Despite its well-documented relation with visceral adiposity (VAT) and cardiometabolic risk (CMR), whether waist circumference (WC) should be measured in addition to body mass index (BMI) remains debated. This study tested the relevance of adding WC to BMI for the estimation of VAT and CMR. In the International Study of Prediction of Intra-abdominal Adiposity and Its Relationship with Cardiometabolic Risk/Intra-abdominal Adiposity, 297 physicians recruited 4,504 patients (29 countries). Both BMI and WC were measured, whereas VAT and liver fat were assessed by computed tomography. A composite CMR score was calculated. From the 4,109 patients included in the present analyses (20 ≤ BMI < 40 kg/m 2 , 47% women), about 30% displayed discordant values for WC and BMI quintiles, despite a strong correlation between the 2 anthropometric variables (r = 0.87 and r = 0.84 for men and women, respectively, p <0.001). Within each single BMI unit, VAT and WC showed substantial variability between subjects (mean difference between 90th and 10th percentiles: 175 cm 2 /16 cm and 137 cm 2 /18 cm for VAT/WC in men and women, respectively). Within each BMI category, increasing gender-specific WC tertiles were associated with significantly higher VAT, liver fat, and with a more adverse CMR profile. In conclusion, this large international cardiometabolic study highlights the frequent discordance between BMI and WC, driven by the substantial variability in VAT for a given BMI. Within each BMI category, WC was cross-sectionally associated with VAT, liver fat, and CMR factors. Thus, WC allows a further refinement of the CMR related to any given BMI.
Abdominal adiposity is recognized to be associated with a cluster of diabetogenic and atherogenic abnormalities associated with an increase in cardiometabolic risk (CMR) and all-cause morbidity/mortality risk, highlighting the importance of specifically considering abdominal adiposity in patients’ overall risk assessment. Although numerous epidemiologic studies have shown that anthropometric indices of abdominal adiposity/body fat distribution, such as waist circumference (WC) and waist-to-hip ratio, are better variables than the body mass index (BMI) to estimate abdominal adiposity, body fat distribution, and CMR, some investigators have questioned the clinical relevance of measuring WC instead of BMI to assess cardiovascular disease (CVD) risk, notably because BMI and WC are highly correlated and both independently predict CMR. The International Study of Prediction of Intra-abdominal Adiposity and Its Relationship with Cardiometabolic Risk/Intra-abdominal Adiposity (INSPIRE ME IAA) is an international study with a planned 3-year prospective evaluation, designed to explore the relations between visceral adiposity (VAT) and liver fat, assessed by computed tomography, CMR profile, and risk of CVD or type 2 diabetes mellitus. The present analyses explore the relevance of measuring WC in addition to BMI in the estimation of VAT and related CMR. We hypothesized that at any specific BMI value, a higher WC would be predictive of higher VAT and related CMR.
Methods
Study patients were recruited from June 2006 to May 2008, by 297 physicians (hospital-based primary care physicians/internists [27%], cardiologists [38%], and endocrinologists/diabetologists [35%]) from 29 countries in Asia, Europe, Latin, and North America ( Supplementary Table 1 ). Recruitment methods and exclusion criteria are described in the online supplement . All patients signed a written informed consent. The present study complies with the Declaration of Helsinki and each participating site received local ethics committee approval.
Age, gender, menopausal status, ethnicity, level of education, smoking status, personal and familial medical history of CVD, diabetes, dyslipidemia, and hypertension, and concomitant medications were recorded using a physician-administered questionnaire. Definitions of type 2 diabetes mellitus and CVD are given in the online supplement . Only subjects with a BMI between 20 and 40 kg/m 2 were included in the present analyses.
WC was measured using a metric tape according to a standardized protocol: patients stood with their feet, shoulder-width apart and arms crossed over the chest, naked upper body, and breathing normally; WC was measured midway between the lowest rib and the iliac crest at the end of the patient’s normal expiration. Weight and height were measured according to standardized methods, and BMI was calculated as weight per height squared. Seated blood pressure (3 times, third measurement recorded) and heart rate (measured over 1 minute) were recorded after 3 minutes of rest. Cross-sectional areas of abdominal visceral and subcutaneous adipose tissues from a computed tomography scan performed at L4-L5 were considered as markers of abdominal adiposity. Another computed tomography scan was performed at T12-L1 level, and the attenuation of the liver expressed in Hounsfield units was used as an estimate of liver fat. The personnel in all imaging centers (58 sites) were trained, and the procedures in all centers were calibrated using a special, engineer-designed phantom filled with lard to ensure that image acquisitions followed a standardized protocol. Use of the calibration phantom ensured consistency of the image acquisition procedures across all sites, thus reducing potential site-related bias. Abdominal and liver images were analyzed centrally in a core laboratory (Center de recherche de l’Institut universitaire de pneumologie et de cardiologie de Québec, Québec, Québec, Canada) with the Slice-O-matic software (Tomovision, Montreal, Canada).
