Evaluation of a Leg-to-Leg Bioimpedance Device in the Estimation of
Abdominal Visceral Fat for the Elderly
e Comparison with CT
, Hsueh-Kuan Lub
, Li-Ming Chiangc
, Jasson Chiangd
, Chung-Liang Laif*
aGeneral Education Center, National Kaohsiung University of Hospitality and Tourism, Kaohsiung, Taiwan,bSport Science Research Center, National Taiwan
University of Sport, Taichung, Taiwan,cDepartment of Hotel Restaurant Tourism Management, East Stroudsburg University, PA, United States,dDepartment
of Physical Education Sports Affairs, Chinese Culture University, Taipei, Taiwan,eResearch Center, Charder Electronic Co., LTD., Taichung, Taiwan, fDepartment of Physical Medicine and Rehabilitation, Taichung Hospital, Ministry of Health and Welfare, Taichung, Taiwan
a r t i c l e i n f o
Received 19 November 2015 Received in revised form 8 March 2016 Accepted 12 March 2016 Available online xxx Keywords: abdominal obesity, body composition, computed tomography, reliability
s u m m a r y
Background: This study aimed to assess the accuracy of the standing leg-to-leg bioelectrical impedance analyzer (LBIA) for estimating abdominal visceral fat with body mass index (BMI) scores ranging from non-overweight to obese in elderly Chinese individuals.
Methods: The abdominal visceral fat area (VFA) at the lumbar level of L4-L5 was measured for 100 elderly (age 68.5± 8.5 years) individuals using LBIA and computed tomography (CT) (respectively represented as VFALBIAand VFACT). The VFA from LBIA were compared with those from CT under various BMI levels. Results: The average BMI of the subjects was 24.3 ± 4.1 kg/m2(58 non-overweight individuals had BMIs< 25 kg/m2, and 42 overweight individuals had BMIs 25 kg/m2). The correlation coefﬁcient (r) of VFALBIAand VFACTwas r¼ 0.707 with a standard estimate of error (SEE) ¼ 21.96 cm2, and the limit of agreement (LOA) was in the range of67.46 to 58.48 cm2. The r value for non-overweight and over-weight was 0.707 (p< 0.001) and 0.356 (p < 0.001), SEE was 21.36 cm2and 20.42 cm2with LOA values ranged between68.40 to 46.00 cm2and61.79 to 71.19 cm2.
Conclusion: The LBIA results showed that each obese group all has large LOA when compared with the CT reference values. The study suggests that the LBIA accuracy for estimating VFA is limited in the Chinese elderly population.
Copyright© 2017, Taiwan Society of Geriatric Emergency & Critical Care Medicine. Published by Elsevier Taiwan LLC. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/).
Obesity has begun to replace undernutrition and infectious disease in becoming the most signiﬁcant contributor to ill health in the modern society around the globe. Obesity has become a serious problem as a result of the prevalence of cheap, processed foodﬁlled with a lot more sugar, salt, and saturated fat over time in developed countries1. Fat metabolism varies in different parts of the body. Existing research indicated that abdominal obesity was an impor-tant health index and had a positive relationship with indicators of metabolic syndrome, such as insulin resistance, hypertension, dyslipidemia, and pathoglycemia2.
The fat distributed around the abdominal skin adipose tissue considered abdominal subcutaneous fat, while the fat separating the organs considered visceral fat. Methods for estimating visceral fat, or the area of abdominal obesity, include anthropometric measurement, such as measuring the indicators of waist circum-ference, waist hip ratio (WHR), body mass index (BMI)3, and the Lange skinfold caliper4. These methods are rapid, simple and non-invasive and are often used in epidemiological studies, but they are limited for their low accuracy. To measure body and abdominal fat, the most prevalent methods recognized in the literature are CT and magnetic resonance imaging (MRI)5,6.
Compared to the aforementioned methods for measuring body fat, bioelectrical impedance analysis (BIA) may obtain electrical impedance signals detected from our physiological tissues and or-gans. This technology offers advantages in that it is rapid, easy, low-cost and non-invasive7. Additionally, a number of agreements and studies on the safety, measurement standardization,
bio-* Corresponding author. Department of Physical Medicine and Rehabilitation, Taichung Hospital, Ministry of Health and Welfare, 199, Section 1, San-Ming Road, Taichung, 403, Taiwan.
E-mail address:email@example.com(C.-L. Lai).
