waist circumference and cardiovaskular risk factors in prepubertal children

9
Waist Circumference and Cardiovascular Risk Factors in Prepubertal Children Claudio Maffeis, Angelo Pietrobelli, Alessandra Grezzani, Silvia Provera, and Luciano Tato ` Abstract MAFFEIS, CLAUDIO, ANGELO PIETROBELLI, ALESSANDRA GREZZANI, SILVIA PROVERA, AND LUCIANO TATO ` . Waist circumference and cardiovascular risk factors in prepubertal children. Obes Res. 2001;9: 179 –187. Objective: Intra-abdominal fat has been identified as being the most clinically relevant type of fat in humans. There- fore, an assessment of body-fat distribution could possibly identify subjects with the highest risk of adverse lipid pro- file and hypertension. Few data on the relationship between body-fat distribution and cardiovascular risk factors are available in children, especially before puberty. Research Methods and Procedures: This cross-sectional study was undertaken to explore the relationship between anthropometric variables, lipid concentrations, and blood pressure (BP) in a sample of 818 prepubertal children (ages 3 to 11 years) and to assess the clinical relevance of waist circumference in identifying prepubertal children with higher cardiovascular risk. Height, weight, triceps and sub- scapular skinfolds, waist circumference, and BP were mea- sured. Plasma levels for triacylglycerol, total cholesterol, high-density lipoprotein (HDL) cholesterol, low-density li- poprotein cholesterol, apolipoprotein A1 (ApoA1), and apo- lipoprotein B (ApoB) were determined. Results: Females were fatter than males (5.8 [3.5] vs. 4.8 [3.3] kg of fat mass; p , 0.01). Males had higher HDL cholesterol and ApoA1/ApoB plasma concentrations than females (p , 0.001 and p , 0.01, respectively). Waist circumference had a higher correlation with systolic and diastolic BP (r 5 0.40 and 0.29, respectively; p , 0.001) than triceps (r 5 0.35 and 0.21, respectively; p , 0.001) and subscapular (r 5 0.28 and 0.16, respectively; p , 0.001) skinfolds and relative body weight (0.33 and 0.23, respectively; p , 0.001). Multivariate linear model analysis showed that ApoA1/ApoB, HDL cholesterol, total choles- terol/HDL cholesterol, and systolic as well as diastolic BP were significantly associated with waist circumference and triceps and subscapular skinfolds, independently of age, gender, and body mass index. Discussion: Waist circumference as well as subscapular and triceps skinfolds may be helpful parameters in identifying prepubertal children with an adverse blood-lipids profile and hypertension. However, waist circumference, which is easy to measure and more easily reproducible than skin- folds, may be considered in clinical practice. Children with a waist circumference greater than the 90th percentile are more likely to have multiple risk factors than children with a waist circumference that is less than or equal to the 90th percentile. Key words: child, adiposity, body composition, plasma lipoproteins, waist circumference Introduction Several epidemiological studies support the hypothesis that the relationship between adiposity and risk of disease begins early in life (1,2). Adipose tissue stores have differ- ent metabolic activity and relationships to disease risk as a function of their distribution in the body. In adults, intra- abdominal adipose tissue (IAAT) is the most clinically relevant type of body fat, apart from total body fat. Meta- bolic complications and adverse health effects of increased IAAT include high blood pressure (BP), hyperinsulinemia, type 2 diabetes, and dyslipidemia (3–5). In prepubertal children, the relationship between body-fat distribution and disease risk factors is not clear. Accurate methods used to assess total body fat (DXA) and body-fat distribution (computed tomography and mag- netic resonance imaging) in humans are not suitable for use in large population studies because of cost, irradiation ex- posure (i.e., computed tomography), and limited availability outside the research setting (6). To obtain a reasonable estimation of body-fat distribution in children, several an- thropometric parameters have been proposed, such as sub- Submitted for publication August 28, 2000. Accepted for publication in final form December 11, 2000. Department of Pediatrics, University of Verona, Polyclinic, Verona, Italy. Address correspondence to Claudio Maffeis, MD, Department of Pediatrics, University of Verona, Polyclinic, 37134 Verona, Italy. E-mail: [email protected] Copyright © 2001 NAASO OBESITY RESEARCH Vol. 9 No. 3 March 2001 179

Upload: chainurridha-sii-dedek

Post on 17-Dec-2015

216 views

Category:

Documents


4 download

DESCRIPTION

waist circumference

TRANSCRIPT

  • Waist Circumference and Cardiovascular RiskFactors in Prepubertal ChildrenClaudio Maffeis, Angelo Pietrobelli, Alessandra Grezzani, Silvia Provera, and Luciano Tato`

