Discrimination of Health Risk by Combined ... - Wiley Online Library

Discrimination of Health Risk by Combined Body Mass Index and Waist Circumference Christopher I. Ardern,* Peter T. Katzmarzyk,* Ian Janssen,† and Robert Ross*‡

Abstract ARDERN, CHRISTOPHER I., PETER T. KATZMARZYK, IAN JANSSEN, AND ROBERT ROSS. Discrimination of health risk by combined body mass index and waist circumference. Obes Res. 2003;11:135–142. Objective: NIH Clinical Guidelines (1998) recommend the measurement of waist circumference (WC, centimeters) within body mass index (BMI, kilograms per square meter) categories as a screening tool for increased health risk. Research Methods and Procedures: The Canada Heart Health Surveys (1986 through 1992) were used to describe the prevalence of the metabolic syndrome in Canada and to test the use of the NIH guidelines for predicting metabolic risk factors. The sample included 7981 participants ages 20 to 74 years who had complete data for WC, BMI, highdensity lipoprotein-cholesterol, triglycerides, diabetic status, and systolic and diastolic blood pressures. National Cholesterol Education Program Adult Treatment Panel III risk categories were used to identify the metabolic syndrome and associated risk factors. Logistic regression was used to test the hypothesis that WC improves the prediction of the metabolic syndrome, within overweight (25 to 29.9 kg/m2) and obese I (30 to 34.9 kg/m2) BMI categories. Results: The prevalence of the metabolic syndrome was 17.0% in men and 13.2% in women. The odds ratios (OR) for the prediction of the metabolic syndrome were elevated in overweight [OR, 1.85; 95% confidence interval (95%CI), 1.02 to 3.35] and obese (OR, 2.35; 95%CI, 1.25 to 4.42) women with a high WC compared with overweight and obese women with a low WC, respectively. On the other hand, WC was not predictive of the metabolic syndrome or component risk factors in men, within BMI categories.

Received for review May 20, 2002. Accepted for publication in final form October 2, 2002. *School of Physical and Health Education, Queen’s University, Kingston, Ontario, Canada; †Jean Mayer USDA Human Nutrition Research Center on Aging, Tufts University, Boston, MA; and ‡Division of Endocrinology and Metabolism, Department of Medicine, Queen’s University, Kingston, Ontario, Canada. Address correspondence to Peter T. Katzmarzyk, School of Physical and Health Education, Queen’s University, Kingston, Ontario, K7L 3N6 Canada. E-mail: [email protected] Copyright © 2003 NAASO

Discussion: In women already at increased health risk because of an elevated BMI, the additional measurement of WC may help identify cardiovascular risk. Key words: cardiovascular risk, anthropometry, Canada Heart Health Surveys

Introduction The metabolic syndrome, operationally defined as the presence of any three of the following factors: central obesity (high waist circumference), hyperglycemia, high blood pressure, low high-density lipoprotein-cholesterol (HDLC),1 or high triglycerides (1), has recently been recognized as a public health concern in the United States (2). While the metabolic syndrome alone is a condition of disabling medical sequelae, it is an intermediary step in the progression of a number of vascular and coronary-related clinical events. Observational evidence (3) suggests that central obesity is a key element of the metabolic syndrome, and recent studies suggest an etiologic role for visceral adipose tissue (4,5). The early identification of this condition and the primary prevention of overweight and obesity in Canada is an important and cost-effective public health priority (6). Lean et al. (7) first proposed the use of waist circumference (WC) as part of clinical cardiovascular risk assessments and interpretation of health risks associated with adiposity. In their sample, a body mass index (BMI) of 25 kg/m2 (overweight) and 30 kg/m2 (obese) corresponded with WC gender-specific action levels: “action level 1” (women, 80 cm; men, 94 cm) and “action level 2” (women, 88 cm; men, 102 cm), respectively. A random sample of men and women of the Netherlands MORGEN project were later used to test the sensitivity and specificity of the two action levels for identifying individual cardiovascular risk

1

Nonstandard abbreviations: HDL-C, high-density lipoprotein-cholesterol; WC, waist circumference; BMI, body mass index; NIH, National Institutes of Health; NCEP, National Cholesterol Education Program; ATP, Adult Treatment Panel; CHHS, Canada Heart Health Surveys; CHD, coronary heart disease; TG, triglyceride(s); LDL-C, low-density lipoproteincholesterol; BP, blood pressure; WHO, World Health Organization; OR, odds ratio; NHANES, National Health and Nutrition Examination Survey.

