Validation of predictive equations against DXA for estimation of body fat composition in Vietnamese children

 MedPharmRes, 2020, 4 11 
*Address correspondence to Thanh V. Kim at Department of Epidemiology, 
Faculty of Public Health, Pham Ngoc Thach University of Medicine, Ho Chi 
Minh City 70000, Vietnam; E-mails: thanhkv@pnt.edu.vn. 
DOI: 10.32895/UMP.MPR.4.2.2 
 © 2020 MedPharmRes 
MedPharmRes 
journal of University of Medicine and Pharmacy at Ho Chi Minh City 
homepage:  and  
Original article 
Validation of predictive equations against DXA for estimation of body fat 
composition in Vietnamese children 
Thanh V. Kima*, Tam M. Doa, Thanh T.K. Trana, Xuan M. Ngob, Hong K. Tanga 
aDepartment of Epidemiology, Faculty of Public Health, Pham Ngoc Thach University of Medicine, Ho Chi Minh City 
70000, Vietnam; 
bPham Ngoc Thach University of Medicine, Ho Chi Minh City 70000, Vietnam. 
Received February 05, 2020: Revised March 26, 2020: Accepted April 20, 2020 
Abstract: Background: Childhood overweight and obesity are becoming more pronounced in Vietnam, so an 
assessment tool of high efficiency in the community is warranted. This study sought to validate skinfold 
thickness (SFT) equations for estimation of body fatness by Goran and Slaughter against DXA to aid in 
assessing obesity. Method: A cross-sectional study was conducted on 144 healthy children (ages 6-17) who 
were conveniently sampled from schools within an urban district. Their anthropometric measurements (height, 
weight, and SFT) and DXA whole-body results were taken to record body fat percentage (BF%). Bland-Altman 
analysis and correlation between bias and body fat were employed to understand the agreement between results 
from each equation and DXA whole body. Result: BF% was 32.2 ± 7.6% (mean ± SD). 52.8% of the children 
were overweight or obese. Bland-Altman plots showed that all four SFT equations had wide limits of agreement 
(LOAs) and largely underestimated the reference BF% by up to 8.90%. Goran equation predicted better when 
BF% decreased, whereas Slaughter equations produced less bias when there was more body fat. Conclusion: 
The prevalence rate of overweight and obesity has become alarming. Besides, Goran and Slaughter equations 
cannot be used as alternatives for DXA scanning to measure body fat due to their underestimation. 
Keywords: Validation; predictive equation; body fat. 
1. INTRODUCTION 
While high-income countries have struggled against 
obesity epidemic, countries at lower end of SDI 
(socialdemographic index) levels witnessed a significant 
increase childhood obesity [1]. Notably in Vietnam, 2014-
2015 surveillance of the National Institute of Nutrition 
showed nearly one in two urban children were obese [2]. Once 
a child becomes overweight or obese, he or she is more likely 
to maintain this condition to adulthood [3, 4]. Furthermore, 
this condition may lead to chronic diseases for these children 
when they grow up [5-7]. Currently, health professionals most 
depend on BMI to diagnose this condition although concerns 
have been raised about this index’s validity [8, 9]. Hence, 
other alternatives have been developed. Until now, the use of 
adiposity for diagnosing childhood obesity is increasingly 
endorsed by contemporary literature [8-12]. 
DXA is becoming a reference method to measure body fat 
[13]. However, due to logistic and financial limitations, 
technical complexity and concerns for radiation, it is not 
feasible to use DXA widely and routinely on the field. 
Therefore, scores of equations using anthropometric 
characteristics were developed to estimate body fat 
percentage because of the usefulness of these measures in 
various settings, especially at the primary care level. In 
12 MedPharmRes, 2020, Vol. 4, No. 2 Kim et al. 
children, they include famous equations developed by 
Slaughter or Goran [14, 15]. These predictive equations were 
built upon specific populations and need validating on other 
populations before being used. So far, to our best knowledge, 
little or no previous studies have confirmed the validity of 
these equations on Vietnamese children. 
The present study primarily aimed to validate current 
predictive equations against DXA on children from 6 to 17 
years of age currently living in Ho Chi Minh city. The 
secondary aim was to describe the anthropometric 
characteristics of the same population. 