    AbstractMAFFEIS, CLAUDIO, ANGELO PIETROBELLI,ALESSANDRA GREZZANI, SILVIA PROVERA, ANDLUCIANO TATO` . Waist circumference and cardiovascularrisk factors in prepubertal children. Obes Res. 2001;9:179187.Objective: Intra-abdominal fat has been identified as beingthe most clinically relevant type of fat in humans. There-fore, an assessment of body-fat distribution could possiblyidentify subjects with the highest risk of adverse lipid pro-file and hypertension. Few data on the relationship betweenbody-fat distribution and cardiovascular risk factors areavailable in children, especially before puberty.Research Methods and Procedures: This cross-sectionalstudy was undertaken to explore the relationship betweenanthropometric variables, lipid concentrations, and bloodpressure (BP) in a sample of 818 prepubertal children (ages3 to 11 years) and to assess the clinical relevance of waistcircumference in identifying prepubertal children withhigher cardiovascular risk. Height, weight, triceps and sub-scapular skinfolds, waist circumference, and BP were mea-sured. Plasma levels for triacylglycerol, total cholesterol,high-density lipoprotein (HDL) cholesterol, low-density li-poprotein cholesterol, apolipoprotein A1 (ApoA1), and apo-lipoprotein B (ApoB) were determined.Results: Females were fatter than males (5.8 [3.5] vs. 4.8[3.3] kg of fat mass; p , 0.01). Males had higher HDLcholesterol and ApoA1/ApoB plasma concentrations thanfemales (p , 0.001 and p , 0.01, respectively). Waistcircumference had a higher correlation with systolic anddiastolic BP (r 5 0.40 and 0.29, respectively; p , 0.001)than triceps (r 5 0.35 and 0.21, respectively; p , 0.001)and subscapular (r 5 0.28 and 0.16, respectively; p ,0.001) skinfolds and relative body weight (0.33 and 0.23,

    respectively; p , 0.001). Multivariate linear model analysisshowed that ApoA1/ApoB, HDL cholesterol, total choles-terol/HDL cholesterol, and systolic as well as diastolic BPwere significantly associated with waist circumference andtriceps and subscapular skinfolds, independently of age,gender, and body mass index.Discussion: Waist circumference as well as subscapular andtriceps skinfolds may be helpful parameters in identifyingprepubertal children with an adverse blood-lipids profileand hypertension. However, waist circumference, which iseasy to measure and more easily reproducible than skin-folds, may be considered in clinical practice. Children witha waist circumference greater than the 90th percentile aremore likely to have multiple risk factors than children witha waist circumference that is less than or equal to the90th percentile.

    Key words: child, adiposity, body composition, plasmalipoproteins, waist circumference

    IntroductionSeveral epidemiological studies support the hypothesis

    that the relationship between adiposity and risk of diseasebegins early in life (1,2). Adipose tissue stores have differ-ent metabolic activity and relationships to disease risk as afunction of their distribution in the body. In adults, intra-abdominal adipose tissue (IAAT) is the most clinicallyrelevant type of body fat, apart from total body fat. Meta-bolic complications and adverse health effects of increasedIAAT include high blood pressure (BP), hyperinsulinemia,type 2 diabetes, and dyslipidemia (35). In prepubertalchildren, the relationship between body-fat distribution anddisease risk factors is not clear.

    Accurate methods used to assess total body fat (DXA)and body-fat distribution (computed tomography and mag-netic resonance imaging) in humans are not suitable for usein large population studies because of cost, irradiation ex-posure (i.e., computed tomography), and limited availabilityoutside the research setting (6). To obtain a reasonableestimation of body-fat distribution in children, several an-thropometric parameters have been proposed, such as sub-

    Submitted for publication August 28, 2000.Accepted for publication in final form December 11, 2000.Department of Pediatrics, University of Verona, Polyclinic, Verona, Italy.Address correspondence to Claudio Maffeis, MD, Department of Pediatrics, University ofVerona, Polyclinic, 37134 Verona, Italy. E-mail: [email protected] 2001 NAASO

    OBESITY RESEARCH Vol. 9 No. 3 March 2001 179

  • cutaneous skinfolds and body circumferences, which areeasy to perform and have a sufficient degree of accuracy.Some anthropometric measures or indexes, such as bodymass index (BMI) and waist circumference, have been usedin a large number of studies on adults to analyze the asso-ciation between adiposity and cardiovascular risk factors(7). Few studies have shown that waist circumference maybe a better predictor of cardiovascular disease than BMI andwaist-to-hip ratio (8). In fact waist circumference in adultsis better correlated with visceral adipose tissue than BMIand waist-to-hip ratio (9). In contrast, the degree of associ-ation between cardiovascular risk factors and anthropomet-ric parameters has not been studied extensively in prepu-bertal children. Therefore, the purposes of the present studywere to explore the relationship between anthropometricvariables, lipid profile, and BP in a group of 818 children of3 to 11 years of age and to assess the clinical relevance ofwaist circumference in identifying prepubertal children withhigher cardiovascular risks.

    Research Methods and ProceduresSubjects

    A sample of 885 3- to 11-year-old children living innortheastern Italy was used. Selection of sampling areas andparticipant inclusion have been described previously (10).Briefly, children living in six areas of northeastern Italywere selected, and a number of school districts within eacharea were chosen to obtain a representative sample of chil-dren from different socioeconomic backgrounds and envi-ronmental conditions. Each child underwent a physical ex-amination by pediatricians who took anthropometricmeasurements. All of the subjects were at the prepubertalstage, as verified by an expert pediatric endocrinologist.Pubertal development was clinically assessed based on Tan-ner stages (11). The parents of each child gave their in-formed consent to participate in the study.

    A total of 67 subjects were not included in the results andstatistical analysis because of missing data. We recordedcomplete data and statistical analysis results for 818 chil-dren (443 males and 375 females).

    The protocol was in accordance with the Helsinki Dec-laration of 1975 as revised in 1983.