OBESITY RESEARCH Vol. 11 No. 1 January 2003

135

Use of BMI-WC Guidelines, Ardern et al.

factors (8). Action level 1 had a sensitivity of 57% to 72% for identifying individual cardiovascular risk factors, and this improved with action level 2, while decreasing false positive predictions (8). Recently the U.S. National Institutes of Health (NIH) released clinical guidelines advocating risk stratification based on measurement of WC (action level 2) within BMI categories as a means to identify those at increased cardiovascular health risk (9). It seems that the Expert Panel of the NIH adopted the higher cut-off based on the increased sensitivity and specificity for high total cholesterol, low HDL-C, and hypertension in the analysis of Han et al. (8). Only recently has evidence (10 –12) been presented to support dichotomizing WC levels in addition to BMI risk categories to improve the prediction of health risk. However, only one of these studies used the recommended WC action level in a population-based study (12); therefore, the use of these guidelines in the Canadian setting remains unclear. The purpose of this paper is 2-fold: 1) to estimate the prevalence and distribution of the metabolic syndrome in Canada and 2) to test the NIH BMI/WC guidelines for the identification of the metabolic syndrome and associated component risk factors according to the National Cholesterol Education Program (NCEP) Adult Treatment Panel (ATP) III risk categories (1).

Research Methods and Procedures Sample Data from the Canada Heart Health Surveys (CHHS), a nationally representative cross-sectional survey of coronary heart disease (CHD)-related knowledge and risk factors, were used. A detailed description of the sampling scheme is available elsewhere (13). Briefly, potential participants were identified within each province’s medical insurance registry and then stratified into six age-by-sex groups based on a complex sampling scheme. Participants were then randomly selected to approximate 2000 responses per province, assuming an 80% response rate (13). Trained registered nurses collected data between 1986 and 1992, beginning with an initial home visit in which a risk factor knowledge questionnaire was administered and two resting blood pressure readings were measured. Approximately 2 weeks later, a clinic visit was conducted to gather 8-hour fasting blood samples and two further blood pressure readings. For both the home visit and clinic blood pressure measurements, participants were asked not to eat or smoke for at least 30 minutes, and blood pressure measurements were taken using the first and fifth Korotkoff sounds (14) after the participant rested quietly for at least 5 minutes. Height, weight, and WC were also measured during the clinic visit. WC was measured in only five provinces; thus, data for this study were available for a total probability 136


sample of 7981 participants from Ontario, Manitoba, Alberta, Que´ bec, and Saskatchewan. In 1991, these five provinces accounted for over 79% of the total population of Canada (15). The CHHS group approved all protocols, and written informed consent was obtained from all participants. Anthropometry Weight (kilograms) and height (centimeters) were measured barefoot, and BMI (kilograms per square meter) was calculated. BMI measurements were truncated at 17.0 kg/m2 at the lower end and 42.0 kg/m2 at the upper end of the distribution, limiting ⬃0.5% of the distribution. WC was measured at the visible narrowing of the waist after a normal exhalation, or in extreme cases of obesity, at the level of the 12th rib. Metabolic Syndrome Eight-hour fasting plasma concentrations of total cholesterol, HDL-C, and triglycerides (TG) were assayed. Lowdensity lipoprotein-cholesterol (LDL-C) levels were calculated indirectly from total cholesterol levels using the Friedewald equation when TG ⬍4.51 mM (16). All blood lipid analyses were conducted at the Lipid Research Laboratory at the University of Toronto. Classification of the metabolic syndrome was based on the NCEP ATP III (1) blood lipid guidelines and is defined as the presence of any three of the following: central obesity (men, WC ⬎ 102 cm; women, WC ⬎ 88 cm); high TG (ⱖ1.69 mM); low HDL-C (men, ⬍1.04 mM; women, ⬍1.29 mM); high blood pressure (BP; systolic ⱖ 130 mm Hg or diastolic ⱖ 85 mm Hg); and high blood glucose (ⱖ6.1 mM) (1). In the present analysis, central obesity was omitted as a component of the metabolic syndrome in the prediction from NIH WC cut-offs (but not in the prevalence estimates), and self-report of physician diagnosed diabetes was used in place of a high blood glucose level. Statistical Analyses Sex- and age-group–specific prevalences of the metabolic syndrome were calculated using the total sample available (n ⫽ 7981). Participants were then grouped into NIH (9) and World Health Organization (WHO) (17) BMI categories (normal weight, 18.5 to 24.9 kg/m2; overweight, 25 to 29.9 kg/m2; obese class I, 30 to 34.9 kg/m2), which excluded some participants from further analysis (underweight and obese classes II and III; n ⫽ 606). Participants were also categorized according to the NIH WC recommendations (9) [men, low WC (ⱕ102 cm) and high WC (⬎102 cm); women, low WC (ⱕ88 cm) and high WC (⬎88 cm)]. The prevalences of metabolic and cardiovascular disease risk factors were first estimated across the NIH BMI and WC categories. Logistic regression was used to predict the metabolic syndrome and individual risk factors from WC by sex within the specific BMI categories of overweight and