2. MATERIALS AND METHOD 
2.1. Study design and sample 
This cross-sectional study recruited 144 children of 
normal health (72 boys and 72 girls) aged 6 – 17. They were 
conveniently sampled from 5 schools (2 elementary, 2 
secondary and 1 high school) in District 10 of Ho Chi Minh 
City during the 2018 – 2019 period. The eligibility criteria 
were children from 6 – 17 years of age, not having any acute 
or chronic illnesses nor using any medications which were 
related to the alteration in body composition. 
2.2. Anthropometric measurements 
The research team came to the schools to collect 
anthropometric data according ... d [19]. 
2.6. Ethics 
The study was conducted under the ethic approval from 
Pham Ngoc Thach University of Medicine. We gave the 
parents a written consent form to sign up for participation after 
they were informed of information related to the study. 
3. RESULTS 
3.1. Participants’ characteristics 
Data from Table 1 show descriptive data from participants 
aged between 6 and 17. More than half were overweight or obese. 
Moreover, though boys had significantly lower body fat, more of 
them were obese than girls. Nearly half of the children were in 
post-puberty. 
3.2. Agreement between predicted and measured BF% 
In Table 2, the significant biases between the two methods 
show that the four SFT equations underestimated the reference 
BF%. The 95% LOAs were wide with the gap fluctuating from 
25.50% (Goran’s) to 35.78% (Slaughter’s). Although the bias 
from Goran equations was the most pronounced, its LOAs was 
the narrowest. Furthermore, the correlation degree between bias 
and average BF% from DXA and each equation was weak to 
moderate, indicating the bias significantly changed across the 
body fat continuum (Figure 1). Specifically, as for the three 
Slaughter equations, the more body fat, the lower the bias. 
Conversely, the bias from Goran equation became larger when 
children had more body fat. 
4. DISCUSSION 
In this study, we found that Goran and three Slaughter SFT 
equations underestimated the BF% measured by DXA scans. 
Also, the difference computed by Goran equation was larger 
when the child has more body fat. Conversely, with three 
Slaughter equations, the difference yielded was larger when a 
child has less body fat. Therefore, we cannot use these equations 
as a replacement for DXA. We also learnt that the proportion of 
children with overweight or obesity had gone up to 52.8%; BF% 
was 32.2 ± 7.6 %. As far as we can tell, this is the first study in 
Vietnam to validate predictive equations against DXA scans with 
the latest nutritional status data from children in Ho Chi Minh 
city. 
Validation of equations for estimating body fat in Vietnamese children MedPharmRes, 2020, Vol. 4, No. 2 13 
Table 1. Descriptive characteristics, by gender 
 Total Boys Girls 
 Mean±SD/ n (%) Mean±SD/ n (%) Mean±SD/ n (%) 
N 144 (100%) 72 (50 %) 72 (50 %) 
Age (year) 11.7 ± 3.2 12.0 ± 3.3 11.5 ± 3.1 
Weight (kg) 45.4 ± 16.0 48.7 ± 17.9 42.0 ± 13.2* 
Height (cm) 145.5 ± 16,3 147.8 ± 18.1 143.1 ± 14.0 
Waist circumference (cm) 71.6 ± 11.6 73.7 ± 12.0 69.6 ± 10.9* 
Hip circumference (cm) 82.4 ± 13.0 84.3 ± 12.5 80.3 ± 13.3 
BF% by DXA (%) 32.2 ± 7.6 30.0 ± 8.1 34.4 ± 6.3*** 
Skinfold thickness 
Triceps (mm) 17.2 ± 7.4 16.6 ± 6.6 17.7 ± 8.2 
Sub-scapular (mm) 14.6 ± 8.2 13.7 ± 5.4 15.7 ± 10.3*** 
Abdominal (mm) 19.2 ± 11.0 19.5 ± 10.0 18.9 ± 11.9 
Supra-iliac (mm) 23.2 ± 12.6 23.7 ± 12.5 22.6 ± 12.8 
Mid-thigh (mm) 23.2 ± 12.8 17.7 ± 8.5 28.7 ± 14.1*** 
Medial Calf (mm) 17.7 ± 9.3 13.5 ± 5.5 21.8 ± 10.5*** 
Overweight - Obesity † 76 (52.8 %) 46 (63.9 %) 30 (41.7 %)** 
Maturational level 
Prepubescent 55 (38.2%) 38 (52.8%) 17 (23.6%) 
Pubescent 25 (17.4%) 2 (2.8%) 23 (31.9%) 
Postpubescent 64 (44.4%) 32 (44.4%) 32 (44.5%) 
* <0.05; ** <0.01; *** <0.001; † WHO 2007 