    Physical CharacteristicsMeasurements of height, weight, triceps and subscapular

    skinfolds, waist circumference, and BP were carried outunder fasting conditions. Body weight was determined tothe nearest 0.5 kg on standard physicians beam scales withthe child wearing only underwear and no shoes. Height wasmeasured to the nearest 0.5 cm on standardized, wall-mounted height boards according to the following protocol:no shoes, heels together, and childs heels, buttocks, shoul-ders, and head touching the vertical wall surface with line-

    of-sight aligned horizontally. BMI was defined as weight/height2 and was expressed in kilograms per squared meter.Each of the standard physicians beam scales and wall-mounted height boards used to measure the children werecalibrated previously, using three different weights and onereference tape. Skinfold thickness was measured to thenearest millimeter three times with a Harpenden skinfoldcaliper on both the right and left sides of the body (CMS;Weighing Equipment Ltd, London, UK). The triceps skin-fold locus is halfway between the acromion and olecranonon the back of the arm measured with the elbow bent. Wemeasured the triceps skinfold with the arm pendant, whereasthe subscapular skinfold was measured just below the tip ofthe scapula (12). Readings were taken 3 seconds after thecaliper jaws were released. In the present analysis, we usedthe mean of right- and left-sided measurements for eachskinfold. Lohmans formulas were used to estimate relativebody fat mass, based on the measurement of triceps andsubscapular skinfolds (12). Waist circumference was mea-sured to the nearest centimeter with a flexible steel tapemeasure while the subjects were in the standing position atthe end of gentle expiration (12). The following anatomicallandmarks were used: laterally, midway between the lowestportion of the rib cage and iliac crest, and anteriorly midwaybetween the xiphoid process of the sternum and the umbi-licus (12). Children were defined as obese on the basis oftheir BMI. In particular, BMI cutoff values were used,which were age- and gender-specific, according to Cole etal. (13). Relative body weight (RBW), calculated as thepercentage of the ratio between weight and body weight atthe 50th percentile for age and gender, was obtained in allthe children. In accordance with the guidelines approved atthe Italian Consensus Conference on Obesity (14), Tannersgrowth tables were used to calculate the RBW% of eachchild (15). The physician made three BP measurements onthe left arm over a period of 30 minutes with the subjectsupine, using a mercury sphygmomanometer. The cuffsused had bladders long enough to circle at least one-half ofthe upper arm without overlapping and widths that coveredat least two-thirds of the upper arm. Systolic BP (SBP)(Korotkoff phase I) and diastolic BP (DBP) (Korotkoffphase V) were measured three times, and the average wasused for analysis.

    Lipids and LipoproteinsFasting venipuncture samples were drawn for lipid

    determinations. Plasma triacylglycerol (TG) and choles-terol were measured enzymatically (Abbott VP, Milan,Italy), using spectrophotometric methods in our HospitalCore laboratory (16,17). The plasma high-density li-poprotein (HDL) cholesterol fraction was obtained afterprecipitation using phosphotungstic reagent. Apolipopro-tein A1 (ApoA1) and apolipoprotein B (ApoB) weremeasured by radial immunodiffusion (18). A 10% sample

    Waist Circumference and Disease Risk, Maffeis et al.

    180 OBESITY RESEARCH Vol. 9 No. 3 March 2001

  • was randomly chosen each day to assess measurementerror, and intraclass correlation coefficients ranged from0.94 (HDL cholesterol) to 0.99 (TG). Our Core labora-tory monitored the accuracy of total cholesterol (TC),HDL cholesterol, TG, ApoA1, and ApoB measurements.The low-density lipoprotein (LDL) cholesterol level wascalculated using the Friedewald formula (LDL choles-terol 5 TC 2 HDL cholesterol 2 TG/5) (19).

    Statistical AnalysisAll statistical analyses were carried out using the SPSS

    software version 9.0 for Windows (SPSS Inc., Chicago, IL)package for personal computers. Baseline variables aredescribed as groups mean and SD. Differences betweengender and between non-obese and obese subjects wereanalyzed using the Students t test for unpaired samples.

    Zero-order correlations were performed first to assess un-adjusted association between body composition parameters,plasma lipids, and BP. A x2 test was run to compare thecorrelation coefficients of the relationships between cardio-vascular risk factors and anthropometric variables (20). Thedegree of association between plasma lipids and BP, ad-justed for age, gender (dummy variable), BMI (covariates),and anthropometric parameters, was calculated using a par-tial correlation multivariate linear model analysis. PlasmaTG was not normally distributed; therefore, it was expressedas its logarithm, which normalized the distribution.

    To assess the effects of waist circumference on the clus-tering of risk factors, the children were divided into normaland increased-risk groups for TC, LDL cholesterol, HDLcholesterol, SBP, and DBP. The cutoff points were 0.9mM/L (35 mg/dL) for HDL cholesterol, 3.4 mM/L (130

    Table 1. Physical characteristics and plasma lipids profile of the 845 prepubertal children

    Total sample (n 5 818) Males (n 5 443) Females (n 5 375)Age (years) 7.7 (2) 7.8 (2) 7.6 (2)Weight (kg) 27.7 (9) 28.2 (9) 27.2 (8)Height (cm) 126 (15) 127 (15) 125 (15)BMI (kg/m2) 17 (2) 17 (2) 17 (3)RBW (%) 104 (15) 106 (14) 102 (15)*Triceps skinfold (mm) 12 (5) 11 (5) 13 (5)*Subscapular skinfold (mm) 8 (5) 7 (5) 9 (5)*Fat mass (kg) 5 (3) 5 (3) 6 (4)*Waist circumference (cm) 57 (7) 57 (7) 56 (7)Systolic blood pressure (mm Hg) 107 (13) 107 (14) 106 (12)Diastolic blood pressure (mm Hg) 65 (11) 66 (11) 65 (11)Total cholesterol