Table 1. Weighted means and SDs for metabolic risk factors in the total sample of normal weight, overweight, and obese class I participants

Figure 1: Age-specific prevalences of the metabolic syndrome in the Canada Heart Health Surveys, 1986 through 1992. Error bars represent SE.

obese class I, as the prevalences of risk factors in normalweight participants were too low for meaningful analysis. This approach is reasonable because little additional power would be expected if WC was measured in a severely obese individual, or conversely, one who is underweight (9). Covariates in the analyses included age, alcohol use (0 ⫽ never drink; 1 ⫽ former drinker; 2 ⫽ current drinker), smoking status [0 ⫽ nonsmoker (never smoked or occasional smoker); 1 ⫽ current smoker (daily smoker)], and income adequacy, as calculated by family income and household size (0 ⫽ low; 1 ⫽ middle; 2 ⫽ high). All analyses were conducted using Stata Intercooled 7 (18) to take into account the complex sampling design of the CHHS, and sampling weights were applied to assure the representativeness of the results.

Results The overall prevalences of the metabolic syndrome in the total sample were 17.0% and 13.2%, in men and women, respectively. In general, there was an increasingly positive relationship between age category (20 to 29, 30 to 39, 40 to 49, 50 to 59, 60 to 69, and 70⫹ years) and the metabolic syndrome, with the exception of men in their 60s (Figure 1). There was a significant sex effect in the relationship between age and the prevalence of the metabolic syndrome (␹2 ⫽ 69.0, p ⬍ 0.0001), mainly because of the differences at older ages. When the sample was restricted to participants 20 to 59 years of age, the sex ⫻ age ⫻ metabolic syndrome interaction was no longer significant (␹2 ⫽ 3.5, p ⫽ 0.33). The clustering of high TG (92.5%), high BP (78.0%), and high HDL-C (75.7%) was most commonly associated with

Age (years) WC (cm) BMI (kg/m2) TG (mM) Systolic BP (mm Hg) Diastolic BP (mm Hg) HDL-C (mM) LDL-C (mM) Total C (mM)

Men (N ⴝ 3872)

Women (N ⴝ 3503)

41.9 ⫾ 17.9 90.9 ⫾ 12.9 25.6 ⫾ 4.22 1.63 ⫾ 1.05 126 ⫾ 17.7 79.4 ⫾ 11.0 1.20 ⫾ 0.36 3.20 ⫾ 1.09 5.13 ⫾ 1.22

42.6 ⫾ 18.4 78.3 ⫾ 13.1 24.5 ⫾ 4.94 1.31 ⫾ 0.85 119 ⫾ 21.3 74.7 ⫾ 11.2 1.46 ⫾ 0.45 3.02 ⫾ 1.17 5.07 ⫾ 1.35

the positive classification of the metabolic syndrome in men. In women, the metabolic syndrome was most commonly identified by the presence of low HDL-C (84.4%), high TG (79.2%), and abdominal obesity (76.9%). Table 1 presents the weighted descriptive statistics for normal-weight, overweight, and obese class I men and women. Table 2 presents the prevalence of selected cardiovascular disease risk factors across the U.S. NIH BMI and WC categories. The prevalence of metabolic risk factors increased across normal-weight, overweight, and obese class I categories in both men and women. Over 65% of obese men and almost 80% of obese women were classified as having a WC measurement above the threshold, whereas only 13% and 27% of overweight men and women, respectively, had high WC measurements. The prevalences of the metabolic syndrome and associated risk factors were higher in the high WC category compared with the low WC category (Table 2). Table 3 presents the results of the logistic regression models conducted within BMI categories (overweight and obese I). Here, normal-weight participants were excluded from the analysis because of sample size limitations; few normal-weight men and women were designated as having a high WC. The adjusted odds ratios (ORs) are from models that included age, smoking status, alcohol consumption, and income adequacy as covariates. For overweight and obese class I men, there was no observed benefit to the combined WC-BMI screening of health risks, after taking into account the covariates. Several ORs were significant in the unadjusted models, suggesting that the effects of WC are somewhat dependent on the measured covariates. The risk of the metabolic syndrome is nearly doubled in overweight women with a high WC compared with those with a low WC (OR, 1.85; 95%CI, 1.02 to 3.35). In obese I women, WC again adds to the prediction of the metabolic syndrome OBESITY RESEARCH Vol. 11 No. 1 January 2003

137


Table 2. Prevalence (mean ⫾ SE) of cardiovascular risk factors across NIH/WHO normal weight, overweight, and obese I categories Men (N ⴝ 3872) BMI category

Metabolic syndrome† Hypertension Diabetes Low HDL-C High LDL-C High TG High WC Current drinker Current smoker Sedentary lifestyle

Women (N ⴝ 3503) BMI category

Normal weight (N ⴝ 1656)