Contents lists available atScienceDirect
International Journal of Gerontology
jo u rn a l h o m e p a g e :w w w . i j g e - o n l i n e . c o m
1873-9598/Copyright© 2017, Taiwan Society of Geriatric Emergency & Critical Care Medicine. Published by Elsevier Taiwan LLC. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
electricity-related variables, validity, clinical applications and re-strictions of BIA have been published8. Ryo et al9applied bioelec-trical impedance technology to measure the impedance of the cross-section of the waist, thereby estimating the abdominal visceral fat area (VFA). Browning et al10performed assessments with the integrated results of both the waist circumference mea-surement and the abdominal VFA measured from the impedance of the cross-section of the waist. Shimomura et al11used the height, weight, gender, age, and the impedance between the legs as esti-mation variables; this design is the same model as the standing leg-to-leg BIA for the estimation of abdominal visceral fat. Compared to the two previous estimations of VFA by BIA, the leg-to-leg BIA is more convenient and has also been widely applied in family health care. However, only a few studies are available on the validation of the method's accuracy. The literature is scarce on the validation of the leg-to-leg BIA in the measurement of abdominal visceral area of Asians and elderly individuals12.
Based on the above considerations, we assumed that the results of VFA measurements for the elderly in Taiwan's Chinese by the leg-to-leg BIA and CT imaging would be similar, with a high correlation, no signiﬁcant systematic deviation, and a small conﬁdence interval. Therefore, the abdominal visceral fat of elderly individuals in Taiwan was measured using both the leg-to-leg BIA and CT imaging, and the results in groups of different genders and non-overweight and overweight individuals were compared to investigate whether the VFA measurement results were signiﬁcantly different between the two methods.
2. Materials and methods 2.1. Study design and subjects
The subjects of this study were selected using a non-random purposive sampling method. Middle-aged and older adults in central Taiwan, 55 years of age or older and with good mobility skills, were recruited via posters. The subjects were required to complete a thorough health questionnaire, including personal in-formation, physical characteristics, health status, and disease his-tory. All participants with endocrine, nutritional and major chronic diseases, such as diabetes mellitus, cancer, kidney dysfunction, and liver diseases, were excluded from this study; after screening, 100 subjects were included. The study procedure was approved by the ethics committee of human trials at the Nantou Tsaotun Psychiatric Center (IRB-103035) and Ta-Li Jen-AI Hospital (IBR-97-02) and was implemented in the Taichung Hospital and Ta-Li Jen-AI Hospital of the Ministry of Health and Welfare.
The body weight of the participants was measured using a Weight-Tronix (Scale Electronic Development, New York, USA) electronic scale. The height of the participants without shoes was measured using a stadiometer (Holtain, Crosswell, Wales, UK). The waist circumference (WC) was measured at a level parallel to the height of the navel, and the hip circumference (HC) was measured at the widest part of the hip using a standard measuring tape. 2.3. Four-plate standing bioelectrical impedance analyzer
A BC-305 (Tanita Corp, Tokyo, Japan) four-plate standing leg-to-leg BIA (hereinafter referred to as LBIA) was used in this study. Based on the built-in estimate equation and the height, sex, and age of the corresponding subject, the VFA at the lumbar level of L4-L5 of the subject could be estimated. With every 10 cm2as a level and with a level resolution of 0.5, the VFA values obtained are presented
as VFALBIA. Before the actual measurement using LBIA, the VFALBIA
wasﬁrst measured for ﬁve participants with an interval of three days to evaluate the reliability the LBIA measurement.
2.4. Computed tomography
A Somatorn Sensation 64 CT system (Siemens Corp., Germany) with its operating software (software version syngo CT2005A) was used to perform a CT scan on the abdominal area. To scan the monolithic image in the middle of the lumbar L4-L5 vertebral area, each subject, wearing only a cotton hood, was asked to lay on the central CT scanning platform and lifted both arms straight over his/ her head. Based on the procedures recommended by Yoshizumi et al5the abdominal visceral fat and the abdominal subcutaneous
fat area were colored in the scanned image for the area calculation, wherein the threshold CT value of adipose tissue was (260 ± 3) -(10 ± 3) Hu. The abdominal cross-sectional area (ACSA), the abdominal VFA, and the abdominal subcutaneous fat area (SFA) at the lumbar level of L4-L5 of the subject obtained by CT scanning were represented as ACSACT, VFACTand SFACT, respectively. Before
the actual CT scanning measurement, the CT scanning for the abdominal cross-sectional area at the lumbar L4-L5 level wasﬁrst performed twice for ﬁve participants over a 3-day interval to analyze the reliability of the CT measurement.
