Author + information
- Received September 24, 2012
- Revision received December 24, 2012
- Accepted January 8, 2013
- Published online August 20, 2013.
- Michael Khoury, MD,
- Cedric Manlhiot, BSc and
- Brian W. McCrindle, MD, MPH⁎ ()
Reprint requests and correspondence:
Dr. Brian W. McCrindle (guarantor), The Hospital for Sick Children, 555 University Avenue, Toronto, Ontario M5G 1X8, Canada
Presented at the American Heart Association Scientific Sessions, November 12 to 14, 2011, Orlando Florida.
Objectives The aim of this study was to determine the utility of waist/height ratio (WHtR) in the specification of cardiometabolic risk in children already stratified by body mass index (BMI).
Background Reflective of its association with cardiometabolic risk, BMI is a commonly used indirect indicator of adiposity in children. The WHtR, a marker of central adiposity, has been advocated as a possibly superior indicator of cardiometabolic risk.
Methods Cross-sectional analysis of 5 National Health and Nutrition Examination Surveys from 1999 to 2008 (ages 5 to 18 years of age). The BMI percentile categories (normal, overweight, and obese) were further stratified on the basis of WHtR (<0.5, 0.5 to <0.6, ≥0.6). Outcome measures were lipid and glycemic profiles, C-reactive protein, liver transaminases, prevalence of hypertension, and metabolic syndrome.
Results Data were available for 14,493 subjects. Overweight and obese subjects with a WHtR <0.5 had a cardiometabolic risk approaching that of subjects with a normal BMI percentile category. Increasing WHtR was significantly associated with increased cardiometabolic risk in overweight and obese subjects, with the greatest associations observed in the obese population. Of obese subjects with WHtR ≥0.6, 26% had elevated non–high-density lipoprotein levels, 18% had elevated C-reactive protein levels, 69% had an elevated homeostatic model assessment–insulin resistance, and 32% had metabolic syndrome.
Conclusions The WHtR further specifies cardiometabolic risk within classifications stratification on the basis of BMI percentile. A significant proportion of obese children with increased WHtRs have abnormal cardiometabolic risk factor levels. The WHtR should be included in the routine screening and assessment of overweight and obese children, and those with an elevated WHtR should undergo a further cardiometabolic risk assessment.
Childhood overweight and obesity is associated with an increased prevalence of numerous cardiometabolic risk factors, including hypertension, hypercholesterolemia, and type 2 diabetes (1–3). The atherosclerotic process begins in childhood, and cardiometabolic risk in these early years tracks into adulthood, potentially resulting in early cardiovascular disease events (4,5). Over recent decades, a general trend of increased cardiometabolic risk factors abnormalities among children has been observed (6). Body mass index (BMI) (kilograms/square millimeters) is a commonly used surrogate measure of adiposity in children. The International Obesity Task Force (7), the U.S. Centers for Disease Control and Prevention (8), and the World Health Organization (9) have formulated population-based age- and sex-specific normal distributions for BMI. It is an established correlate of numerous adiposity-related cardiometabolic risk factors in the pediatric population (10,11), and numerous committees and organizations have endorsed its use (7,8,12,13). However, BMI by definition cannot distinguish between fat and fat-free mass. Therefore, an elevated BMI might not necessarily reflect increased adiposity (14–16).
Excessive central adiposity has been shown to be particularly pathogenic through the development of adiposopathy (17). The waist/height ratio (WHtR) (waist circumference divided by height, both measured in centimeters) has been advocated as an effective and convenient measure of central adiposity that could potentially be superior to BMI alone in determining cardiometabolic risk (18–22). However, studies assessing the use of WHtR with BMI have been limited, specifically by subject populations that are not diverse and the lack of assessment of inflammatory markers to further explore the impact of adiposopathy in central adiposity. Therefore, through the use of a large and diverse subject population and the incorporation of inflammatory markers, we sought to determine the clinical utility of the WHtR in further specifying the cardiometabolic risk assessment of children already categorized by BMI.
