KINANTHROPOMETRY Differences in the obesity screening ability of 19 anthropometric parameters in young Japanese females: Comparisons of direct measurements, conventional and novel indices

Aim: The present study aimed to examine the usefulness of anthropometric parameters for obesity screening in young Japanese females by assessing their associations with indicators of adiposity obtained from a dual energy x-ray absorptiometry (DXA). Methods: Screening ability of 19 anthropometric parameters was examined using a total of 50 young Japanese females who completed detailed anthropometry and a whole-body DXA scan. Anthropometric parameters were categorized into 1) measured variables, 2) conventional indices, and 3) novel indices and their correlations with body fat variables obtained from DXA were investigated. Using a percentage body fat (%BF) of 30.0% as a cut-off point of obesity, the Area Under the Curve (AUC) was observed from the Receiver Operating Characteristics (ROC) analysis and cut-off points of anthropometric parameters were determined. Results: While body mass correlated highly with total fat tissue mass in this sample (r = 0.847), body mass index (BMI) and waist circumference (WC) correlated most strongly with trunk fat and android fat tissues respectively (r = 0.820 and 0.865). However, all body composition variables were correlated with the sum of eight skinfolds (Sum8SF) if %BF was used (r ranged 0.672 – 0.834). Among anthropometric parameters examined, Ʃ8SF showed highest AUC for %BF Total , %BF Gynoid and %BF IAAT while Ʃ2SF and abdominal circumference (AbC) showed highest AUC for %BF Trunk and %BF Android respectively. Conclusion: Directly measured variables and conventional indices showed moderate to strong correlations with results from DXA. However, the sum of skinfolds, particularly Sum8SF, showed stronger correlations and superior screening ability for obesity. Although many novel indices have been utilized to screen obesity and metabolic abnormalities, observed results indicated that these indices may not necessarily better than measured values or conventional indices. Further investigations to confirm proposed cut-off points are warranted. present study aimed to investigate the usefulness of existing anthropometric parameters for obesity screening in young Japanese Results confirmed usefulness of BMI, WC, AbC and also AHtR but also showed the better screening ability of skinfolds. On the other hand, novel indices including BRI, BAI, C-Index, WHHR and ABSI were found to be weakly correlated with results obtained from DXA and had poor screening abilities. Although many novel indices have been utilized to screen obesity and metabolic abnormalities, observed results indicated that these indices may not necessarily be better than measured values or conventional indices. In order to screen at-risk individuals accurately and promote lifestyle modification, it is important to select anthropometric parameters for the target population and also for accumulation of particular region of the focus. individuals study on sample including determined