2.5. Experimental procedures
Each experiment in this study was started at 2:00 p.m. daily. Before the test, the participants fasted for 4 h, with no intense exercise for 24 h and no consumption of alcohol or diuretics for one week. After the urine in the bladder was emptied before the test, the weight, height, waist circumference, hip circumference, bioelectrical impedance analysis and CT scanning measurements were sequentially collected. All measurements for each subject were completed within 2 h. The anthropometric measurement, CT scanning, and the operation of the bio-impedance measurement in this study were performed by research assistants and radiologists. 2.6. Statistical analysis
The data analysis in this study was performed using SPSS ver. 17 (SPSS Inc., Chicago, IL, USA). The results were presented as means± SD. The correlation of VFALBIAand VFACTwas described
using Pearson's product moment correlation, and their relationship was represented with a simple linear regression equation. The Bland-Altman method13 was used to describe the difference (bias± SD) and the limit of agreement (95% conﬁdence interval, bias± 2 SD) of VFABIAand VFACT. The difference between the two
methods was compared using an independent samples t-test. Moreover, the participants were grouped according to gender and different levels of BMI, in which the group with BMI<25 kg/m2was
deﬁned as the non-overweight group and the group with BMI 25 kg/m2was deﬁned as the overweight and obese group.
In this study, a total of 100 subjects with an average age of 68.5 ± 8.5 years were included. The average BMIs were 25.1± 3.5 kg/m2for the male subjects and 23.7± 3.7 kg/m2for the
female subjects. The subjects' characteristics and measurement results are shown inTable 1. The calculated correlation coefﬁcients of VFACTand VFALBIAfor the male, female and total subjects were
0.565, 0.548, and 0.707 respectively in the non-overweight group; 0.600, 0.398, and 0.365 in the overweight group; and 0.540, 0.682, and 0.707 in the total group. The distribution of the VFACTand
VFALBIAof all subjects is shown inFig. 1, where the solid line is the
regression line: VFACT¼ 0.490 VFALBIAþ 56.054 (r2¼ 0.50, standard
estimate of error (SEE)¼ 21.960 cm2, p< 0.001), the 50% of the
VFALBIAvariation may be explained by VFACT. Further, VFACTand
VFALBIAshowed signiﬁcant difference after applying independent
sample t-test, this result may cause the accuracy of LBIA in esti-mating VFA to be limited.
The diagram showed inFig. 1 indicates a majority of female participants were distributed on the upper left while male partic-ipants distributed on the lower right of the equivalent line (y¼ x). Two groups both have its own corresponded correlation coefﬁcient, but when calculated together, they complement one another. As result, total subject group would have a larger correlation coef ﬁ-cient value than male and female groups.
The bias and the ranges of the LOA with Bland-Altman analysis for VFACT and VFALBIA of the male and female subjects were
14.4 cm2,39.4 to 68.3 cm2and21.97 cm2, and71.3 to 27.4 cm2,
respectively. The bias and the LOA ranges for VFACTand VFALBIAof
the total subjects were4.49 cm2and67.46 to 58.48 cm2. The
results are shown in Fig. 2, where the solid line represents the difference between VFALBIA and VFACT, which increased
propor-tionately with the increase of (VFACTand VFALBIA)/2 (p< 0.01).
The mean differences between VFALBIAand VFACTof the
non-overweight group and the non-overweight group were11.2 ± 28.6 and 4.7± 33.2 cm2respectively for the total subjects, 5.4± 26.1 and
23.5 ± 25.0 cm2 for the male subjects, and 21.8 ± 24.6
and20.4 ± 25.5 cm2for the female subjects, as shown inFig. 3.
Independent samples t-tests revealed signiﬁcant different BMIs for the VFALBIA and VFACT measures of the male, female and total
subjects (p< 0.001).
