol (mmol/l) | 5.58 ± 0.04 | 5.38 ± 0.04 | <0.001 | 5.21 ± 0.03 | 5.46 ± 0.04 | <0.001 | <0.001 | ||||
| LDL cholesterol (mmol/l) | 3.39 ± 0.03 | 3.47 ± 0.04 | 0.105 | 3.29 ± 0.03 | 3.59 ± 0.04 | <0.001 | 0.027 | ||||
| Current cigarette smoking (%) | 35.4 (32.3–38.7) | 24.8 (21.6–28.3) | <0.001 | 30.5 (27.9–33.2) | 23.5 (20.2–26.3) | <0.001 | 0.023 | ||||
| Women | |||||||||||
| n | 1,260 | 958 | — | 1,677 | 1,167 | — | — | ||||
| Age (years)‡ | 43.9 ± 0.31 | 52.3 ± 0.35 | — | 43.4 ± 0.27 | 51.8 ± 0.32 | — | — | ||||
| Mexican Americans (%)‡ | 58.4 | 58.0 | — | 69.8 | 70.1 | — | — | ||||
| Socioeconomic index (0–100)‡ | 51 | 53 | — | 47 | 47 | — | — | ||||
| BMI (kg/m2) | 26.7 ± 0.17 | 27.6 ± 0.20 | <0.001 | 28.0 ± 0.15 | 29.4 ± 0.18 | <0.001 | <0.001 | ||||
| Elevated waist circumference (%) | 32.1 (29.3–35.0)§ | 47.4 (44.0–50.9) | <0.001 | 40.9 (38.4–43.5) | 60.9 (57.7–63.9) | <0.001 | <0.001 | ||||
| Hypertriglyceridemia (%) | 27.7 (25.1–30.5) | 29.8 (26.9–32.9) | 0.355 | 29.1 (26.8–31.5) | 36.3 (33.3–39.3) | <0.001 | 0.703 | ||||
| Low HDL cholesterol (%) | 19.8 (17.6–22.2) | 58.0 (54.6–61.3) | <0.001 | 54.5 (51.9–57.0) | 66.0 (63.0–68.8) | <0.001 | <0.001 | ||||
| High blood pressure (%) | 17.0 (14.8–19.5) | 23.4 (20.7–26.4) | <0.001 | 25.4 (23.1–27.9) | 27.5 (24.8–30.5) | 0.075 | <0.001 | ||||
| Elevated fasting glucose (%) | 20.7 (18.3–23.3) | 16.2 (14.0–18.7) | 0.002 | 15.4 (13.7–17.4) | 20.2 (17.8–22.7) | <0.001 | <0.001 | ||||
| Type 2 diabetes (%) | 8.9 (7.4–10.8) | 7.3 (5.9–9.0) | 0.082 | 9.8 (8.4–11.4) | 12.2 (10.4–14.3) | 0.344 | 0.685 | ||||
| Total cholesterol (mmol/l) | 5.48 ± 0.03 | 5.30 ± 0.03 | <0.001 | 5.17 ± 0.03 | 5.45 ± 0.03 | <0.001 | <0.001 | ||||
| LDL cholesterol (mmol/l) | 3.18 ± 0.03 | 3.31 ± 0.03 | <0.001 | 3.18 ± 0.02 | 3.49 ± 0.03 | <0.001 | 0.900 | ||||
| Current cigarette smoking (%) | 24.6 (22.1–27.2) | 16.2 (13.9–18.8) | <0.001 | 20.7 (18.8–22.8) | 14.0 (12.1–16.3) | <0.001 | 0.014 | ||||
Data are means ± SE or odds ratio (95% CI).
P value was adjusted for age, age squared, ethnic origin, and socioeconomic status and accounted for the fact that follow-up data were repeated measurements of the same individuals.
P value was adjusted for age, age squared, ethnic origin, and socioeconomic status.
Indicates nonadjusted values.
Prevalence calculated using imputed waist circumference.