Introducción
Obesity is defined as a condition with excessive accumulation of body fat that may impair health (World Health Organization 1997). Obese individuals, particularly those with excessive accumulation of visceral and ectopic fat are known to have increased risks of metabolic abnormalities including hypertension, hyperglycaemia, hyperlipidaemia that subsequently increase risks of cardio-and celebro-vascular diseases, type II diabetes mellitus and some forms of cancers. (Basen-Engquist and Chang 2011, McMorrow et al.2015, Saponaro et al.2015, Ross et al. 2020. While obesity has been estimated to continue increasing around the globe (NCD Risk Factor Collaboration (NCD-RisC) 2016), obesity and related health problems are considered to cause adverse impacts to individuals as well as to development of countries and the global economy (Specchia et al. 2015, Okunogbe et al. 2021. Therefore, a reduction in obesity has been an important global public health challenge. Early screening and prevention strategies are key components to minimize the influence of obesity and related health problems. Since development of obesity-related diseases are not only associated with total amount of fat accumulation but also its distribution, an emphasis has been put on screening individuals with a greater health risk (The Examination Committee of Criteria for 'Obesity Disease' in Japan 2002). For this purpose, anthropometric parameters (i.e. measured variables and indices calculated from measured variables) are a common tool to screen obesity and related health conditions. Many anthropometric parameters such as body mass index (BMI), waist circumference (WC), waist-to-hip ratio (WHR), and waist-to-height ratio (WHtR) have cut-off points that indicate obesity and risk of metabolic abnormalities (World Health Organization 1997, Ashwell and Hsieh 2005, International Diabetes Federation 2006, World Health Organization 2011. In addition, a number of other circumferences (e.g. neck and arm) have been examined for their usefulness to screen overweight and obesity in both adults and adolescents (Ferretti Rde et al. 2015, Chaput et al. 2016. In recent years, more complex indices such as the conicity index (C-Index), the body roundness index (BRI), the body adiposity index (BAI), and a body shape index (ABSI) have also been proposed (Valdez et al., 1993, Bergman et al. 2011, Krakauer and Krakauer 2012, Thomas et al. 2013).
These anthropometric parameters are, however, not equally useful to all individuals and groups as relationships between metabolic abnormalities and body composition as well as relationships between body composition and anthropometric parameters vary between race/ethnicity, age, and sex (WHO expert consultation 2004, Huxley et al. 2010, Ross et al. 2020. As a result, investigation of the usefulness of anthropometric parameters of interest and appropriateness of existing cut-off points for the population of the focus will become important. In addition, compared with studies focused on relationship between body composition and common anthropometric indices (e.g. BMI, WC, and WHtR), studies investigated skinfolds and novel anthropometric parameters comprehensively in Asia are scarce (Kagawa et al. 2008b, Kagawa et al. 2010, Yamborisut et al., 2012, Liu et al. 2020. The present study therefore aimed to examine the usefulness of anthropometric parameters in young Japanese females by assessing their association with indicators of adiposity obtained from a dual energy x-ray absorptiometry (DXA).

Participants
Japanese female university students (aged between 18 and 30 years) enrolled in the School of Applied Nutrition were invited to participate in the study. Individuals with 1) genetic and chronic illnesses, 2) currently under prescribed medications that may influence body composition, 3) excessive level of radiation exposure during the past year, 4) presence of any metallic plates or bolt in the body, and 5) likelihood of being pregnant were excluded from the study. Potential participants were given verbal and written explanation on the study, including its rationale, aim, methods, associated risks, voluntary nature of their participation, their right to withdraw from the study without DOI: 10.34256/ijk2117 adverse consequences, and confidentiality of the data. Participants who agreed to join the study signed a written informed consent prior to commence assessments. Participants received an incentive of ¥2,000 upon successful completion of all assessments. The study was approved by the Human Research Ethics Committee of Kagawa Nutrition University prior to commence of recruitment process (approval number: 254).

Anthropometry
Participants underwent detailed anthropometric assessments including four basic measurements (stature, body mass, arm span, and demi-arm span), eight skinfolds, 14 circumferences, nine lengths and nine breadths. In the present study, anthropometric variables required to calculate anthropometric parameters were analyzed (i.e., Stature, body mass, eight skinfolds, seven circumferences [neck, arm, wrist, waist, abdominal, hip, mid-thigh]). All measurements except abdominal circumference were conducted according to the standard protocol of the International Society for the Advancement of Kinanthropometry (ISAK) (Stewart et al. 2012). Abdominal circumference was measured at the level of the umbilicus based on the protocol to assess metabolic syndrome in Japan (Japanese Society for the Study of Obesity 2016). Stature was measured using a stadiometer to the nearest 0.1 cm and body mass was measured using a single-frequency bioelectrical impedance analysis (SFBIA) device (Karadascan HBF-361, Omron Corp., Japan) to the nearest 0.1 kg. Skinfolds were measured using a Harpenden skinfold caliper (British Indicators Ltd., England) and measured to the nearest 0.1 mm whereas circumferences were measured using a steel tape measure (W606PM, Lufkin, the United States) to the nearest 0.1 cm. All participants were instructed to wear light clothes with no socks and shoes and measured by a Level three anthropometrist accredited by ISAK. An intra-tester technical errors of measurement (TEM) of all analyzed variables were within an internationally acceptable standard of ≤ 10% for skinfolds and ≤ 2% for the rest of the measurement sites (Gore et al.1996, Wang et al. 2000.
In the present study, a total of 19 anthropometric parameters were investigated. Parameters were divided into three categories of 1) parameters from direct measurements (body mass, six circumferences [neck, arm, waist, abdominal, hip, and mid-thigh]