    (mmol/L) 4.72 (0.80) 4.74 (0.79) 4.69 (0.82)(mg/dL) 182.3 (31) 183.1 (30.6) 181.2 (31.8)

    LDL cholesterol(mmol/L) 2.91 (0.69) 2.88 (0.68) 2.94 (0.72)(mg/dL) 112.4 (26.9) 111.5 (26.2) 113.5 (27.8)

    HDL cholesterol(mmol/L) 1.49 (0.37) 1.54 (0.38) 1.43 (0.35)(mg/dL) 57.5 (14.4) 59.6 (14.8) 55.1 (13.7)

    TC/HDL cholesterol 3.32 (0.90) 3.22 (0.84) 3.44 (0.88)*TG

    (mmol/L) 0.69 (0.31) 0.68 (0.33) 0.72 (0.27)(mg/dL) 61.5 (27) 60 (29.3) 63.4 (24.2)

    ApoA1/ApoB 2.13 (0.69) 2.20 (0.76) 2.04 (0.60)*

    Data are shown as mean 6 SD.* p , 0.01. p , 0.001.

    Waist Circumference and Disease Risk, Maffeis et al.

    OBESITY RESEARCH Vol. 9 No. 3 March 2001 181

  • mg/dL) for LDL cholesterol, 4.7 mM/L (180 mg/dL) forTC, and 90th percentile for SBP and DBP (21,22). How-ever, to include intercorrelated variables in the same cate-gory, we included LDL cholesterol and HDL cholesterol inthe analysis; they were not correlated and excluded TC,which was highly correlated with LDL cholesterol and HDLcholesterol. Children with SBP or DBP higher than the 90thpercentile were considered hypertensive. Therefore, threecardiovascular risk-factor categories were considered (LDLcholesterol, HDL cholesterol, and BP). The percentage ofsubjects with no, one, two, or three risk factors was calcu-

    lated. The MannWhitney test for ordinal data was used tocompare the positivity for risk-factor categories (none, one,two, or three) of the two groups of children: Group 1,children with a waist circumference less than the 90thpercentile; Group 2, children with a waist circumferencegreater than the 90th percentile. To assess the predictionlevel of waist circumference on the probability of havingcardiovascular risk factors (HDL cholesterol, LDL choles-terol, ApoA1/ApoB, and BP), we performed a multivariatelogistic regression analysis with backward stepping of vari-ables and an evaluation of the model using three goodness-of-fit x2 statistics. In all the analyses, a probability level ofp , 0.05 was used to indicate statistical significance.

    ResultsThe physical characteristics of the 818 prepubertal chil-

    dren (443 males and 375 females) are shown in Table 1.Triceps and subscapular skinfold measurements and fatmass were significantly higher in females than in males(p , 0.01). RBW% was significantly higher in males thanin females (p , 0.01). No other significant differences inthe anthropometric parameters between genders werefound. Regarding plasma lipids, TC and TG showed nodifferences between genders. HDL cholesterol and ApoA1/ApoB were significantly lower in females than males (p ,0.001 and p , 0.01, respectively). TC/HDL cholesterol wassignificantly higher in females than males (p , 0.01).

    The physical characteristics of the children divided intoobese and non-obese groups are shown in Table 2. Asexpected, all of the anthropometric variables and BP weresignificantly higher (p , 0.01) in the obese children than inthe non-obese children. In the obese children, the LDLcholesterol and TC/HDL cholesterol were significantlyhigher (p , 0.05 and p , 0.01, respectively) and HDLcholesterol was significantly lower (p , 0.05) than in thenon-obese children. No other differences were found in thelipids profile between obese and non-obese subjects.

    Body composition parameters were significantly corre-lated with each other (p , 0.001; Table 3) as well as withSBP and DBP. The correlation between SBP and waistcircumference (r 5 0.40; p , 0.001) was significantlyhigher (x2 5 9.08; p , 0.05) than those with tricepsskinfold (r 5 0.35; p , 0.001), subscapular skinfold (r 50.28; p , 0.001), and RBW (r 5 0.33; p , 0.001). Thecorrelation between DBP and waist circumference (r 50.29; p , 0.001) was higher, but not significantly higher (x25 7.87; p 5 not significant), than those with triceps skin-fold (r 5 0.21; p , 0.001), subscapular skinfold (r 5 0.16;p , 0.001), and RBW (r 5 0.23; p , 0.001). Triceps andsubscapular skinfolds were correlated with TG (r 5 0.09and 0.14, respectively; p , 0.001). Subscapular skinfoldshowed a negative correlation with HDL cholesterol (r 520.11; p , 0.001). Subscapular skinfold was positivelycorrelated with TC/HDL cholesterol (r 5 0.11; p , 0.001).