When evaluating the coefﬁcient of variation (CV) in LBIA, it was measured 5 times with a 15 min interval between each measure-ment per day. The within-day coefﬁcient of variation (CV; 100 [SD/mean]) for VFALBIAwas 0.13e0.25%. The corresponding
between-day CV was 0.32e0.53%. The correlation coefﬁcient r and Cronbach's alpha of the VFACTmeasured by CT scanning fromﬁve
participants were both 0.99. 4. Discussion
Using CT imaging to determine the nature of the tissue and to accurately calculate the abdominal visceral fat and abdominal subcutaneous fat with related software has shown a high degree of repeatability and is now considered the gold standard for the detection of fat distribution and content14. Compared to the tradi-tional supine posture hand-to-leg BIA, the estimation of body composition by standing BIA is not only safe, convenient, and non-invasive, but widely applied in clinical practice, research, and per-sonal health monitoring15. However, its built-in VFA estimation
equation is proprietary to the manufacturer and is not available for appropriate comparisons; thus, this study of the assessment and veriﬁcation of commercial BIA measurements is particularly important.
In the present study, the abdominal VFACTof the female subjects
was lower than that of the male subjects, and the SFACTof the male
subjects was lower than that of the female subjects. Theseﬁndings are similar to the VFA results measured by MRI in a study by Browning et al10for the elderly and a study by Wang et al16for adults. When the VFA is greater than the 100 cm2cut-off point, the risk for coronary artery disease and diabetes mellitus is
Descriptive characteristics of the subjects, divided by BMI group.
Male Female Total
BMI< 25 BMI 25 Total BMI< 25 BMI 25 Total BMI< 25 BMI 25 Total n¼ 24 n¼ 24 n¼ 48 n¼ 34 n¼ 18 n¼ 52 n¼ 58 n¼ 42 n¼ 100 Age (years) 77.0± 7.4 65.5± 6.4 71.2± 9.0 65.8± 7.5 66.62± 7.3 66.1± 7.4 70.4± 9.3 66.0± 6.7 68.5± 8.5 BMI (kg/m2) 22.4± 2.0 27.8± 2.2 25.1± 3.5 21.5± 1.7 27.8± 2.7 23.7± 3.7 21.9± 1.9 27.8± 2.4 24.4± 3.6 WHR 0.90± 0.02 0.95± 0.04 0.93± 0.04 0.84± 0.06 0.89± 0.05 0.87± 0.06 0.86± 0.06 0.93± 0.05 0.89± 0.06 ACSACT(cm2) 478.4± 85.6 638.1± 93.4 558.2± 119.8 454.8± 115.7 632.0± 101.9 516.1± 139.2 464.5± 104.2 35.5± 95.9 536.3± 131.3 SFACT(cm2) 93.7± 36.8 182.6± 72.9 138.2± 72.6 154.7± 54.3 267.6± 65.7 193.8± 79.3 129.5± 56.3 219.0± 81.1 167.1± 80.7 VFACT(cm2) 111.1± 24.0 129.3± 20.6 120.2± 23.9 79.5± 26.9 116.0± 21.7 92.5± 30.8 92.6± 30.0 124.0± 21.7 105.8± 30.9 VFALBIA(cm2) 116.5± 31.7 152.8± 15.7 134.6± 30.8 56.7± 24.6 96.7± 24.6 70.5± 31.0 81.4± 40.5 128.8± 34.4 101.3± 44.5
WHR: Waist-hip ratio; ACSA, abdominal cross-sectional area; SFA, subcutaneous fat area; VFA, visceral fat area; CT, computed tomography; LBIA: standing leg-to-leg bioelectrical impedance analysis; Data are presented as means± SD.
Fig. 1. Regression analysis and scatterplot between VFALBIAand VFACTfor males and
females. The lines of identity and regression are shown in theﬁgure.
Fig. 2. Bland-Altman plots of agreement in VFA between CT and LBIA. The linear line represents the regression line (y¼ 0.421 x e 48.054, r ¼ 0.477, p < 0.05).
substantially increased17. In this study, the average VFACTof the
subjects was 105.8± 30.9 cm2. However, the average BMI of the
subjects in our study was 24.4± 3.6 kg/m2, which did not reach the
overweight standard of the World Health Organization (WHO). Based on the VFA alone, the risk for type 2 diabetes mellitus was relatively high. The BMI of the male subjects in the non-overweight group was only 22.4± 2.0 kg/m2, but the VFA was 111.1± 24.0 cm2,
while the BMI of the total group of male subjects was 25.1 cm2, and
the VFA was as high as 120.2± 23.9 cm2. Although the
determi-nation of the obesity level using BMI is simple and easy, the results do not correlate with gender and age; therefore, this measurement is applicable for general and trend assessments of a population with a large sample scale and is not appropriate to estimate individual abdominal obesity. To further assess the risk of abdominal obesity on cardiovascular disease or type 2 diabetes, particularly among the elderly, VFA also provides a reference value. However, for the fe-male subjects of this study in both the non-overweight and
overweight groups, VFALBIA calculations were undervalued by
21.8± 24.6 and 20.4 ± 25.5 cm2compared to VFA
underes-timation in the VFA of the subjects using LBIA may result in un-derestimations in the risk of cardiovascular disease or type 2 diabetes.