Table 2. Biases and 95% limits of agreement between predicted and measured BF% 
Bias (%) 95% Limits of Agreement (%) 
r 
 Mean 95% CI Lower Upper 
Goran 8.90 7.90 - 10.00 -3.80 21.70 0.44* 
Slaughter 
 (Tri+Calf) 
6.02 4.60 - 7.45 -10.90 23.00 -0.31* 
Slaughter 
(Tri+SS) Whites 
5.95 4.50 - 7.40 -11.70 23.60 -0.28* 
Slaughter 
(Tri+SS) Blacks 
6.46 4.96 - 7.97 -11.43 24.35 -0.31* 
* p<0.001; Bias: BF% by DXA minuses values from SFT equations; 95% limits of agreement: ± 2 SD of the mean difference 
between two methods; r: correlation between bias and average of BF% form two methods. Abbreviation: CI: Confidence Interval; 
Tri+Calf: Triceps and Medial Calf; Tri+SS: Triceps and Subscapular. 
The first finding is the underestimation of BF% by predicted 
equations. This could be explained by the differences in 
anthropometric and ethnic characteristics. Firstly, Goran used a 
Caucasian sample to develop his equations; Slaughter’s sample 
also included Caucasian and Black children [14, 15]. A study by 
Deurenberg et al [20] showed that Asian people might have 3–
5% higher body fat compared with Caucasian people with similar 
BMIs. This can be because Asian people accumulate a greater 
amount of abdominal fat tissue [21]. Secondly, roughly one in 
two children in our sample were overweight or obese, while 
Goran’s and Slaughter’s development samples were of the 
normal weight range. Hence, bias and LOAs would become 
wider. In other studies [22-25], the underestimation was also 
reported and mean biases ranged from 2.9 to 11.1%. Only 
14 MedPharmRes, 2020, Vol. 4, No. 2 Kim et al. 
Wickramasinghe et al [26] showed that Slaughter equations for 
white children overestimated by 5.9% ± 8.4 (mean bias ± 
standard deviation). 
The wide LOAs and significant change of bias across the 
body fat continuum also described in previous studies [22-26]. 
These studies shared the same bias trends by Slaughter’s [22-26] 
and Goran’s [24, 25]. The correlation was negative between BF% 
bias from Slaughter equation and average BF% in children with 
excessive fat in Gonzalez-Ruiz’s study and type 1 diabetes 
children in Sarnblad’s study [22, 23]. However, in the healthy 
group in Samblad’s study and Wickramasinghe et al, no 
significant correlation was observed [23, 26]. 
Furthermore, the figure that 52.8% of the children were 
overweight or obese might imply the growing trend of 
overweight and obesity in urban area. In 2014, Thuy et al [27] 
found that the prevalence in urban secondary schools in Hanoi 
was 36.2%. Earlier in 2010, the figure in Phuong et al [28] were 
27.5% of children from urban secondary schools in Ho Chi Minh 
city. 
The strength of our study is that we used DXA as a reference 
method. This technique has high validity and reliability in 
measuring BF% [13]. Furthermore, the use of Bland-Altman 
analysis was more accurate than merely using correlation or t-test 
to evaluate the agreement between two methods. However, we 
find it hard to generalize our results because our sample was well-
characterised. Five conveniently-sampled schools were adjacent 
to each other and located near the city centre. Also, our sample 
was relatively small and could not be representative of children 
of Ho Chi Minh city. 
SFT equations are convenient tools to measure body 
composition on the field. Therefore, more and more research has 
been done to validate predictive equations on a specific 
population. Our results raise a need to develop new 
anthropometric equations on Vietnamese children. Also, larger-
scaled studies are warranted to determine the nutrition status of 
children in urban and in rural schools as well. 
5. CONCLUSION 
Figure 1. Comparison of BF% measured by DXA in comparison with Goran and three Slaughter equations, 
displayed as Bland-Altman plots. 