Prevalence of the metabolic syndrome stratified by sex, cohort, period of examination, and baseline diabetes status
| Cohort point estimate | Cohort 1 | Cohort 2 | P value† | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1979–1982 | 1987–1990 | P value* | 1984–1988 | 1991–1996 | P value* | ||||||
| Baseline | Follow-up | Baseline | Follow-up | 1979–1982 vs. 1984–1988 | |||||||
| Men | |||||||||||
| All individuals | 20.4 (17.8–23.2)‡ | 29.1 (25.8–32.7) | <0.001 | 29.3 (26.7–32.0) | 38.0 (34.6–41.6) | <0.001 | <0.001 | ||||
| Nondiabetic individuals | 15.5 (13.1–18.2)‡ | 25.8 (22.4–29.5) | <0.001 | 23.3 (20.8–26.0) | 30.4 (26.9–34.1) | <0.001 | <0.001 | ||||
| Diabetic individuals | 57.5 (47.3–67.2)‡ | 66.2 (55.7–75.3) | 0.072 | 79.4 (70.8–85.9) | 85.2 (78.3–90.2) | 0.246 | <0.001 | ||||
| Women | |||||||||||
| All individuals | 16.3 (14.2–18.7)‡ | 28.8 (25.8–32.0) | <0.001 | 26.3 (24.0–28.6) | 38.6 (35.5–41.8) | <0.001 | <0.001 | ||||
| Nondiabetic individuals | 10.8 (9.0–12.9)‡ | 22.6 (19.8–25.7) | <0.001 | 18.7 (16.6–20.9) | 29.5 (26.4–32.8) | <0.001 | <0.001 | ||||
| Diabetic individuals | 65.8 (56.1–74.3)‡ | 86.7 (79.1–91.8) | <0.001 | 83.1 (76.8–88.0) | 86.9 (81.8–90.7) | 0.323 | 0.001 | ||||
| Cohort point estimate | Cohort 1 | Cohort 2 | P value† | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1979–1982 | 1987–1990 | P value* | 1984–1988 | 1991–1996 | P value* | ||||||
| Baseline | Follow-up | Baseline | Follow-up | 1979–1982 vs. 1984–1988 | |||||||
| Men | |||||||||||
| All individuals | 20.4 (17.8–23.2)‡ | 29.1 (25.8–32.7) | <0.001 | 29.3 (26.7–32.0) | 38.0 (34.6–41.6) | <0.001 | <0.001 | ||||
| Nondiabetic individuals | 15.5 (13.1–18.2)‡ | 25.8 (22.4–29.5) | <0.001 | 23.3 (20.8–26.0) | 30.4 (26.9–34.1) | <0.001 | <0.001 | ||||
| Diabetic individuals | 57.5 (47.3–67.2)‡ | 66.2 (55.7–75.3) | 0.072 | 79.4 (70.8–85.9) | 85.2 (78.3–90.2) | 0.246 | <0.001 | ||||
| Women | |||||||||||
| All individuals | 16.3 (14.2–18.7)‡ | 28.8 (25.8–32.0) | <0.001 | 26.3 (24.0–28.6) | 38.6 (35.5–41.8) | <0.001 | <0.001 | ||||
| Nondiabetic individuals | 10.8 (9.0–12.9)‡ | 22.6 (19.8–25.7) | <0.001 | 18.7 (16.6–20.9) | 29.5 (26.4–32.8) | <0.001 | <0.001 | ||||
| Diabetic individuals | 65.8 (56.1–74.3)‡ | 86.7 (79.1–91.8) | <0.001 | 83.1 (76.8–88.0) | 86.9 (81.8–90.7) | 0.323 | 0.001 | ||||
Data are odds ratio (95% CI).
P value was adjusted for age, age squared, ethnic origin, and socioeconomic status and accounted for the fact that follow-up data were repeated measurements of the same individuals.
P value was adjusted for age, age squared, ethnic origin, and socioeconomic status.
Prevalence calculated using elevated waist circumference imputed from weight, height, sex, and ethnic origin.