Overweight (N ⴝ 1749)

Obese I (N ⴝ 500)

Normal weight (N ⴝ 2112)

Overweight (N ⴝ 1035)

Obese I (N ⴝ 408)

5.6 ⫾ 1.2 14.7 ⫾ 2.2 2.5 ⫾ 0.4 23.3 ⫾ 1.3 10.1 ⫾ 1.0 8.2 ⫾ 1.0 0.2 ⫾ 0.1 90.3 ⫾ 1.7 28.4 ⫾ 2.0 36.2 ⫾ 2.0

18.6 ⫾ 1.2 27.8 ⫾ 2.1 5.5 ⫾ 2.1 38.6 ⫾ 1.1 15.2 ⫾ 1.2 26.0 ⫾ 1.5 12.9 ⫾ 0.9 91.7 ⫾ 2.1 19.0 ⫾ 0.9 41.5 ⫾ 3.0

49.0 ⫾ 6.3* 34.6 ⫾ 4.8* 6.9 ⫾ 1.5* 49.3 ⫾ 4.7* 19.8 ⫾ 4.1* 30.7 ⫾ 4.0* 65.3 ⫾ 3.1* 91.5 ⫾ 1.3* 31.2 ⫾ 9.2* 44.1 ⫾ 4.9*

2.9 ⫾ 0.3 9.6 ⫾ 0.6 2.7 ⫾ 0.4 6.8 ⫾ 1.2 9.8 ⫾ 1.1 5.7 ⫾ 0.6 1.9 ⫾ 0.8 85.9 ⫾ 1.0 26.3 ⫾ 1.1 29.4 ⫾ 3.5

17.6 ⫾ 3.9 24.8 ⫾ 2.5 4.0 ⫾ 0.8 10.1 ⫾ 1.5 15.7 ⫾ 3.2 14.0 ⫾ 2.3 26.5 ⫾ 1.7 82.6 ⫾ 5.0 18.3 ⫾ 2.7 31.7 ⫾ 6.6

46.5 ⫾ 3.9* 32.3 ⫾ 3.7* 6.1 ⫾ 1.1* 29.6 ⫾ 6.6* 19.8 ⫾ 4.3* 31.2 ⫾ 5.3* 78.1 ⫾ 4.6* 77.5 ⫾ 5.2* 21.5 ⫾ 1.9* 47.8 ⫾ 2.8*

WC category

Metabolic syndrome‡ Hypertension Diabetes Low HDL-C High LDL-C High TG Current drinker Current smoker Sedentary lifestyle

WC category

Low (N ⴝ 307)

High (N ⴝ 598)

Low (N ⴝ 2937)

High (N ⴝ 618)

9.5 ⫾ 0.5 19.8 ⫾ 1.5 3.5 ⫾ 0.9 31.3 ⫾ 1.0 12.7 ⫾ 0.7 16.6 ⫾ 0.9 91.1 ⫾ 1.7 24.4 ⫾ 1.6 38.2 ⫾ 1.3

21.6 ⫾ 5.5* 43.0 ⫾ 1.6* 9.7 ⫾ 2.7* 43.8 ⫾ 2.9* 18.9 ⫾ 3.0* 34.2 ⫾ 4.3* 91.2 ⫾ 1.3* 24.2 ⫾ 3.7* 49.3 ⫾ 8.8*

3.7 ⫾ 0.5 12.9 ⫾ 0.6 3.1 ⫾ 0.5 7.9 ⫾ 0.6 11.0 ⫾ 1.4 7.1 ⫾ 0.7 85.1 ⫾ 1.9 24.6 ⫾ 1.0 28.5 ⫾ 3.9

23.1 ⫾ 2.7* 32.0 ⫾ 5.5* 5.7 ⫾ 1.0* 21.6 ⫾ 3.2* 21.7 ⫾ 6.0* 29.0 ⫾ 3.3* 79.2 ⫾ 1.9* 19.3 ⫾ 1.8* 48.6 ⫾ 4.3*

␹2 test across BMI and WC categories (* p ⬍ 0.05). † Any three: high BP (ⱖ130 mm Hg systolic BP or ⱖ85 mm Hg diastolic BP); diabetes (physician diagnosed); low HDL (men: ⬍1.03 mM; women: ⬍1.29 mM); high TG (ⱖ1.69 mM); high WC (men: ⱖ102 cm; women: ⱖ88 cm); hypertension (systolic BP ⱖ 140 mm Hg or diastolic BP ⱖ 90 mm Hg or BP medication); high LDL (ⱖ4.13 mM); low HDL (⬍1.03 mM); high TG (ⱖ2.26 mM); self-reported current drinker; self-reported current smoker; self-reported sedentary lifestyle. ‡ Definition excludes high WC.