National Health and Nutrition Examination Survey
Data for this study were obtained from 5 National Health and Nutrition Examination Surveys (NHANES) between the years 1999 and 2008 (representing 5 2-year cycles) (23) for subjects 5 to 18 years of age. The NHANES surveys are complex multi-stage area probability samples of the U.S. non-institutionalized population, designed to be representative of the general U.S. population. It is conducted by the National Center for Health Statistics of the Centers for Disease Control. To produce reliable statistics, the survey oversamples certain populations, including African Americans and Hispanics. The survey includes a physical examination and blood sampling for various laboratory measures. Blood samples are taken from all but the very young. Fasting, although encouraged, was not mandatory for children. The NHANES 1999 to 2008 surveys underwent ethics approval by the National Center for Health Statistics Institutional Review Board and Research Ethics Board. Informed consent was obtained directly from participants 12 years of age and older, whereas written assent was obtained for children 7 to 11 years of age. Written parental consent was obtained for all children <18 years of age.
Standing height was measured with a stadiometer with a fixed vertical backboard and an adjustable headpiece. Weight was obtained with a digital weight scale with the subject in a standard examination gown (23). Height and weight measurements were then converted to age- and sex-adjusted BMI percentiles and z-scores with the World Health Organization growth standards (9). Subjects were categorized, on the basis of their BMI percentile, as overweight (85% to <95%) and obese (≥95%) (9).
Waist circumference was measured to the nearest 0.1 cm at the end of a normal expiration with a steel measuring tape placed at the top of the iliac crest, with the patient in a standing position. The WHtR was calculated (waist circumference in centimeters divided by height in centimeters) and classified into 3 categories (<0.5, 0.5 to <0.6, ≥0.6). These categories were selected to provide similar categorization as a previous study involving WHtR (24). The WHtR category and BMI category were combined to form a 9-category variable. There were some groups with too few subjects to allow for adequate statistical analysis. As such, the 4 subjects with BMI <85% and WHtR ≥0.6 and the 13 subjects with BMI 85% to <95% and WHtR ≥0.6 were combined with the 1,437 subjects with BMI ≥95% and WHtR ≥0.6. These specific combinations were chosen to maintain subjects with WHtR ≥0.6 in 1 group.
Lipid and glycemic profile, inflammation
Patients were considered to be fasting if they reported at least 12 h of fasting before blood sampling. Complete lipid profile was obtained, including total cholesterol (TC), high-density lipoprotein cholesterol (HDL), triglyceride (TG), low-density lipoprotein (LDL) (calculated with the Friedewald calculation: LDL = TC − HDL − [TG/5]) (25), and non–HDL-cholesterol. The ratios of HDL to TC and TG to HDL were calculated. Values for LDL, TG, and TG/HDL ratio are reported only for patients with fasting samples. The TC, HDL, non-HDL, and HDL/TC are reported for all patients and separately for the subgroup with fasting samples. Non-fasting TC was considered to be borderline-high if 4.4 to <5.2 mmol/l (170 to <200 mg/dl) and high if ≥5.2 mmol/l (≥200 mg/dl) (26). Non-fasting non-HDL cholesterol was considered to be borderline-high if 3.10 to <3.75 mmol/l (120 to <145 mg/dl) and high if ≥3.75 mmol/l (≥145 mg/dl) (26). Fasting TG levels >1.7 mmol/l (>150 mg/dl) were considered elevated (26).
Fasting glucose and insulin were measured, with glucose values ≥6.1 mmol/l (≥110 mg/dl) (26) and insulin values >118 pmol/l (>17 μIU/ml, reference value used by our academic institution, on the basis of an internal study) considered elevated. Insulin resistance was quantified with the homeostatic model assessment (HOMA-IR), with fasting plasma glucose and insulin measurements (HOMA-IR = [glucose × insulin]/22.5). A value of ≥3.16 was considered abnormal (27). Hemoglobin A1c was measured as a surrogate marker of blood glucose control, with values >6.5% considered abnormal (28).
Degree of background inflammation was measured by level of C-reactive protein (CRP), measured through an ultra-sensitive assay. Values 0.7 to <3 mg/l were considered to be mildly elevated, and those ≥3 mg/l were considered to be elevated (29). Both aspartate aminotransferase and alanine aminotransferase were used as surrogate markers of liver inflammation. For both measures, values >40 U/l were considered abnormal (30).
Replicated measurements were used to obtain systolic and diastolic blood pressure levels. Up to 4 consecutive blood pressure readings were obtained after the maximum inflation level was determined, and the subject rested quietly in a sitting position for 5 min. Mean values of the replicated measurements were used to determine blood pressure (23). Systolic and diastolic blood pressure were converted to age, sex, and height-based z-scores, which were used to classify each subject as pre-hypertensive (90th to <95th percentile), stage 1 hypertensive (≥95th to <99th percentile), or stage 2 hypertensive (≥99th percentile) (31).