    Table 2. Physical characteristics and plasma lipidsprofile of the children divided into two groups: obesechildren and non-obese children

    Obese(n 5 43)

    Non-obese(n 5 775)

    Age (years) 8.1 (2) 7.7 (2)Weight (kg) 41.4 (10) 27.2 (8)*Height (cm) 132 (13) 126 (15)BMI (kg/m2) 23 (2) 17 (2)*RBW (%) 141 (10) 102 (12)*Triceps skin (mm) 22 (5) 12 (4)*Subscapular skin (mm) 19 (7) 7 (4)*Fat mass (kg) 13 (4) 5 (3)*Waist circumference (cm) 71 (8) 56 (6)*Systolic blood pressure

    (mm Hg)119 (14) 106 (13)*

    Diastolic blood pressure(mm Hg)

    72 (11) 65 (11)*

    Total cholesterol(mmol/L) 4.83 (0.77) 4.71 (0.8)(mg/dL) 186.6 (29.9) 181.9 (31.1)

    LDL cholesterol(mmol/L) 3.11 (0.66) 2.89 (0.69)(mg/dL) 120.5 (25.5) 111.7 (26.9)

    HDL cholesterol(mmol/L) 1.37 (0.33) 1.49 (0.37)(mg/dL) 52.9 (12.8) 57.9 (14.4)

    TC/HDL cholesterol 3.66 (0.80) 3.28 (0.86)TG

    (mmol/L) 0.75 (0.31) 0.69 (0.30)(mg/dL) 67.8 (27.3) 61.2 (27)

    ApoA1/ApoB 1.95 (0.49) 2.15 (0.70)

    Data are shown as mean 6 SD.* p , 0.001. p , 0.01. p , 0.05.

    Waist Circumference and Disease Risk, Maffeis et al.

    182 OBESITY RESEARCH Vol. 9 No. 3 March 2001

  • Tabl

    e3.

    Corre

    latio

    nm

    atrix

    ofa

    nth

    ropo

    met

    ricpa

    ram

    eter

    s,B

    P,an

    dpl

    asm

    alip

    ids

    inth

    e81

    8pr

    epub

    erta

    lchi

    ldre

    n

    Age

    RBW

    Tric

    eps

    skin

    fold

    Subs

    capu

    lar

    skin

    fold

    Wai

    stSB

    PD

    BPTC

    HD

    Lch

    oles

    tero

    lLD

    Lch

    oles

    tero

    lTC

    /HD

    Lch

    oles

    tero

    lTG

    Age

    RB

    W

    Tric

    eps

    skin

    fold

    0.29

    *0.

    69*

    Subs

    capu

    lars

    kinf

    old

    0.21

    *0.

    67*

    0.72

    *

    Wai

    st0.

    57*

    0.78

    *0.

    71*

    0.66

    *

    SBP

    0.35

    *0.

    33*

    0.35

    *0.

    28*

    0.40

    *

    DB

    P0.

    12*

    0.23

    *0.

    21*

    0.16

    *0.

    29*

    0.59

    *

    TC0.

    12*

    0.03

    0.02

    0.01

    0.04

    20.

    030.

    00

    HD

    Lch

    oles

    tero

    l0.

    30*

    20.

    012

    0.05

    20.

    11*

    0.05

    0.06

    20.

    040.

    41*

    LDL

    chol

    este

    rol

    20.

    005

    0.02

    0.03

    0.04

    0.01

    20.

    07

    0.00

    40.

    90*

    0.00

    8

    TC/H

    DL

    chol

    este

    rol

    0.27

    *0.

    010.

    050.

    11*

    20.

    062

    0.10

    *0.

    040.

    26*

    20.

    71*

    0.58

    *

    TG0.

    09*

    0.09

    *0.

    09*

    0.14

    *0.

    060.

    020.

    10*

    0.17

    *2

    0.33

    *0.

    16*

    0.49

    *

    Apo

    A1/

    Apo

    B0.

    23*

    0.03

    20.

    062

    0.05

    0.06

    0.13

    *0.

    12*

    20.

    21*

    0.53

    *2

    0.49

    *2

    0.66

    *2

    0.22

    *

    *p

    ,0.

    001.

    p,

    0.05

    .

    Waist Circumference and Disease Risk, Maffeis et al.

    OBESITY RESEARCH Vol. 9 No. 3 March 2001 183

  • No other correlations were found with body compositionparameters. ApoA1/ApoB was correlated with SBP (r 50.13; p , 0.001) and DBP (r 5 0.12; p , 0.001).

    Eight sets of multivariate linear model analysis wererun using ApoA1/ApoB, HDL cholesterol, TC/HDL cho-lesterol, LDL cholesterol, triacylglycerol logarithm, TC,SBP, and DBP as dependent variables, adjusted for age,gender, and BMI (covariates), and waist circumferenceand triceps and subscapular skinfolds as independentvariables (Table 4). Waist circumference, as well assubscapular and triceps skinfolds, was significantly as-sociated with ApoA1/ApoB, HDL cholesterol, TC/HDLcholesterol, SBP, and DBP (p , 0.01).

    The effects of waist circumference on the clustering ofrisk factors in the total sample are shown in Table 5. Thewaist circumference (90th percentile) of the boys and girls

    at different ages is shown in Figure 1. Approximately 19%of children with a waist circumference that was greater thanthe 90th percentile had two or more risk factors, comparedwith 9% of children with a waist circumference that wasless than or equal to the 90th percentile and 9.5% predictedby chance alone. Approximately 35% of the children with awaist circumference that was greater than the 90th percen-tile had no risk factors, compared with 80% of children witha waist circumference less than or equal to the 90th percen-tile and a predicted number of 48% (p , 0.01). Multivariatelogistic regression analysis revealed that children with awaist circumference above the 90th percentile for sex andage have a higher probability of having cardiovascular riskfactors. In particular, these children have a significantlygreater risk of having lower HDL cholesterol (odds ratio 50.97; 95% confidence interval: 0.96 to 0.99; p , 0.01) and