(OR, 2.35; 95%CI, 1.25 to 4.42) and low HDL-C levels (OR, 5.24; 95%CI, 2.08 to 13.20). Similar to the results in men, several ORs that were significant in the unadjusted models were not significant after the inclusion of the covariates.

Discussion Prevalence of the Metabolic Syndrome The results indicate that the prevalences of the metabolic syndrome in Canada (1986 through 1992) were 17.0% in men and 13.2% in women. These estimates are substantially lower than those recently reported for the third wave of the 138


National Health and Nutrition Examination Survey [NHANES III (1988 through 1994)], a nationally representative sample of the United States; using the same ATP III guidelines, Ford et al. (2) found the prevalences to be 24.0% and 23.4% in men and women, respectively. It is unlikely, however, that ethnic or racial differences could explain the differences in results because the prevalence of the metabolic syndrome in the white cohort of NHANES (23.8%) is not unlike that of the entire sample (2). The difference between metabolic syndrome estimates using NHANES III data (2) and the present analysis is largely because of differences in the prevalences of abdom-


Table 3. Prediction of the metabolic syndrome and associated risk factors from NIH waist circumference categories within NIH/WHO overweight and obese class I participants Overweight Unadjusted Metabolic parameter Men Metabolic syndrome High LDL-C Low HDL-C High total C High TG Hypertension Diabetes Women Metabolic syndrome High LDL-C Low HDL-C High total C High TG Hypertension Diabetes

Obese I Adjusted*

Unadjusted

Adjusted*

OR

95% CI

OR

95% CI

OR

95% CI

OR

95% CI

1.98 1.09 0.94 1.09 1.57 2.85 3.19

(0.73 to 5.35) (0.62 to 1.92) (0.40 to 2.19) (0.76 to 1.56) (0.50 to 4.95) (1.66 to 4.86)† (1.88 to 5.41)†

1.42 0.86 0.85 0.99 1.42 1.49 1.66

(0.65 to 3.10) (0.48 to 1.54) (0.41 to 1.79) (0.66 to 1.50) (0.50 to 4.09) (0.80 to 2.78) (0.85 to 3.24)

1.51 0.95 1.07 0.83 1.39 2.37 1.72

(1.02 to 2.24)† (0.57 to 1.58) (0.73 to 1.56) (0.42 to 1.63) (0.95 to 2.05) (1.47 to 3.82)† (0.91 to 3.26)

0.88 0.69 1.19 0.70 1.42 1.62 0.49

(0.46 to 1.69) (0.35 to 1.36) (0.27 to 5.18) (0.35 to 1.41) (0.51 to 3.94) (0.71 to 3.72) (0.13 to 1.84)

2.75 1.80 0.72 2.46 3.26 1.86 1.20

(1.71 to 4.44)† (1.08 to 3.26)† (0.38 to 1.36) (1.01 to 6.01)† (1.16 to 9.18)† (0.79 to 4.39) (0.55 to 2.64)

1.85 1.10 0.91 1.77 2.80 1.08 0.99

(1.02 to 3.35)† (0.63 to 1.91) (0.54 to 1.51) (0.61 to 5.13) (0.99 to 7.91) (0.40 to 2.96) (0.41 to 2.36)

2.50 1.96 4.00 0.86 2.37 1.21 1.88

(1.36 to 4.61)† (0.47 to 8.19) (1.44 to 11.1)† (0.26 to 2.85) (1.12 to 5.04)† (0.47 to 3.14) (0.59 to 5.95)

2.35 0.87 5.24 0.22 2.55 0.57 2.28

(1.25 to 4.42)† (0.22 to 3.49) (2.08 to 13.2)† (0.07 to 0.71) (0.97 to 6.71) (0.20 to 1.60) (0.59 to 8.75)

* Logistic regression adjusted for age, smoking status, alcohol consumption, and income adequacy. ORs indicate “high WC” groups relative to the “low WC” reference group (OR ⫽ 1.00) within each overweight and obese I BMI category. † p ⬍ 0.05.

inal obesity (high WC) and high blood glucose/diabetes between the two surveys. The prevalences in NHANES III vs. CHHS were quite similar for high BP (34.0% vs. 34.1%), high TG (30.0% vs. 30.2%), and low HDL-C (37.1% vs. 34.8%); however, the prevalence of high blood glucose/diabetes (12.6% vs. 4.3%) was almost triple, and the prevalence of abdominal obesity was more than double in the NHANES III vs. CHHS (38.6% vs. 16.5%). In the current study, physician-diagnosed diabetes was used in lieu of measured blood glucose, which may underestimate the prevalence of diabetes by as much as 50% (19). This suggests that the use of self-reported diabetes in the CHHS may be a conservative estimate of the actual prevalence of high blood glucose levels, thus accounting for some of the difference between the U.S. NHANES and CHHS findings. Part of the difference in the prevalences of high WC may be because of differences in the protocol for measuring WC between NHANES III and the CHHS; however, it is doubtful that it explains the difference entirely. The prevalences of abdominal obesity obtained in the present study, 14.4% and 18.8% for men and women, respectively, seem reasonable given that the WC cut-off corresponds to a BMI of 30 kg/m2 (7) and the prevalence of