Plasma fasting glucose; insulin; triglycerides; total, high-density lipoprotein, and low-density lipoprotein cholesterol; apolipoprotein B; apolipoprotein A1; high-sensitive C-reactive protein; adiponectin; and fibrinogen were assessed in 1 central laboratory that was accredited by the College of American Pathologists’ Laboratory Accreditation Program (MDS Pharma Services, Paris, France) using standardized methodologies as described in the online supplement .
A global CMR score was calculated. A subscore of 0 (lower risk level) or 1 (higher risk level) was set for each risk factor (CVD, low liver attenuation, elevated triglycerides, low high-density lipoprotein cholesterol, elevated apolipoprotein B, elevated blood pressure, elevated fasting glycemia, and elevated high-sensitive C-reactive protein) according to gender-specific thresholds (details provided in the online supplement ), and the global CMR score was calculated for patients who had no missing data (1,630 men and 1,446 women), as the sum of the 8 individual risk factor subscores.
Patients’ characteristics are presented by gender as mean ± SD or percentages. Variables were log-transformed for statistical analyses when the distribution was skewed. Differences between men and women for continuous variables were assessed using analysis of variance (adjusting for age, ethnicity, and physician’s specialty). Further statistical analyses included additional adjustment for smoking status and education level. First, within each gender, the distribution of VAT and WC for every single BMI unit from 20 to 39 kg/m 2 is presented using a box-and-whisker plot (median, 10th, 25th, 75th, and 90th percentiles). Second, quintiles of WC and BMI were calculated to assess the concordance between BMI and WC, defined as the percentage of patients for whom the WC fell within the same quintile group as BMI. Third, patients were divided by gender-specific tertiles of WC within each BMI category (lean [20 ≤ BMI < 25 kg/m 2 ], overweight [25 ≤ BMI < 30 kg/m 2 ], and obese [30 ≤ BMI < 40 kg/m 2 ]) and within each BMI unit. The relations between parameters (VAT, WC, BMI, and CMR score) were assessed within each gender using a multivariable linear regression analysis, and Pearson’s partial correlation coefficients are reported. An analysis of covariance followed by a post hoc test was also performed to detect differences in VAT, liver attenuation, and CMR score between WC tertile groups within each BMI category or unit. To determine whether the addition of WC would improve the model explaining variability in VAT and in the CMR score when both BMI and WC are used as continuous variables, a linear regression analysis was performed including VAT or CMR score as dependant variable and both BMI and residuals of WC adjusted for BMI, to avoid collinearity, as independent variables. A logistic model was used to calculate gender-specific odds ratios for type 2 diabetes, CVD, and each risk factor in the CMR score between the third and the first WC tertile groups in each BMI category. Results were considered significant at the p <0.05 level. All statistical analyses were performed using the SAS Statistical Package, version 9.2 (SAS Institute Inc., Cary, North Carolina).
Results
From the 4,504 patients initially recruited for the INSPIRE ME IAA study, 4,109 people (2,174 men and 1,935 women) were included in the present analyses (exclusion of patients with high-sensitive C-reactive protein level <10 mg/ml, with BMI <20 kg/m 2 or ≥40 kg/m 2 ). Characteristics of the study population are given in Table 1 . The population included 38% from Asia, 25% from Europe, 21% from Latin America, and 16% from North America. Regarding medical treatment, 36.3% of men were treated for diabetes, 53.2% for dyslipidemia, and 67.4% for hypertension. For women, these frequencies were 32.0%, 41.4%, and 60.3%, respectively.
| Variable | Men (n=2174) | Women (n=1935) |
|---|---|---|
| Overweight (25≤BMI<30 kg/m 2 ) | 44.6 % | 37.2 % |
| Obesity (30≤BMI<40 kg/m 2 ) | 31.4 % | 34.1 % |
| Metabolic syndrome | 62.3 % | 62.1 % |
| Cardiovascular diseases | 32.0 % | 15.5 % |
| Type 2 diabetes mellitus | 49.5 % | 43.1 % |
| Current smokers | 20.0 % | 8.8 % |
| Age (years) | 56.4 ± 8.1 | 57.1 ± 6.7 |
| Body Mass Index, BMI (kg/m 2 ) | 28.2 ± 4.2 | 28.2 ± 4.8 |
| Waist circumference, WC (cm) | 100 ± 12 | 92 ± 13 |
| Visceral Adipose Tissue, VAT (cm 2 ) | 182 ± 82 | 149 ± 63 |
| Systolic blood pressure (mm Hg) | 131 ± 17 | 130 ± 19 |
| Diastolic blood pressure (mm Hg) | 80 ± 11 | 78 ± 11 |
| Heart Rate (beats/min) | 71 ± 11 | 73 ± 10 |
| Triglycerides (mg/dL) | 69.6 ± 50.3 | 62.6 ± 46.0 |
| Total cholesterol (mg/dL) | 178.7 ± 40.2 | 197.2 ± 40.2 |
| Low-density lipoprotein (LDL) cholesterol (mg/dl) | 105.6 ± 34.4 | 117.6 ± 34.8 |
| High-density lipoprotein (HDL) cholesterol (mg/dl) | 45.2 ± 12.8 | 53.8 ± 15.1 |
| Apolipoprotein B (g/L) | 0.86 ± 0.24 | 0.89 ± 0.24 |
| Apolipoprotein A1 (g/L) | 1.31 ± 0.24 | 1.48 ± 0.27 |
| Total cholesterol/HDL cholesterol | 4.19 ± 1.46 | 3.93 ± 1.39 |
| Fasting glycaemia (mmol/L) | 6.7 ± 2.2 | 6.3 ± 2.1 |
| Fasting insulin (pmol/L) | 89 ± 104 | 78 ± 86 |
| High-sensitive C-reactive protein (mg/L) | 1.9 ± 1.8 | 2.4 ± 2.2 |
| Fibrinogen (g/L) | 3.2 ± 0.9 | 3.5 ± 0.9 |
| Adiponectin (μg/mL) | 5.3 ± 5.1 | 8.0 ± 5.8 |
a Data are given as percentages for frequencies or as mean ± standard deviation for other variables.