    Table 4. Multivariate linear model analysis using the total sample

    Variables F R2 p

    Independent variable: waist circumferenceTC 1.09 0.01 NSLDL cholesterol 0.86 20.01 NSHDL cholesterol 2.89 0.14 ,0.001TC/HDL cholesterol 2.20 0.09 ,0.001LogTG 1.33 0.03 ,0.05ApoA1/ApoB 2.08 0.08 ,0.001SBP 4.19 0.21 ,0.001DBP 2.63 0.12 ,0.001

    Independent variable: subscapular skinfoldTC 1.25 0.03 NSLDL cholesterol 1.21 0.03 NSHDL cholesterol 2.09 0.13 ,0.001TC/HDL cholesterol 1.85 0.10 ,0.001LogTG 1.40 0.05 ,0.01ApoA1/ApoB 1.73 0.09 ,0.001SBP 3.47 0.25 ,0.001DBP 2.38 0.16 ,0.001

    Independent variable: triceps skinfoldTC 1.12 0.02 NSLDL cholesterol 0.99 20.001 NSHDL cholesterol 2.21 0.16 ,0.001TC/HDL cholesterol 1.70 0.10 ,0.001LogTG 1.12 0.02 NSApoA1/ApoB 1.43 0.07 ,0.01SBP 3.01 0.25 ,0.001DBP 1.96 0.13 ,0.001

    Age, gender, and BMI were used as covariates.NS, Not significant.

    Waist Circumference and Disease Risk, Maffeis et al.

    184 OBESITY RESEARCH Vol. 9 No. 3 March 2001

  • higher BP (odds ratio 5 2.3; 95% confidence interval: 1.41to 3.72; p , 0.001) than subjects with a waist circumferencethat is less than the 90th percentile.

    DiscussionThe results of this study show that the plasma lipids

    profile and BP in prepubertal children are significantlyassociated with anthropometric indexes of body-fat distri-bution. In particular, using multivariate linear model anal-ysis with waist circumference and triceps and subscapularskinfolds as independent variables, we found a significantcorrelation between these variables and ApoA1/ApoB,

    HDL cholesterol, TC/HDL cholesterol, SBP, and DBP. Thehigh degree of association between waist circumference andtriceps and subscapular skinfolds suggests the use, with anegligible difference, of one of these three different anthro-pometric parameters. However, waist circumference hassome relevant advantages compared with skinfolds: 1) theinter- and intraindividual reproducibility of the measure-ments is higher in the former than in the latter, and 2) inclinical practice, the measurement of waist circumferenceinstead of skinfolds is easier and offers more accurateresults for the pediatrician (3). Moreover, de Ridder et al.demonstrated that waist circumference is a good measurefor truncal fat in girls (23). In addition, Goran et al. foundthat waist circumference was strongly correlated with sub-cutaneous adipose tissue (24). Based on the aforementionedconsideration, we have discussed the results obtained usingwaist circumference as the dependent variable.

    Chan et al. showed that in adults, the best determinationof cardiovascular risk can be achieved by using waist cir-cumference as a measure of body-fat distribution (25). Inprepubertal children, waist circumference, adjusted for ageand gender, significantly contributed the explanation ofinterindividual variability of HDL cholesterol and SBP.Including weight and height among the independent vari-ables in the regression analyses did not affect the predict-ability of waist circumference. In our sample, LDL choles-terol did not show a correlation with waist circumference asfound previously in studies on adults (26,27). The differ-ence between the findings for adults and children may beexplained by several factors. First, the hormonal pattern inprepubertal children is different from adults, particularlyregarding sex hormones. Testosterone and estradiol have

    Figure 1. Waist circumference (90th percentile) of males and females of different ages.

    Table 5. Number and percentage of subjects with no,one, two, or three risk factors by percentileof waist

    Number ofrisk factors

    Waist

    90th percentile*(n 5 75)

    0 368 (79.6%) 26 (35%)1 308 (41%) 35 (47%)2 64 (9%) 14 (19%)3 3 (0.4%) 0 (0%)

    Risk factors included are HDL cholesterol, LDL cholesterol,and BP.* p , 0.01.

    Waist Circumference and Disease Risk, Maffeis et al.

    OBESITY RESEARCH Vol. 9 No. 3 March 2001 185

  • been shown to affect body-fat distribution and lipid metab-olism in humans (28). Another contributing factor may bethe level of physical activity. BP and plasma lipids profile,in particular HDL cholesterol and LDL cholesterol, areassociated with fitness (29). Modern lifestyles promote sed-entary behavior and reduce the practice of sports or orga-nized physical exercise in children and adults (30). Inadults, a low level of fitness has been positively associatedwith mortality, apart from other risk factors such as obesity,smoking, alcohol intake, and a parental history of ischemicheart disease (31).

    Few studies are available on the relationship betweenwaist circumference and cardiovascular risk factors in pre-pubertal children (3,32,33). The findings of our study agreewith the results of the Bogalusa Heart Study. Both studiesfound that waist circumference had a consistent associationwith cardiovascular risk factors. As in our subjects, the 5- to9-year-old children of Bogalusa showed an inverse associ-ation between waist circumference and HDL cholesterol.