Metabolic syndrome was defined according to the National Cholesterol Education Program Adult Treatment Panel III pediatric definition (32,33), which requires the presence of 3 or more of the following: 1) waist circumference >90th percentile for age and sex (34); 2) fasting TG ≥1.25 mmol/l (≥110 mg/dl); 3) HDL <1.03 mmol/l (<40 mg/dl); 4) blood pressure >90th percentile for age, height and sex; and 5) fasting glucose ≥5.6 mmol/l (≥100 mg/dl).
Data are reported as means with SD, median with 5th and 95th percentiles, and frequencies as appropriate. All continuous variables with the exception of CRP level (which was categorized) had a distribution close enough to normal to be treated as normal. Linear, logistic, and ordinal logistic regression models (as appropriate), adjusted for age at testing and sex, were used to determine the relative contribution of WHtR and BMI z-score (modeled as continuous variables) on cardiometabolic risk factors. Reported p values represent the statistical significance of the contribution of each element (WHtR or BMI z-score) while adjusting for the other one. Additionally, linear and logistic regression models were also created with the normal BMI percentile category (<85%) and WHtR <0.5 as the reference category and comparing all other WHtR/BMI percentile category combinations. These models were also used to determine the effect of WHtR category within BMI percentile categories with WHtR <0.5 as the reference category. An interaction term was used to determine the difference between male and female subjects in regard to the association between worsening cardiometabolic risk factors and increased adiposity. We opted not to perform a weighted analysis, despite the weighted sample strategy used by NHANES, because our objective was to find associations at an individual level and not to report population estimates. As such any reported prevalence in this study should not be taken as representing national estimates. All statistical analyses were performed with SAS statistical software (version 9.3, SAS Institute, Cary North Carolina).
Data were available for 14,493 subjects (age range 5 to 18 years, 50% male, 4,238 [31%] with fasting blood work). Demographic data, lipid profile, glycemic profile, inflammation, blood pressure, and metabolic syndrome stratified by BMI percentile and WHtR categories are reported in Table 1. Increased BMI category was universally associated with increased cardiometabolic risk factors, including all elements of lipid and glycemic profile, inflammation, blood pressure, and metabolic syndrome. With few exceptions, WHtR was found to further stratify cardiometabolic risk factor levels beyond BMI percentile category alone (Table 1). Patients with the highest level of adiposity (WHtR ≥0.6 and BMI ≥95th percentile) were found to have a high proportion of abnormal levels in cardiometabolic risk factors.
Mean values for selected cardiometabolic risk factors (Fig. 1) and proportion of abnormal findings (Fig. 2) across BMI percentile and WHtR categories, adjusted for age and sex, are presented in Figures 1 and 2. For most variables, increased BMI percentile category was associated with increased cardiovascular risk factors. Elevated WHtR category, regardless of BMI percentile category was associated with increased cardiometabolic risk factors, whereas high BMI percentile category but low WHtR category had cardiometabolic risk factor levels approaching that of the normal BMI category population. The association between increased adiposity and cardiometabolic risk factors was mediated by sex (Table 2). Although increased adiposity was associated with increased cardiometabolic risk in both sexes, the increase was proportionally more marked in male subjects than in female subjects. A significant proportion of obese subjects with a WHtR ≥0.6 had abnormal cardiometabolic risk factor levels (18% to 78%, depending on the cardiometabolic risk factor) (Fig. 2). Those patients with both elevated BMI category and elevated WHtR category were at significantly higher risk for all measures at a higher level than expected by either measure alone, suggesting a distinct pathological process at that level of adiposity.
Our study sought to investigate the role of WHtR in the cardiometabolic risk assessment of normal, overweight, and obese children categorized by BMI percentile category. We found that an increased WHtR category was associated with worsened cardiometabolic risk and an increased frequency of abnormal cardiometabolic risk factor levels. Subjects categorized as overweight and obese by BMI percentile with a WHtR <0.5 had a cardiometabolic risk that was approaching that of subjects with a normal BMI percentile category. This is of particular importance for overweight subjects, because 55% had a WHtR <0.5, indicating that a large proportion of overweight children might have a cardiometabolic risk that is not as high as previously perceived. Ninety-seven percent of subjects with a normal BMI category had a WHtR <0.5, indicating that subjects with normal BMIs are unlikely to have elevated levels of central adiposity. Conversely, the majority of obese subjects had an elevated waist measure.