Y-axis represents bias of BF% measured by DXA minus BF% measured by the equations. X-axis represents 
average of BF% measured by DXA and the equations. Central dash lines represent the mean biases between 2 
methods. Dotted lines on the either sides represent upper and lower limits of agreement. Solid diagonal lines are 
fitted line representing the relationship between biases and average BF% by each pair of methods. Abbreviations: 
Tri+Calf: Triceps and Medial Calf; Tri+SS: Triceps and Subscapular; BF%: body fat percentage; DXA: dual 
energy X-ray absorptiometry; r: correlation between bias and average of BF% form two methods. 
Validation of equations for estimating body fat in Vietnamese children MedPharmRes, 2020, Vol. 4, No. 2 15 
We found that Slaughter and Goran equations cannot predict 
accurately BF% of children in Ho Chi Minh city and the 
development of new anthropometric equations is warranted. 
Moreover, with the preliminary results showing an alarming 
figure of over 50% of children with overweight or obese, we need 
a large-scale study to have a bigger view of the situation. 
ACKNOWLEDGEMENTS 
General: We would like to send our gratefulness to 
Chamber of Education, People Committee District 10; 
Administration Committee of Thien Ho Duong Primary 
School, Le Thi Rieng Primary School, Tran Phu Junior High 
School, Hoang Van Thu Junior High School, Nguyen Du High 
School for your cooperation and assistance throughout the 
study. 
Funding: This study was supported by Research Fund 
provided by Pham Ngoc Thach University of Medicine. 
Author Contributions: TK substantially contributes to 
conceptual design, acquisition of data, drafting the article; TD 
contributes to acquisition of data, data analysis and 
interpretation, and revising the article; TT contributes to data 
acquisition and data analysis; XN contributes to design the 
study, critically revises the article; HT substantially 
contributes to study design, critically revises the article, 
makes final approval for submission. 
Competing interests: The authors declare no conflict of 
interest. 
REFERENCES 
1. New England Journal of Medicine. Health Effects of Overweight and 
Obesity in 195 Countries over 25 Years. 2017;377(1):13-27. 
2. Cao HTT. Phòng chống Thừa cân - Béo phì ở trẻ nhỏ: National Institute 
of Nutrition. 2018. 
3. Ward ZJ, Long MW, Resch SC, Giles CM, Cradock AL, Gortmaker SL. 
Simulation of Growth Trajectories of Childhood Obesity into Adulthood. 
New England Journal of Medicine. 2017;377(22):2145-53. 
4. Singh AS, Mulder C, Twisk JW, Van Mechelen W, Chinapaw MJ. 
Tracking of childhood overweight into adulthood: a systematic review of 
the literature. Obesity Reviews. 2008;9(5):474-88. 
5. Biro FM, Wien M. Childhood obesity and adult morbidities–. The 
American journal of clinical nutrition. 2010;91(5):1499S-505S. 
6. Halfon N, Larson K, Slusser W. Associations between obesity and 
comorbid mental health, developmental, and physical health conditions 
in a nationally representative sample of US children aged 10 to 17. 
Academic pediatrics. 2013;13(1):6-13. 
7. Park MH, Falconer C, Viner Ra, Kinra S. The impact of childhood 
obesity on morbidity and mortality in adulthood: a systematic review. 
Obesity reviews. 2012;13(11):985-1000. 
8. Javed A, Jumean M, Murad MH, Okorodudu D, Kumar S, Somers VK, 
et al. Diagnostic performance of body mass index to identify obesity as 
defined by body adiposity in children and adolescents: a systematic 
review and meta-analysis. Pediatric Obesity. 2015;10(3):234-44. 
9. Romero-Corral A, Somers VK, Sierra-Johnson J, Thomas RJ, Collazo-
Clavell M, Korinek J, et al. Accuracy of body mass index in diagnosing 
obesity in the adult general population. International journal of obesity. 
2008;32(6):959. 
10. Le TT, Chu DT, Hanh NTH. Percentage Body Fat is As a Good Indicator 
for Determining Adolescents Who Are Overweight or Obese: A Cross-
Sectional Study in Vietnam. Osong public health and research 
perspectives. 2019;10(2):108. 