    Few studies have discussed the clustering of risk factorsamong children and adolescents (22,34,35). In our study,the clustering was significantly higher in children with awaist circumference greater than the 90th percentile than inchildren with a waist circumference less than the 90thpercentile. The choice of the 90th percentile was based onthe association between truncal fat and waist circumferenceaccording to Taylor et al. (36). In our study, the 90thpercentile of waist circumference is very similar to the 80thpercentile in the study by Taylor et al. (36), chosen by thoseauthors as the point closest to one on the correspondingreceiver operating characteristic curve. Our analysis dem-onstrated that children with a waist circumference that isabove the 90th percentile for sex and age are more likely tohave multiple risk factors than children with a waist cir-cumference that is equal to or below the 90th percentile.Children with a waist circumference greater than the 90thpercentile are less likely to have no risk factors.

    Epidemiological and clinical investigations haveshown that the relationship between obesity and cardio-vascular risk factors begins early in life (1,2,5). Accord-ing to the criteria used to define obesity in the sample ofchildren we recruited for this study, we analyzed thesimple correlation between anthropometric indexes andplasma lipids profile in a subsample of 43 obese prepu-bertal children. In this sample, SBP and DBP showed abetter correlation (r 5 0.45 and 0.39, respectively; p ,0.01) with waist circumference than the total sample.Moreover, waist circumference showed a negative corre-lation with HDL cholesterol (r 5 20.31; p , 0.05; datanot shown). Prepubertal obese children could have alarger intra-abdominal fat store, which could explain ourfinding. Waist circumference measurement is not able todiscriminate between IAAT and subcutaneous adiposetissue (2). However, Taylor et al. (36) found recently that

    waist circumference correctly identified a high propor-tion of children and adolescents with high trunk-fat massas measured by a state-of-the-art measurement. Theyconcluded that waist circumference is a simple techniquethat could be used to screen for high central obesity inchildren (36). Therefore, we cannot conclude that therelationship between waist circumference and cardiovas-cular risk factors in these children is due to IAAT or totalfat. Despite the important effect of body fat location onthe development of metabolic disturbances in prepubertalchildren, we confirmed that waist circumference is asensitive marker of cardiovascular risk. This findinghighlights the potential uses of waist measurement, usingcommon anthropometric parameters, in identifying sub-groups of obese children at higher metabolic risk (37).

    In conclusion, waist circumference in prepubertal chil-dren adjusted for age, gender, and BMI is independentlyassociated with cardiovascular risk factors. Measurement ofwaist circumference may be a good choice in clinical prac-tice. Waist circumference may facilitate the detection ofindividuals with cardiovascular risk factors in childhoodbecause it is easy to measure and has a good interindividualreproducibility. Longitudinal studies performed in childrenshould verify whether, as in adults, changes in waist cir-cumference will indicate changes in cardiovascular riskfactors during growth.

    AcknowledgmentsThis study was supported by the National Research

    Council, Rome, Italy, contact no. 96.03441.CT04 and byNestle` Italiana Spa, Italy.

    References1. Owens S, Gutin B, Ferguson M, Allison J, Karp W, Le NA.

    Visceral adipose tissue and cardiovascular risk factors inobese children. J Pediatr. 1998;133:415.

    2. Goran MI, Gower BA. Relation between visceral fat anddisease risk in children and adolescents. Am J Clin Nutr.1999;70(suppl):149S56S.

    3. Freedman DS, Serdula MK, Srinivasan SR, Berenson GS.Relation of circumferences and skinfold thicknesses to lipidand insulin concentrations in children and adolescents: theBogalusa Heart Study. Am J Clin Nutr. 1999;69:30817.

    4. Daniels SR, Morrison JA, Sprecher DL, Khoury P, Kim-ball TR. Association of body fat distribution and cardiovas-cular risk factors in children and adolescents. Circulation.1999;99:5415.

    5. Goran MI, Gower BA. Abdominal obesity and cardiovascu-lar risk in children. Coron Art Dis. 1998;9:4837.

    6. Roche AF, Lohman TG, Heymsfield SB. Human Body Com-position. Champaign, IL: Human Kinetics; 1996.

    7. Lean MEJ, Han TS, Seidell JC. Impairment of health andquality of life in people with large waist circumference. Lan-cet. 1998;351:8536.

    8. Han TS, Feskens EJ, Lean ME, Seidell JC. Associations ofbody composition with type 2 diabetes mellitus. Diabet Med.1998;15:12935.

    Waist Circumference and Disease Risk, Maffeis et al.

    186 OBESITY RESEARCH Vol. 9 No. 3 March 2001

  • 9. Okosun IS, Prewitt TE, Liao Y, Cooper RS. Association ofwaist circumference with ApoB to ApoAI ratio in black andwhite Americans. Int J Obes Relat Metab Disord. 1999;23:498504.

    10. Maffeis C, Schutz Y, Piccoli R, Gonfiantini E, Pinelli L.Prevalence of obesity in children in northeast Italy. Int J ObesRelat Metab Disord. 1993;17:28794.

    11. Tanner JM. Growth at Adolescence, 2nd ed. Oxford: Black-well; 1962.

    12. Lohman TG. Applicability of body composition techniquesand constants for children and youth. Exerc Sport Sci Rev.1986;14:32557.

    13. Cole TJ, Bellizzi MC, Flegal KM, Dietz WH. Establishing astandard definition for child overweight and obesity world-wide: international survey. BMJ. 2000;320:12403.

    14. Crepaldi C, Belfiore F, Bosello O, et al. Consensus Confer-ence Italiana. Sovrappeso, Obesita` e Salute (CCI S.O.S 91).Ann Ital Med Intern. 1991;6:35975.