Body composition assessment
A whole-body scan was conducted on participants using a dual energy x-ray absorptiometry (DXA: Lunar iDXA with enCORE 2011, version 13.60.033, GE Medical Systems Lunar, Madison, WI, USA). Participants were instructed to remove all jewelries and metals prior to the measurement and positioned supine on a DXA bed by a trained operator. Scanned images were divided into body regions according to the manufacture's instruction and lean mass, fat mass, bone mineral content and bone mineral density of total body as well as different regions were determined. In addition, regional percentage body fat (%BF) of total body (%BFTotal) was calculated using an equation fat tissue/(fat tissue + lean tissue + bone tissue) x 100. Along with %BFTotal, %BF of the trunk (%BFTrunk), the android (%BFAndroid), the gynoid and (%BFGynoid) regions were determined. Furthermore, based on a previous study that indicated association with the intra-abdominal adipose tissue (IAAT), %BF of specific region of interest (ROI) from the bottom of ribs to the top of iliac crest (%BFIAAT) was determined (Hill et al. 2007).

Statistical Analysis
Of 68 female students recruited, 51 participants completed both anthropometry and whole-body DXA scan and one participant was excluded due to an embedded bolt in the body, resulting in 50 participants for analysis. Normality of data was examined using the Shapiro-Wilk test. Variables were expressed in both mean and standard deviation as well as median and percentiles. Pearson correlation coefficients were calculated to assess relationship between anthropometric parameters and absolute and percentage fat values obtained from DXA. The receiver operating characteristics (ROC) analysis was conducted and the area under the curve (AUC) with sensitivity and specificity for each anthropometric parameter were determined. Based on past studies that suggest 30% ≤ as a cut-off point for obesity for adult women (Huenemann et al., 1966, Wilmore et al., 1986, Bray, 1993, optimal cut-off points of anthropometric parameters for each %BF variable (i.e. %BFTotal, %BFTrunk, %BFAndroid, %BFGynoid, and %BFIAAT) were determined using the Index of Union (IU: defined as (|Sensitivity -AUC| + |Specificity -AUC|)) (Unal, 2017). All statistical analysis was conducted using SPSS statistical package (version 27.0, IBM, Tokyo) and significance level of 5% was applied unless otherwise stated. The force the matter exerts in a standard gravitational field. (Stewart et al., 2012) Neck circumference (cm) [NC] The circumference of the neck immediately superior to the thyroid cartilage, and perpendicular to the long axis of the neck. (Stewart et al., 2012) Arm circumference (cm) [AC] The circumference of the arm at the level of the Mid-acromiale-radiale site, perpendicular to the long axis of the arm. (Stewart et al., 2012) Waist The circumference of the abdomen at its narrowest point between the lower costal (10th rib) border and the top of the iliac crest, perpendicular to the long axis of the trunk. (Stewart et al., 2012) Abdominal The circumference of the abdomen at the level of umbilicus, perpendicular to the long axis of the trunk.
(Japanese Society for the Study of Obesity, 2016) Hip circumference (cm) [HC] The circumference of the buttocks at the level of their greatest posterior protuberance, perpendicular to the long axis of the trunk. (Stewart et al., 2012) Mid-thigh circumference (cm) [MTC] The circumference of the thigh measured at the level of the Mid-trochanterion-tibiale laterale site, perpendicular to its long axis. (Stewart et al., 2012)