Several studies have been reported on the validation of the standing leg-to-leg BIA for VFA. For example, a study by Wang et al16 showed that the correlation coefﬁcient, bias, and LOA of VFAMRI(VFA obtained by MRI) and VFABIA using a Tanita BC-532
device (Tanita, Tokyo, Japan) for adults were 0.77, 40 cm2, and6.0 to 87.0 cm2, respectively. Bosy-Westphal et al18estimated
VFA using the same BIA device, and the correlation coefﬁcient, bias, and LOA with VFAMRIwere 0.83, 2.7 cm2, and24.8 to 54.5 cm2,
respectively. Subjects average age in Wang et al16 and
Bosy-Westphal et al18were 48 and 45 year old adults while the current study mainly focuses on the older adults segment (68.5 ± 7.4). Further, different BIA brand, model, and methodology could affect the correlation coefﬁcient value and reliability analysis. Based on the above considerations, the accuracy and reliability analysis needed to be reported separately by different age groups.
Although the estimation of VFA using a standing leg-to-leg BIA is convenient and safe, the available literature and the results of this study indicated that the correlation coefﬁcient of VFALBIA with
VFACTor VFAMRIreﬂects a high correlation between these values
(r> 0.7), but the LOA was almost always greater than 80 cm2,
indicating the restriction of its application in clinical practice, research and individual health care. In this study, the comparison of the VFA measurement results by CT and by LBIA showed that, regardless of gender or obesity level, the measurement results of the two methods were signiﬁcantly different.
In addition to the standing leg-to-leg BIA for VFA estimation, standing hand-to-leg BIA is also available for the estimation of VFA. A study by Wang et al16using an Omron HBF-359 device (Omron, Kyoto, Japan) found that the correlation coefﬁcient, bias, and LOA of VFALBIAand VFAMRI were 0.84, 39.0 cm2, and 2.0 to 81.0 cm2,
respectively. A study by Bosy-Westphal et al18using an Omron HBF-500 device (Omron, Medizintechnik, Mannheim, Germany) found that the correlation coefﬁcient, bias, and LOA of VFALBIAand VFAMRI
were 0.92, 1.9 cm2, and34.3 to 38.1 cm2, respectively. Compared
to the standing leg-to-leg BIA device, the standing hand-to-leg BIA device showed a higher correlation with a smaller LOA range in the VFA estimation. This variation might have occurred because the accuracies of the VFA estimate models from different manufac-turers were different; moreover, one of the actual factors for this difference is that the impedance value derived using the leg-to-leg model did not pass through the abdomen, while that derived using the hand-to-leg model did. Therefore, in estimating abdominal visceral fat, the performance of the hand-to-leg model may be better, though the reason for this difference remains to be further clariﬁed. For the limited accuracy in VFA estimations by LBIA dis-cussed in this study, the existing simple VFA measurement method is still based on the more relevant variables in anthropometric measurements, such as waist circumference, age and abdominal skinfold thickness19, to estimate the abdominal visceral fat, as these measures are convenient and somewhat accurate.
This study's subjects were limited to the Chinese elderly in Taiwan, with no consideration for other populations. Additionally, subjects with abnormal body water ratios caused by bad behavior, severe obesity, or other diseases were not included in this study. In addition, for the investigation of abdominal VFA, this and most other studies have conducted monolithic scans at the lumbar L4-L5 level. However, a study by Shen et al20found that the VFA and the total volume of the visceral adipose tissue measured by the monolithic image at 5e10 cm above the L4-L5 lumbar spine showed the highest relevance. Thus, the optimal anatomical
Fig. 3. VFA-dependent bias of bioelectrical electrical analysis (VFABIA) compared with
CT (VFACT) in (a) males and females combined, (b) males, and (c) females. Data are
presented as the mean differences± SD. Means with difference symbols are signiﬁ-cantly different, p< 0.001** (one-way ANOVA).
position of VFA measurement for Chinese elderly individuals by monolithic imaging needs to be further explored.