Odds ratios for having the metabolic syndrome between periods of examination*
| Odds ratio for having the metabolic syndrome in 1987–1990 (follow-up) relative to 1979–1982 (baseline) in cohort 1 participants | ||||||||
|---|---|---|---|---|---|---|---|---|
| Period of examination | Model 1 | Model 2 | Model 3 | Model 4 | ||||
| Mexican-American men | Non-Hispanic white men | Mexican-American women | Non-Hispanic white women | |||||
| 1987–1990 vs. 1979–1982 | 1.67 (1.29–2.16) | 1.63 (1.11–2.39) | 2.08 (1.62–2.67) | 2.35 (1.72–3.21) | ||||
| Odds ratio for having the metabolic syndrome in 1991–1996 (follow-up) relative to 1984–1988 (baseline) in cohort 2 participants | ||||||||
| Period of examination | Model 5 | Model 6 | Model 7 | Model 8 | ||||
| Mexican-American men | Non-Hispanic white men | Mexican-American women | Non-Hispanic white women | |||||
| 1991–1996 vs. 1984–1988 | 1.40 (1.14–1.71) | 1.61 (1.18–2.19) | 1.74 (1.46–2.06) | 1.63 (1.19–2.24) | ||||
| Odds ratio for having the metabolic syndrome in 1984–1988, 1987–1990, and 1991–1996 relative to 1979–1982 using data from both cohorts | ||||||||
| Period of examination | Model 9 | Model 10 | Model 11 | Model 12 | ||||
| Mexican-American men | Non-Hispanic white men | Mexican-American women | Non-Hispanic white women | |||||
| 1984–1988 vs. 1979–1982 | 1.70 (1.31–2.22) | 1.53 (1.07–2.19) | 1.86 (1.46–2.38) | 1.80 (1.23–2.63) | ||||
| 1987–1990 vs. 1979–1982 | 1.66 (1.30–2.11) | 1.64 (1.20–2.25) | 1.99 (1.57–2.51) | 2.25 (1.66–3.06) | ||||
| 1991–1996 vs. 1979–1982 | 2.38 (1.79–3.16) | 2.47 (1.71–3.58) | 3.30 (2.55–4.27) | 3.03 (2.08–4.39) | ||||
| Odds ratio for having the metabolic syndrome in 1987–1990 (follow-up) relative to 1979–1982 (baseline) in cohort 1 participants | ||||||||
|---|---|---|---|---|---|---|---|---|
| Period of examination | Model 1 | Model 2 | Model 3 | Model 4 | ||||
| Mexican-American men | Non-Hispanic white men | Mexican-American women | Non-Hispanic white women | |||||
| 1987–1990 vs. 1979–1982 | 1.67 (1.29–2.16) | 1.63 (1.11–2.39) | 2.08 (1.62–2.67) | 2.35 (1.72–3.21) | ||||
| Odds ratio for having the metabolic syndrome in 1991–1996 (follow-up) relative to 1984–1988 (baseline) in cohort 2 participants | ||||||||
| Period of examination | Model 5 | Model 6 | Model 7 | Model 8 | ||||
| Mexican-American men | Non-Hispanic white men | Mexican-American women | Non-Hispanic white women | |||||
| 1991–1996 vs. 1984–1988 | 1.40 (1.14–1.71) | 1.61 (1.18–2.19) | 1.74 (1.46–2.06) | 1.63 (1.19–2.24) | ||||
| Odds ratio for having the metabolic syndrome in 1984–1988, 1987–1990, and 1991–1996 relative to 1979–1982 using data from both cohorts | ||||||||
| Period of examination | Model 9 | Model 10 | Model 11 | Model 12 | ||||
| Mexican-American men | Non-Hispanic white men | Mexican-American women | Non-Hispanic white women | |||||
| 1984–1988 vs. 1979–1982 | 1.70 (1.31–2.22) | 1.53 (1.07–2.19) | 1.86 (1.46–2.38) | 1.80 (1.23–2.63) | ||||
| 1987–1990 vs. 1979–1982 | 1.66 (1.30–2.11) | 1.64 (1.20–2.25) | 1.99 (1.57–2.51) | 2.25 (1.66–3.06) | ||||
| 1991–1996 vs. 1979–1982 | 2.38 (1.79–3.16) | 2.47 (1.71–3.58) | 3.30 (2.55–4.27) | 3.03 (2.08–4.39) | ||||
Data are odds ratio (95% CI).
Each generalized linear model had age, age squared, socioeconomic status, and period of examination as independent variables and accounted for the fact that follow-up data represented repeated measurements of the same individuals.
Article Information
This work was supported by grants from the National Heart, Lung, and Blood Institute (RO1-HL24799 and RO1-HL36820).
References
A table elsewhere in this issue shows conventional and Système International (SI) units and conversion factors for many substances.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