obesity in the CHHS based on a BMI ⱖ30 kg/m2 was 13% in men and 14% in women (20). When a regression analysis was conducted on the present CHHS sample, WC cut-offs associated with overweight (women, 80 cm; men, 90 cm) and obese I BMI categories (women, 90 cm; men, 102 cm) were very similar to those proposed by Lean et al. (7). Furthermore, the correlation between BMI and WC in the present sample was 0.79, so it is not surprising to obtain similar prevalence estimates when using the BMI and WC cut-offs. Prediction of Risk Factors from WC The standardized use of anthropometric measurements has recently been advised for the prediction of health risks in primary care and private practice (21). Because of the difficulty in calculating and explaining BMI scores to patients, WC has been recommended as a single measure of cardiovascular and metabolic health risk (7,22). However, the U.S. NIH has recommended a two-tiered classification system using both BMI and WC (9). We are aware of only three analyses that have investigated the use of combined BMI-WC classifications for risk stratification (10 –12). OBESITY RESEARCH Vol. 11 No. 1 January 2003

139


Most recently, Janssen et al. (12) assessed the use of the two-tiered NIH BMI/WC stratification for the prediction of the metabolic syndrome and associated risk factors using the NHANES III dataset. The results indicated that use of the WC categories within both overweight men and women added to the prediction of several risk factors and the metabolic syndrome. Among obese participants, WC was predictive of most risk factors and the metabolic syndrome in women, but only of the metabolic syndrome in men. The results for men are not consistent with those of the present study, although the results for women are similar. Potential reasons for the difference in predictive capacity between Canadian and American men may be complicated, only one factor of which may include differences in the measurement protocol for WC. Even within the United States there are inconsistencies. For example, technicians in NHANES III measured WC midway between the iliac crest and the 12th rib; however, 4 years later, in 1998, the Practical Guide for the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults indicates that WC should be measured at the level of the iliac crest (9). Another potential explanation may again be the reported differences in the prevalence of obesity between Canada (men, 15.4%; women, 14.4%; 1998 National Population Health Survey) (23) and the United States (men, 17.7%; women, 18.1%; 1998 Behavioral Risk Factor Surveillance System) (24). Iwao et al. (11) examined the influence of age on the prediction of CHD risk factors from WC and BMI in a self-selected sample of upper-middle-class men and women from the Baltimore Longitudinal Study of Aging. Logistic regression, corrected for BMI classification [normal weight (BMI ⬍ 25 kg/m2) vs. overweight (BMI ⱖ 25 kg/m2)], was used to test the ability of action level 1 (men, 94 cm; women, 80 cm) categories to predict individual CHD risk factors as defined by ATP II criteria. WC substantially improved the prediction of most individual risk factors beyond that explained by BMI in both men and women under 65 years of age, but provided little improvement in older subjects. Another analysis using data from the Iowa Women’s Health Study (55 to 69 years old) indicated that WC stratification within normal weight, overweight, and obese I was a better predictor of diabetes, hypertension, and cardiovascular-related mortality than BMI alone in older women (10). It is likely that any relationship between WC, BMI, and the metabolic syndrome will vary according to age, gender, and ethnicity. In this sample, limited data on hormone replacement therapies were available, limiting the potential for investigation into the influence of fat redistribution and hormonal status on the predictive capacity of WC and BMI on metabolic parameters. Metabolic risk factors that were most likely to be responsible for a positive metabolic syndrome classification may also explain some of the observed 140