Clinical utility of the WHtR
Because BMI measurement cannot differentiate between fat and fat-free mass or serve as an indicator of the anatomical distribution of mass, an increased BMI might not reflect increased central adiposity. Waist measures, conversely, are an effective indicator of central fat distribution, serving as a strong marker of cardiometabolic risk in the pediatric population (18–22,35,36). Our group elected to focus our statistical analysis on the utility of WHtR in further specifying cardiometabolic risk. In the previous work of our group (24), WHtR and waist circumference percentile showed very similar associations with cardiometabolic risk in the normal, overweight, and obese population. The WHtR might be advantageous, because it is a convenient marker that does not require conversion to z-scores or percentiles (18,19,21,22). It has been previously suggested that a WHtR of 0.5 might serve as a consistent threshold of increased risk across sex and ethnicity, from childhood to adulthood, and might be easier for patients and families to understand (19,37), thus creating the potential for an important public health message of “keep your waist circumference half your height.”
Although strong associations were observed between increasing WHtR and worsening cardiometabolic risk factor levels in both male and female subjects, these associations seemed stronger in male subjects. Post-pubertal female subjects have a predisposition for gluteofemoral fat deposition, an anatomic area of adiposity that has been shown to play a protective role in the development of cardiometabolic risk (38). Therefore, an increase in the WHtR in some female subjects might be potentially offset by the presence of increased gluteofemoral fat relative to male subjects. To the best of our knowledge, this finding has not been previously shown and needs to be further explored.
Review of previous published data
Although guidelines from the major American, British, and Canadian medical associations acknowledge the potential added benefits of using waist measures, they are not in full agreement with respect to its use in screening the pediatric population (39–41). The present study shows the incorporation of waist measures improves cardiometabolic risk specification among children, suggesting that waist measures should be considered in routine pediatric screening.
Our group recently assessed the association of waist measures with non-fasting lipid and blood pressure screening assessments of 4,384 adolescents enrolled in the Heart Niagara, Inc., Healthy Heart Schools Program in the Niagara Region of Ontario, Canada (24). Increasing WHtR categories corresponded with a worsening non-fasting lipid profile and increased odds of a higher degree of hypertension, with the greatest associations occurring in obese subjects. Overweight and obese subjects with WHtRs <0.5 were found to have significantly different mean lipid values relative to normal subjects. This study was limited, because it assessed only 14- to 15-year-olds through the use of non-fasting lipids and blood pressure data and did not include inflammatory markers. Our present study similarly showed an increased WHtR to be associated with increased cardiometabolic risk. In addition, we found that overweight and obese subjects with a WHtR <0.5 had a cardiometabolic risk approaching that of subjects with a normal BMI.
Obesity in childhood is associated with an increased prevalence of other cardiometabolic risk factors (3,6). Furthermore, cardiometabolic risk factors in childhood track into adulthood (2,4), posing a significant public health concern. Our study of 14- to 15-year-olds in Ontario, Canada found an increased prevalence of elevated cardiometabolic risk factors in obese children with a WHtR ≥0.6. For example, 25% had elevated non-HDL levels, and 17% had stage 1 or 2 hypertensive blood pressure levels (24). Two other school-based studies have shown similar results (42,43). In the present study, approximately 10% of subjects were obese and had a WHtR ≥0.6. This group had an increased prevalence of elevated cardiometabolic risk factor levels (Fig. 2). For example, 18% had an elevated CRP, 32% met the modified criteria for metabolic syndrome, and 69% had an elevated HOMA-IR index (Fig. 2). In contrast, overweight and obese subjects with WHtR <0.5 did not tend to have a significantly increased prevalence of abnormal cardiometabolic risk factor levels relative to subjects with a normal BMI. Given that 55% of subjects in our study categorized as overweight by BMI had a WHtR <0.5, the use of WHtR in pediatric screening might serve to further alert the clinician to evaluate for additional cardiometabolic risk factors, particularly in overweight patients.
A number of studies have found waist measures to provide an added benefit in the assessment of cardiometabolic risk for children categorized by BMI (18,20,44–46). However, many of these studies were derived from subject populations that are not representative of the American pediatric population, either through the use of data from the Bogalusa Heart Study (18,19,45,46), a Caucasian population (20), or a biracial population (44). The present study, through the use of NHANES data, is derived from a large and diverse subject population. In addition, the present study incorporated a wide range of fasting and non-fasting cardiometabolic risk factors, including inflammatory markers and a modified definition for metabolic syndrome.