    15. Tanner JM, Whitehouse RH, Takaishi M. Standards frombirth to maturity for height, weight, height and weight veloc-ity: British children, 1965. Arch Dis Child. 1966;41:45495.

    16. Deeg R, Ziegenhorn J. Kinetic enzymatic method of auto-mated determination of total cholesterol in serum. Clin Chem.1983;29:1798802.

    17. Nagele U, Hagele EO, Sauer G, et al. Reagent for theenzymatic determination of serum total triacylglycerols withimproved lipolytic efficiency. J Clin Chem Clin Biochem.1984;26:17759.

    18. Bachorik PS, Lovejoy KL, Carroll MD, Johnson CL. Apo-lipoprotein B and A1 distributions in the United States, 19881991: results of the National Health and Nutrition Examina-tion Survey III (NHANES III). Clin Chem. 1997;43:236478.

    19. Friedewald WT, Levy R, Fredrickson DS. Estimation of theconcentration of low density lipoprotein cholesterol in plasmawithout use of the preparative ultracentrifuge. Clin Chem.1972;18:499502.

    20. Zar JH. Simple linear correlation. In: Biostatistical Analysis,2nd ed. Englewood Cliffs, NJ: Prentice Hall; 1984, pp. 3156.

    21. Update on the 1987 Task Force Report on High BloodPressure in Children and Adolescents. A working groupreport from the National High Blood Pressure Education Pro-gram. National High Blood Pressure Education ProgramWorking Group on Hypertension Control in Children andAdolescents. Pediatrics. 1996;98:64958.

    22. Freedman DS, Dietz WH, Srinivasan SR, Berenson GS.The relation of overweight to cardiovascular risk factorsamong children and adolescents: the Bogalusa Heart Study.Pediatrics. 1999;103:117582.

    23. de Ridder CM, de Boer RW, Seidell JC, et al. Body fatdistribution in pubertal girls quantified by magnetic resonanceimaging. Int J Obes Relat Metab Disord. 1992;16:4439.

    24. Goran MI, Gower BA, Treuth M, Nagy TR. Prediction ofintra-abdominal and subcutaneous abdominal adipose tissue inhealthy pre-pubertal children. Int J Obes Relat Metab Disord.1998;22:54958.

    25. Chan JM, Rimm EB, Colditz GA, Stampfer MJ, WillettWC. Obesity, fat distribution, and weight gain as risk factorsfor clinical diabetes in men. Diabetes Care. 1994;17:9619.

    26. Pascot A, Lemieux S, Lemieux I, et al. Age-related increasein visceral adipose tissue and body fat and the metabolic riskprofile of premenopausal women. Diabetes Care. 1999;22:14718.

    27. Rainwater DL, Mitchell BD, Comuzzie AG, Haffner SM.Relationship of low-density lipoprotein particle size and mea-sures of adiposity. Int J Obes Relat Metab Disord. 1999;23:1809.

    28. de Ridder CM, Thijssen JH, Bruning PF, Van den BrandeJL, Zonderland ML, Erich WB. Body fat mass, body fatdistribution, and pubertal development: a longitudinal study ofphysical and hormonal sexual maturation of girls. J ClinEndocrinol Metab. 1992;75:4426.

    29. Gutin B, Islam S, Manos T, Cucuzzo N, Smith C, StachuraME. Relation of percentage of body fat and maximal aerobiccapacity to risk factors for atherosclerosis and diabetes inblack and white seven- to eleven-year-old children. J Pediatr.1994;125:84752.

    30. National Institutes of Health Consensus DevelopmentPanel on Physical Activity and Cardiovascular Health.Physical activity and cardiovascular health. JAMA. 1996;276:2416.

    31. Malina RM. Tracking of physical activity and physical fitnessacross the lifespan. Res Q Exerc Sport. 1996;67:S4857.

    32. Chu NF, Rimm EB, Wang DJ, Liou HS, Shieh SM. Rela-tionship between anthropometric variables and lipid levelsamong school children: the Taipei Children Hearth Study. IntJ Obes Relat Metab Disord. 1998;22:6672.

    33. Savva SC, Tornaritis M, Savva ME, et al. Waist circumfer-ence and waist-to-height ratio are better predictors of cardio-vascular disease risk factors in children than body mass index.Int J Obes Relat Metab Disord. 2000;24:14538.

    34. Morrison JA, Sprecher DL, Barton BA, Waclawiw MA,Daniels SR. Overweight, fat patterning, and cardiovasculardisease risk factors in black and white girls: the NationalHeart, Lung, and Blood Institute Growth and Health Study.J Pediatr. 1999;135:45864.

    35. Burns TL, Mall PP, Lauer RM. The relation between pon-derosity and coronary risk factors in children and their rela-tives: the Muscatine Ponderosity Family Study. Am J Epide-miol. 1989;129:97387.

    36. Taylor RW, Jones IE, Williams SM, Goulding A. Evalua-tion of waist circumference, waist-to-hip ratio, and the conic-ity index as screening tools for high trunk fat mass, as mea-sured by dual-energy X-ray absorptiometry, in children aged319 y. Am J Clin Nutr. 2000;72:4905.

    37. Berenson GS, Srinivasan SR, Bao W, Newman WP III,Tracy RE, Wattigney WA. Association between multiplecardiovascular risk factors and atherosclerosis in children andyoung adults. The Bogalusa Heart Study. N Engl J Med.1998;338:16506.

    Waist Circumference and Disease Risk, Maffeis et al.

    OBESITY RESEARCH Vol. 9 No. 3 March 2001 187