11. De-Lorenzo A, Bianchi A, Maroni P, Iannarelli A, Di Daniele N, 
Iacopino L, et al. Adiposity rather than BMI determines metabolic risk. 
International journal of cardiology. 2013;166(1):111-7. 
12. Ellis KJ, Abrams SA, Wong WW. Monitoring childhood obesity: 
assessment of the weight/height2 index. American Journal of 
Epidemiology. 1999;150(9):939-46. 
13. Hu FB. Measurements of adiposity and body composition. Obesity 
epidemiology. 2008;416:53-83. 
14. Slaughter MH, Lohman T, Boileau R, Horswill C, Stillman R, Van Loan 
M, et al. Skinfold equations for estimation of body fatness in children 
and youth. Human biology. 1988:709-23. 
15. Goran MI, Driscoll P, Johnson R, Nagy TR, Hunter G. Cross-calibration 
of body-composition techniques against dual-energy X-ray 
absorptiometry in young children. The American journal of clinical 
nutrition. 1996;63(3):299-305. 
16. Lohman TG, Roche AF, Martorell R. Anthropometric standardization 
reference manual: Human kinetics books Champaign; 1988. 
17. Onis MD, Onyango AW, Borghi E, Siyam A, Nishida C, Siekmann J. 
Development of a WHO growth reference for school-aged children and 
adolescents. Bulletin of the World health Organization. 2007;85:660-7. 
18. World Health Organization. Physical status: The use of and interpretation 
of anthropometry, Report of a WHO Expert Committee. 1995. 
19. Deurenberg P, Deurenberg-Yap M, Guricci S. Asians are different from 
Caucasians and from each other in their body mass index/body fat per 
cent relationship. Obesity reviews : an official journal of the International 
Association for the Study of Obesity. 2002;3(3):141-6. 
20. Lear SA, Humphries KH, Kohli S, Chockalingam A, Frohlich JJ, 
Birmingham CL. Visceral adipose tissue accumulation differs according 
to ethnic background: results of the Multicultural Community Health 
Assessment Trial (M-CHAT). Am J Clin Nutr. 2007;86(2):353-9. 
21. González-Ruíz K, Medrano M, Correa-Bautista J, García-Hermoso A, 
Prieto-Benavides D, Tordecilla-Sanders A, et al. Comparison of 
bioelectrical impedance analysis, slaughter skinfold-thickness equations, 
and dual-energy x-ray absorptiometry for estimating body fat percentage 
in colombian children and adolescents with excess of adiposity. 
Nutrients. 2018;10(8):1086. 
22. Särnblad S, Magnuson A, Ekelund U, Åman J. Body fat measurement in 
adolescent girls with type 1 diabetes: a comparison of skinfold equations 
against dual‐energy X‐ray absorptiometry. Acta Paediatrica. 
2016;105(10):1211-5. 
23. Noradilah MJ, Ang YN, Kamaruddin NA, Deurenberg P, Ismail MN, 
Poh BK. Assessing body fat of children by skinfold thickness, 
bioelectrical impedance analysis, and dual-energy x-ray absorptiometry: 
A validation study among Malay children aged 7 to 11 years. Asia Pacific 
Journal of Public Health. 2016;28(5_suppl):74S-84S. 
24. Hussain Z, Jafar T, uz Zaman M, Parveen R, Saeed F. Correlations of 
skin fold thickness and validation of prediction equations using DEXA 
as the gold standard for estimation of body fat composition in Pakistani 
children. BMJ open. 2014;4(4):e004194. 
25. Wickramasinghe VP, Lamabadusuriay SP, Cleghorn GJ, Davies PS. Use 
of skin-fold thickness in Sri Lankan children: comparison of several 
prediction equations. The Indian Journal of Pediatrics. 
2008;75(12):1237-42. 
26. Pham TTP, Matsushita Y, Dinh LTK, Do TV, Nguyen TTT, Bui AT, et 
al. Prevalence and associated factors of overweight and obesity among 
schoolchildren in Hanoi, Vietnam. BMC public health. 
2019;19(1):1478-. 
27. Nguyen PVN, Hong TK, Hoang T, Robert AR. High prevalence of 
overweight among adolescents in Ho Chi Minh City, Vietnam. BMC 
Public Health. 2013;13(1):141.