To the best our knowledge, previous studies assessing the role of waist measures in children categorized by BMI have not incorporated surrogate markers of inflammation such as CRP and liver transaminases. The prevalence of increased cardiometabolic risk factor levels in children with increased central adiposity might be attributed to the inflammatory state of adiposopathy (17). Adiposopathy refers to the presence of pathological, anatomic, and/or functional adipose tissue disturbances that are promoted by a positive caloric balance in genetically and environmentally susceptible individuals. This results in adverse endocrine and immune responses that might promote cardiovascular disease or worsen metabolic disease (17). In such an inflammatory state, CRP might be increased both through direct production by adipocytes and through increased hepatic production (47–49). Adiposopathy is more likely to develop in the setting of excessive central or visceral adiposity (17). This might help explain the large discrepancy in the prevalence of metabolic syndrome and elevated CRP and liver transaminases among obese children with a high WHtR relative to those with a lower WHtR found in the present study. Therefore, obese children with increased central adiposity but without any other diagnosed cardiometabolic disease might not be “healthy” as previously perceived due to the underlying pro-inflammatory state of adiposopathy.
The primary limitation of this study was that only associations but not causality can be inferred, given its cross-sectional design. The statistical analysis involved the categorization of BMI and WHtR according to clinically used standards to optimize the applicability of the results. It is difficult to assess the independence of the WHtR relationship within BMI categories to BMI gradient within BMI categories itself, because both variables are highly collinear. Although WHtR seems to further specify cardiometabolic risk in this study, longitudinal studies are required to track its relationship to cardiometabolic risk throughout childhood and into adulthood. Pubertal stage information of the subjects was not considered, and the data were not stratified on the basis of ethnicity or sex. Although pubertal stage and ethnicity might confer to different absolute levels of cardiometabolic risk, there is no indication that the associations with increased WHtR within each BMI category would vary. Further studies might be warranted to specifically address variations on the basis of these factors.
The WHtR seems to further specify the cardiometabolic risk assessment of overweight and obese children. A significant proportion of obese subjects with a high WHtR had abnormal cardiometabolic risk factor levels and met the criteria for metabolic syndrome. Overweight and obese children with a low WHtR seem to have a cardiometabolic risk that approaches that of normal BMI children. Given that most overweight subjects had a low WHtR, WHtR measurement might be a particularly useful tool in this population, helping the clinician further stratify cardiometabolic risk. The WHtR seems to be an important discriminating measurement in the cardiometabolic risk assessment of children and should thus be included as part of routine pediatric screening of overweight and obese children.
The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- body mass index
- C-reactive protein
- high-density lipoprotein
- homeostatic model assessment of insulin resistance
- low-density lipoprotein
- National Health and Nutrition Examination Survey
- total cholesterol
- waist/height ratio
- Received September 24, 2012.
- Revision received December 24, 2012.
- Accepted January 8, 2013.
- American College of Cardiology Foundation
- Freedman D.S.,
- Dietz W.H.,
- Srinivasan S.R.,
- Berenson G.S.
- Steinberger J.,
- Daniels S.R.,
- Eckel R.H.,
- et al.
- Ford E.S.,
- Mokdad A.H.,
- Ajani U.A.
- Cole T.J.,
- Bellizzi M.C.,
- Flegal K.M.,
- Dietz W.H.
- Himes J.H.,
- Dietz W.H.
- Demerath E.W.,
- Schubert C.M.,
- Maynard L.M.,
- et al.
- Bays H.E.
- Freedman D.S.,
- Kahn H.S.,
- Mei Z.,
- et al.
- Centers for Disease Control and Prevention; National Center for Health Statistics
- Friedewald W.T.,
- Levy R.I.,
- Fredrickson D.S.
- Kavey R.E.,
- Allada V.,
- Daniels S.R.,
- et al.
- Pearson T.A.,
- Mensah G.A.,
- Alexander R.W.,
- et al.
- Barlow S.E.
- Lau D.C.,
- Douketis J.D.,
- Morrison K.M.,
- Hramiak I.M.,
- Sharma A.M.,
- Ur E.
- Baumer J.H.
- Freedman D.S.,
- Dietz W.H.,
- Srinivasan S.R.,
- Berenson G.S.
- Janssen I.,
- Katzmarzyk P.T.,
- Srinivasan S.R.,
- et al.
- Calabro P.,
- Chang D.W.,
- Willerson J.T.,
- Yeh E.T.