Results
Average age, stature, and body mass of participants were 19.4 ± 1.5 years old, 159.3 ± 5.2 cm, and 52.3 ± 5.8 kg respectively (Table 2). A whole-body DXA scan revealed that participants had approximately 35 kg of lean mass and 15 kg of fat mass. When expressed in percentages, participants had an average %BFTotal of 27.4 ± 4.8% on average. However, a proportion of fat accumulation was different between the region of interest and %BFGynoid was higher than the other regions examined (%BFTrunk: 27.4 ± 6.1%, %BFAndroid: 29.7 ± 8.0%, %BFGynoid, 38.0 ± 4.2%, %BFIAAT: 23.0 ± 6.5%). A total of 19 anthropometric parameters were examined for their usefulness as an obesity screening tool. Table 3 shows results of anthropometric parameters. Their correlations with results from DXA which expressed in tissue mass and percentages were shown in Table 4. When total and regional tissue masses were used as indicators of adiposity (Table 4a), body mass was found to be the best anthropometric parameter for total and gynoid fat tissues (r = 0.847 and r = 0.864 respectively, p<0.01). On the other hand, trunk fat tissue was most strongly correlated with BMI (r = 0.820, p<0.01), followed by arm circumference (r = 0.811, p<0.01) and Sum8SF (r = 0.806, p<0.01). Waist circumference (WC) was the best parameter for fat accumulations of the android region (r = 0.865, p<0.01) and also for the IAAT (r = 0.776, p<0.01).

DOI: 10.34256/ijk2117
Compared with conventional indices (i.e. BMI, WHR, WHtR, AHtR, and WHI), correlations of novel indices were lower and both C-Index and WHHR showed weak to moderate correlations with the android fat tissues and IAAT. In the case of ABSI, a negative correlation with the gynoid fat tissue was observed. When percentage values were used (Table 4b), Sum8SF was found to be the anthropometric parameter that most strongly correlated with all variables obtained from DXA (r ranged between 0.672 and 0.834, p<0.01). Other direct measurements except neck circumference (NC) and conventional indices except WHR showed weak to moderate correlations. Novel indices also showed weak to moderate correlations with DXA results but ABSI did not significantly correlate with any of DXA results.