In both genders, BMI showed a strong correlation with WC ( Figure 1 , p <0.0001). Both BMI and WC were significantly correlated with VAT (r = 0.75 and r = 0.68 in men and r = 0.72 and r = 0.65 in women, respectively, p <0.001). Figure 2 shows the distribution of VAT and WC per unit of BMI; the mean difference between the 90th and the 10th percentiles of VAT within each single BMI unit ranged from 94 to 302 cm 2 (mean 175 cm 2 ) and from 65 to 249 cm 2 (mean 137 cm 2 ) for VAT and from 12 to 33 cm (mean 16 cm) and from 14 to 23 cm (mean 18 cm) for WC in men and women, respectively.
Figure 3 shows the quintile distribution of WC for each quintile group of BMI by gender. If BMI and WC were concordant for all patients, they would all be in the same-rank quintile group for both parameters and 100% of the patients would be on the 45° diagonal of the WC-BMI quintiles matrix. In the lowest and highest BMI quintile groups, around 70% of the patients were concordant for BMI and WC, where <50% were concordant in the intermediate quintile groups.
In both genders, VAT increased significantly across the WC tertiles, within each of the 3 BMI categories or unit, as illustrated in Figures 4 and 5 . Liver attenuation (inversely correlated to liver fat content) also decreased significantly across WC tertile groups in both genders. The CMR score, reflecting the number of cardiometabolic abnormalities, was significantly correlated to VAT, WC, and BMI in men and women ( Supplementary Table 2 ). The comparison of the CMR scores by WC tertile groups in each BMI category is listed in Table 2 . In both gender in all 3 categories of BMI, the CMR score increased across WC tertile groups, with a significantly higher CMR score in the upper tertiles in comparison with the lowest. In obese men and women, the CMR score was further increased in the third WC tertile group compared with the middle one.
| Men | WC tertiles | |||||
|---|---|---|---|---|---|---|
| T1 | T2 | T3 | ||||
| BMI < 25 kg/m 2 | WC ≤ 84 cm | 2.1 ± 0.1 | 84 < WC ≤ 90 cm | 2.5 ± 0.1** | WC > 90 cm | 2.7 ± 0.1*** |
| 25 kg/m 2 ≤ BMI < 30 kg/m 2 | WC ≤ 95 cm | 2.7 ± 0.1 | 95 < WC ≤ 101 cm | 3.3 ± 0.1** | WC > 101 cm | 3.6 ± 0.1*** |
| BMI ≥ 30 kg/m 2 | WC ≤ 108 cm | 3.7 ± 0.1 | 108 < WC ≤ 116 cm | 4.1 ± 0.1* | WC > 116 cm | 4.5 ± 0.1 **† |
| Women | WC tertiles | |||||
|---|---|---|---|---|---|---|
| T1 | T2 | T3 | ||||
| BMI < 25 kg/m 2 | WC ≤ 76 cm | 1.5 ± 0.1 | 76 < WC ≤ 83 cm | 1.9 ± 0.1** | WC > 83 cm | 2.7 ± 0.1***††† |
| 25 kg/m 2 ≤ BMI < 30 kg/m 2 | WC ≤ 87 cm | 2.5 ± 0.1 | 87 < WC ≤ 93 cm | 3.3 ± 0.1*** | WC > 93 cm | 3.8 ± 0.1***†† |
| BMI ≥ 30 kg/m 2 | WC ≤ 100 cm | 3.4 ± 0.1 | 100 < WC ≤ 108 cm | 3.8 ± 0.1 | WC > 108 cm | 4.6 ± 0.1**†† |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