differences in the combined predictive ability of BMI and WC in men (high TG; high BP; low HDL-C) and women (low HDL-C; high TG; high WC). It has also been suggested that individual WC and BMI thresholds for health risk are influenced by the prevalence of risk factors within a population (22). When action level 1 of Lean et al. (7) was tested against BMI cut-offs in 19 populations of the WHO MONICA study, the sensitivity and specificity varied considerably, with WC generally performing the best in those populations with a high prevalence of obesity (25). Although we were unable to address the influence of ethnicity in the current study, it is clear that there is a strong effect of age on the prediction of the metabolic syndrome. When age was eliminated from the logistic regression models presented in Table 3, the ORs for WC categories were inflated, and diabetes was then predicted by the additional measurement of WC in overweight (OR, 3.12; 95%CI, 1.82 to 5.34), but not obese, men (results not shown). When participants were stratified into younger (⬍55 years) and older (ⱖ55 years) age categories, there were differences in the strength of the associations among younger and older women; however, the results were not significant in men in either age group. The ORs for the metabolic syndrome and high TG levels were higher in younger (metabolic syndrome: OR, 8.27; 95%CI, 2.27 to 30.1; high TG: OR, 4.68; 95%CI, 1.59 to 13.8) than older (metabolic syndrome: OR, 0.86; 95%CI, 0.37 to 2.01; high TG: OR, 2.05; 95%CI, 0.43 to 9.82) overweight women. In obese women, the ORs for the metabolic syndrome (younger: OR, 2.94; 95%CI, 1.17 to 7.42; older: OR, 0.78; 95%CI, 0.13 to 4.82), low HDL-C (younger: OR, 7.16; 95%CI, 1.84 to 27.9; older: OR, 0.57; 95%CI, 0.09 to 3.60), and high TG (younger: OR, 3.79; 95%CI, 1.14 to 12.6; older: OR, 0.76; 95%CI, 0.06 to 9.48) were also higher in younger vs. older women. Based on analyses from the CHHS, Dobbelstyn et al. (22) have recommended adjusting the WC cut-off according to 10-year age groups and the specific CHD risk factor(s) of interest, within a 10-cm range beginning at a baseline level of 90 cm in men and 80 cm in women. Corrections could then also be made for different risk factors of interest to improve specificity. However, these adjustments would be very complicated in clinical practice, and the analyses were conducted on a predominately white sample, suggesting again that these recommendations be interpreted with caution, particularly because the combined influence of BMI and WC was not considered. The lack of an improvement in the prediction of diabetes with the combined measurement of WC and BMI in the present study is supported by results of a biracial cohort (26), indicating no improvement in sensitivity beyond WC or BMI alone. On the other hand, there is evidence that central obesity is a significant predictor of the development of diabetes beyond that explained by obesity alone (27,28). Because fasting blood glucose samples were not collected in


the CHHS, we relied on a self-report measure of physiciandiagnosed diabetes to define the presence of high blood glucose for the metabolic syndrome, allowing for a stricter criterion for blood glucose concentration [7.0 mM for diabetes (29) but 6.1 mM as an ATP III risk factor]. Use of self-report measures may underrepresent diabetes by as much as 50% (19). A strength of this study is that it was conducted on a population-based, random selection of Canadians from five different provinces. However, the cross-sectional nature of the design does not allow for cause-effect conclusions to be made, and the stage of disease progression or comorbidities are not known (30). The analyses also use NCEP ATP III guidelines specifically designed for the identification of the metabolic syndrome, which places the results within a clinical context. Based on the current analyses, the fact that the additional screening of WC predicts health risk in women but not men suggests that current WC cut-offs may be more appropriate in women and that they should continue to be examined in men. However, this does not suggest that these BMI and WC combinations are the optimal cut-offs for identification of the metabolic syndrome in women. No research to date has performed analyses to determine the appropriate cutoffs for WC when stratified by BMI category, but if the combined screening of increased health risk by WC and BMI is to be recommended, this should be a priority for future research.

Acknowledgments This research was supported by the Heart and Stroke Foundation of Ontario (Grant T 4946). Dr. I. Janssen is also supported by a Canadian Institute of Health Research Postdoctoral Fellowship. The analyses are based on the Canadian Heart Health Database 1986 through 1992, which contains anonymous data collected in a coordinated series of heart health surveys carried out in the 10 provinces of Canada between 1986 and 1992. The database was constructed by the Conference of Primary Investigators of provincial heart health programs from survey questions and clinical measures that were common to all surveys. All computations on these microdata were prepared by C. Ardern and P. Katzmarzyk of Queen’s University, and the responsibility for the use and interpretation of these data is entirely that of the authors. References 1. The Expert Panel. Executive Summary of The Third Report of The National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol In Adults (Adult Treatment Panel III). JAMA 2001;285:2486 –97. 2. Ford ES, Giles WH, Dietz WH. Prevalence of the metabolic syndrome among U.S. adults: findings from the Third National Health and Nutrition Examination Survey JAMA. 2002; 287:356 –9.