In summary, the results of abdominal VFA assessment obtained using a four-plate standing bioelectrical impedance analyzer and CT scanning in Chinese elderly individuals were highly positive correlated, but the LOA range was large. Interpretations and ap-plications of the estimate results of individual's VFA by leg-to-leg BIA needs to be made with care.
Conﬂict of interests
The authors declare no conﬂict of interest. References
1. Kopelman PG. Obesity as a medical problem. Nature. 2000;404:635e643. 2. Despres JP, Lemieux I. Abdominal obesity and metabolic syndrome. Nature.
3. Molarius A, Seidell JC. Selection of anthropometric indicators for classiﬁcation of abdominal fatnesse a critical review. Int J Obes Relat Metab Disord. 1998;22: 719e727.
4. Bonora E, Micciolo R, Ghiatas AA, et al. Is it possible to derive a reliable estimate of human visceral and subcutaneous abdominal adipose tissue from simple anthropometric measurements? Metabolism. 1995;44:1617e1725.
5. Yoshizumi T, Nakamura T, Yamane M, et al. Abdominal fat: standardized technique for measurement at CT. Radiology. 1999;211:283e286.
6. Goodwin K, Syme C, Abrahamowicz M, et al. Routine clinical measures of adiposity as predictors of visceral fat in adolescence: a population-based magnetic resonance imaging study. PLOS One. 2013;8. e79896.
7. Kyle UG, Bosaeus I, De Lorenzo AD, et al. Bioelectrical impedance analysise part I: review of principles and methods. Clin Nutr. 2004;23:1226e1243. 8. Bioelectrical impedance analysis in body composition measurement. Proceedings
of a National Institutes of Health Technology Assessment Conference. Bethesda, Maryland, December 12e14, 1994. Am J Clin Nutr. 1996;64:387Se532S. 9. Ryo M, MaedA K, Onda T, et al. A new simple method for the measurement of
visceral fat accumulation by bioelectrical impedance. Diabetes Care. 2005;28: 451e453.
10. Browning LM, Mugridge O, Chatﬁeld MD, et al. Validity of a new abdominal bioelectrical impedance device to measure abdominal and visceral fat: com-parison with MRI. Obesity. 2010;18:2385e2391.
11. Shimomura M, Kodama M, Sato H. Method and System for Estimating Visceral Fat Area. U.S. Patent No. 6930, 352 B2, Jul, 2005.
12. Eickemberg M, Oliverira CC, Roriz AK, et al. Bioelectrical impedance and visceral fat: a comparison with computed tomography in adults and elderly. Arq Bras Endoc Rinol Metabol. 2013;57:27e32.
13. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;8476:307e310. 14. Zhao B, Colville J, Kalaigian J, et al. Automated quantiﬁcation of body fat
dis-tribution on volumetric computed tomography. J Comput Assist Tomogr. 2006;30:777e783.
15. Kyle UG, Bosaeus I, Lorenzo AD, et al. Bioelectrical impedance analysise part II: utilization in clinical practice. Clin Nutr. 2004;23:1430e1453.
16. Wang JG, Zhang YI, Chen HE, et al. Comparison of two bioelectrical impedance analysis devices with dual energy X-ray absorptiometry and magnetic reso-nance imaging in the estimation of body composition. J Strength Cond Res. 2013;27:236e243.
17. Okauchi Y, Nishizawa H, Funahashi T, et al. Reduction of visceral fat is asso-ciated with decrease in the number of metabolic risk factors in Japanese men. Diabetes Care. 2007;30:2392e2394.
18. Bosy-Westphal A, Later W, Hitze B, et al. Accuracy of bioelectrical impedance consumer devices for measurement of body composition in comparison to whole body magnetic resonance imaging and dual X-ray absorptiometry. Obes Facts. 2008;1:319e324.
19. Chen CH, Chen YY, Chuang CL, et al. The study of anthropometric estimates in the visceral fat of healthy individuals. Nutr J. 2014;13:46.
20. Shen W, Punyanita M, Wang Z, et al. Visceral adipose tissue: relations between single-slice areas and total volume. Am J Clin Nutr. 2004;80:271e278.