DOI: 10.34256/ijk2117
Discussion Anthropometry is a cost-effective, non-invasive, simple and convenient technique that has been utilized for assessments of growth, nutritional status, disease and metabolic risks. To date, a wide range of anthropometric parameters have been proposed as useful screening tools for obesity and metabolic abnormalities. Among anthropometric parameters examined, common indices such as BMI, WC, and AbC showed moderate to strong correlations with DXA results. Although r values vary by correlations with fat tissue mass or percentages, BMI was generally higher in correlations than WC and AbC except for the android region. This may indicate that BMI reflects overall adiposity while WC and AbC reflect abdominal fat accumulation, which partially support current screening process being practiced around the world (World Health Organization 1997, The Examination Committee of Criteria for 'Obesity Disease' in Japan 2002, International Diabetes Federation 2006. However, AUC showed that AHtR was higher than WC and AbC in most cases, suggesting better screening ability of AHtR over WC or AbC. The observed differences may partially due to using %BF values to determine AUC as correlations of WC and AbC with %BF variables were lower than those obtained from fat tissue mass. While WC and AbC have been commonly applied to determine metabolic risks internationally (International Diabetes Federation, 2006, Ross et al., 2020, application of AHtR may be a better option if screening is based on %BF. Cut-off points proposed from BMI and AHtR ranged from 19.7 kg/m 2 to 21.4 kg/m 2 and 0.44 to 0.48 respectively, depending on the region of focus. While these values were lower than internationally recognized cut-off points (World Health Organization 1997, Ashwell andHsieh 2005), cut-off points for %BFTotal was consistent to previous study obtained from Japanese females (BMI: 21.1 kg/m 2 vs 21.9 kg/m 2 , AHtR: 0.47 vs 0.49) (Kagawa et al. 2006, Kagawa et al., 2008b. This may reflect index-%BF relationships specific to this population and therefore validity of proposed cut-off values than the universal cut-off points proposed for all people. Although the study indicated the usefulness of BMI and AHtR, it is important to acknowledge limitations of these indices. BMI is only an index of "heaviness" and unable to distinguish body composition of individuals even with the use of a population-specific cut-off point. For AHtR, definition of "waist" or "abdominal" circumference will become an issue. In the present study, AHtR and WHtR were treated separately as measurement sites of WC and AbC were different. However, it is common to treat AbC as one way of measuring "waist" (International Diabetes Federation 2006) as different definitions for WC exist (World Health Organization 2011, Stewart et al. 2012). The present study showed that WC and AbC as well as related indices have different correlations with body fat and hence cause differences in usefulness as obesity screening tools depending on the definition used. Since "waist" is a common measurement site as well as an important site to identify abdominal, especially a visceral fat accumulation. In the present study, seven out of 10 indices categorized either as "conventional" or "novel" indices used WC in their calculation. Despite the possibility of AHtR or WHtR as useful screening tool for obesity, it is important to understand the impacts of technical factors on the screening ability of anthropometric parameters. In the case of waist-related indices, it is important to first identify the definition of "waist" that yields optimum screening ability of the parameter, and standardize both definition and measurement protocols of "waist" in order to allow comparisons between studies.
Alternative to BMI and AHtR, this study showed the usefulness of the sum of skinfolds. Ʃ8SF was found to be the best anthropometric parameter to identify excessive fat accumulations of the whole-body (%BFTotal) as well as two body regions (%BFGynoid, and %BFIAAT) while Ʃ2SF was found to be the best parameter to screen fat accumulation of the trunk region (%BFTrunk). Ʃ2SF and Ʃ8SF were both highly corelated with fat tissues than WHR and WHtR and highly correlated with %BF than body mass, BMI, WHtR, WHR, and WC, which were partially consistent to a study by Boeke and colleagues (2013). Also the Ʃ2SF cut-off point established from the study was very close to previously reported cut-off points for the comparable population (32.7 mm vs 35.0 mm) (Kagawa et al. 2010), suggesting its validity. While the number of studies that apply sum of skinfolds for obesity screening is limited compared to those using BMI, WHtR and WC, findings from the present study emphasize usefulness of skinfolds in obesity screening and frequent application should be considered.
In recent years, AC has been proposed as a potential parameter for obesity (Chaput et al. 2016). Although its screening ability for android fat and IAAT was not as high as AHtR or AbC, AC showed high AUC for %BFTotal and %BFTrunk, following sum of skinfolds. Considering its simplicity and easiness compared with skinfolds, AC has potential to assess obesity risk especially at a situation where weighing scale (or stadiometer) is not available.
Compared with directly measured anthropometric variables and conventional indices, novel indices (i.e. BRI, BAI, C-Index, WHHR and ABSI) showed weaker correlations with results from DXA. Compared with a study of similar sample size, correlation of BRI was stronger while BAI and ABSI were comparable although ABSI negatively correlated in the current study (Yang et al. 2020). From these results, despite being proposed as better screening tools, novel indices may be no more useful for obesity screening than directly measured or conventional indices.

Conclusion
The present study aimed to investigate the usefulness of existing anthropometric parameters for obesity screening in young Japanese females. Results confirmed usefulness of BMI, WC, AbC and also AHtR but also showed the better screening ability of skinfolds. On the other hand, novel indices including BRI, BAI, C-Index, WHHR and ABSI were found to be weakly correlated with results obtained from DXA and had poor screening abilities. Although many novel indices have been utilized to screen obesity and metabolic abnormalities, observed results indicated that these indices may not necessarily be better than measured values or conventional indices. In order to screen at-risk individuals accurately and promote lifestyle modification, it is important to select appropriate anthropometric parameters for the target population and also for fat accumulation of particular region of the focus.
This was the first study to investigate a wide range of anthropometric parameters for their usefulness for obesity screening comprehensively. Results may be useful when considering appropriate anthropometric parameters to assess obesity status among young Japanese females whose "masked" obesity (i.e. obese individuals with normal body mass) is common and ordinary screening approach may be missed. However, it is important to acknowledge the limitations of the present study. The study was conducted on a relatively small sample of young Japanese females. To generalize the results, the study needs to be replicated using a larger sample size including a wider age range. In addition, the present study determined cut-off points of each anthropometric parameter using the Index of Union (UI), was a new approach proposed as a superior to previously known methods such as the Youden Index and the Concordance Probability method (Unal 2017). Since previous studies that proposed cut-off points of anthropometric parameters used different analysis techniques, future research to confirm the findings are warranted.