3. Lemieux S. Contribution of visceral obesity to the insulin resistance syndrome. Can J Appl Physiol. 2001;26:273–90. 4. Masuzaki H, Paterson J, Shinyama H, et al. A transgenic model of visceral obesity and the metabolic syndrome. Science. 2001;294:2166 –70. 5. Pascot A, Despre´s JP, Lemieux I, et al. Deterioration of the metabolic risk profile in women. Respective contributions of impaired glucose tolerance and visceral fat accumulation. Diabetes Care. 2001;24:902– 8. 6. Ramos F, Baglivo HP, Ramirez AJ, Sanchez R. The metabolic syndrome and related cardiovascular risk. Curr Hypertens Rep. 2001;3:100 – 6. 7. Lean ME, Han TS, Morrison CE. Waist circumference as a measure for indicating need for weight management. BMJ. 1995;311:158 – 64. 8. Han T, Leer E, Seidell J, Lean M. Waist circumference action levels in the identification of cardiovascular risk factors: prevalence study in a random sample. BMJ. 1995;311: 1401–5. 9. National Institutes of Health. The Practical Guide to the Identification, Evaluation and Treatment of Overweight and Obesity in Adults. Bethesda, MD: U.S. National Institutes of Health; 2000. 10. Folsom AR, Kushi LH, Anderson KE, et al. Associations of general and abdominal obesity with multiple health outcomes in older women: the Iowa Women’s Health Study. Arch Intern Med. 2000;160:2117–28. 11. Iwao S, Iwao N, Muller DC, Elahi D, Shimokata H, Andres R. Does waist circumference add to the predictive power of the body mass index for coronary risk? Obes Res. 2001;9: 685–95. 12. Janssen I, Katzmarzyk PT, Ross R. Body mass index, waist circumference, and health risk: evidence in support of current NIH guidelines. Arch Intern Med 2002;162:2074 –9. 13. MacLean DR, Petrasovits A, Nargundkar M, et al. Canadian Heart Health Surveys: a profile of cardiovascular risk. Survey methods and data analysis. CMAJ. 1992;146:1969 –74. 14. Joffres MR, Hamet P, Rabkin SW, Gelskey D, Hogan K, Fodor G. Prevalence, control and awareness of high blood pressure among Canadian adults. CMAJ. 1992;146: 1997–2005. 15. Statistics Canada. Catalogue no. 82F0075XCB Ed. Statistics Canada. Ottawa, Canada; 2001. 16. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem. 1972;18:499 –502. 17. World Health Organization. Obesity: preventing and managing the global epidemic. Report of a WHO consultation. Geneva: World Health Organization; 1998. 18. StataCorp. Stata Statistical Software: Release 7.0. College Station, TX: Stata Corp.; 2001. 19. Harris MI, Flegal KM, Cowie CC, et al. Prevalence of diabetes, impaired fasting glucose, and impaired glucose tolerance in U.S. adults. The Third National Health and Nutrition Examination Survey, 1988 –1994. Diabetes Care. 1998;21: 518 –24. OBESITY RESEARCH Vol. 11 No. 1 January 2003

141


20. Macdonald SM, Reeder BA, Chen Y, De´ spres JP. Obesity in Canada: a descriptive analysis. CMAJ. 1997;157(Suppl 1):S3–S9. 21. Seidell JC, Kahn HS, Williamson DF, Lissner L, Valdez R. Report from a Centers for Disease Control and Prevention Workshop on use of adult anthropometry for public health and primary health care. Am J Clin Nutr. 2001;73:123– 6. 22. Dobbelsteyn CJ, Joffres MR, MacLean DR, Flowerdew G. A comparative evaluation of waist circumference, waist-to-hip ratio and body mass index as indicators of cardiovascular risk factors. Int J Obes Relat Metab Disord. 2001;25:652– 61. 23. Katzmarzyk PT. The Canadian obesity epidemic: an historical perspective. Obes Res 2002;10:666 –74. 24. Mokdad AH, Bowman BA, Ford ES, Vinicor F, Marks JS, Kaplan JP. The continuing epidemics of obesity and diabetes in the United States. JAMA. 2001;286:1195–200. 25. Molarius A, Seidell JC, Sans S, Tuomilehto J, Kuulasmaa K. Varying sensitivity of waist action levels to identify sub-

142


26.

27.

28.

29.

30.

jects with overweight or obesity in 19 populations of the WHO MONICA Project. J Clin Epidemiol. 1999;52:1213–24. Stevens J, Couper D, Pankow J, et al. Sensitivity and specificity of anthropometrics for the prediction of diabetes in a biracial cohort. Obes Res. 2001;9:696 –705. Chan JM, Rimm EB, Colditz GA, Stampfer MJ, Willett WC. Obesity, fat distribution, and weight gain as risk factors for clinical diabetes in men. Diabetes Care. 1994;17:961–9. Hartz AJ, Rupley DC Jr, Kalkhoff RD, Rimm AA. Relationship of obesity to diabetes: influence of obesity level and body fat distribution. Prev Med. 1983;12:351–7. Alberti KGMM, Zimmet PZ. Definition, diagnosis and classification of diabetes mellitus and its complications. Part I: diagnosis and classification of diabetes mellitus. Provisional report of a WHO consultation. Diabet Med. 1998;15:539-53. Molarius A, Seidell JC. Selection of anthropometric indicators for classification of abdominal fatness—a critical review. Int J Obes Relat Metab Disord. 1998;22:719 –27.