Trunk Fat Estimation According To Gender In Childhood Through Basic Somatic Readings. An Opportunity for Improving Girl’S Evaluation

Background: Easy appraisal of trunk fat and the eventual cardiometabolic risk of pediatric obesity is a present requirement. Its estimators Waist-to-Height Ratio (WHtR) and Relative Fat Mass pediatric (RFMp, expressed as % of total fat content), can be a complement of widely accepted BMI. Methods: 472 children (39% boys), classied as underweight, normal weight, overweight and obese (Nutritional groups, NG) according to BMI Z-score after their strictly obtained anthropometric initial data, with ad hoc exclusion criteria. Calculation of WHtR and RFMp was done for each nutritional group and sex, as associations through multiple linear regression (MLR) and differences among sexes (medians, IQR). Results: Age (mean (95% CI)), 10.8 y (10.1- 11.1). Estimators values in: All group WHtR 0.5 (0.49 – 0.51), RFMp% 32.3 (31.7 – 33.0); Underweight, WHtR 0.41 (0.40 – 0.41), RFMp (%) 22.8 (21.2 – 24.4); Normal weight, WHtR 0.46 (0.45 – 0.47), RFMp(%) 28.9 (28.1 – 29.7); Overweight, WHtR 0.51 (0.50 – 0.52), RFMp(%) 34.2 (33.3 – 35.1) and Obese WHtR 0.56 (0.55 – 0.57), RFMp(%) 37.8 (36.9 – 38.6). Associations (NG; independent variables): In NG adjusted R² values were between 0.74-0.78. In All group Beta coecients were for RFMp 3.36 (P< 0.001) for girls; for waist circumference (WC 2.97 (P< 0.001) and for WHtR were respectively -0.01 (p< 0.001) and 0.03 (p< 0.001) Sex differences: BMI were in NG without gender differences (Mann-W’s U). (median (M a primary categorizer underweight, or from the whole and signicant levels in in all underweights (p <0.004). (trunk) in the All group


Introduction
Excessive abdominal fat deposit is associated to the obesity comorbidities whether in adult (1)(2)(3) or in children and adolescents (4,5). At this age the prime consequences are the elevated blood pressure or subtle metabolic disturbances out of the numerous and coexistent clinical deviations that may appear inconspicuously at these early moments. Body mass index whether expressed as Z-score or percentage ways (BMI Zs, BMI %) are widely accepted to identify malnutrition in pediatric ages but are not enough to alert about present or future cardiometabolic risks in overweight or obese children; in contrast in adults waist circumference readings are accepted to be favoured (6).Therefore simple tests based on waist circumference have shown a reasonable association to cardiovascular risks in children (7) and adolescents (8) and were associated with accurate estimations of fat mass percentage (9).
Prevalence of abdominal obesity is not very well known in pediatric ages, rst because of different methods (and names) for assessing it, probably the most common: Dual X-ray Absorptiometry, DXA, (central fat' or 'trunk fat'. Computed tomography and magnetic resonance imaging are considered the gold standard for quantitative measurement of abdominal adipose tissue compartments (visceral subcutaneous…) although less frequently used due to small but signi cant radiation and cost. Anthropometry is considered the basic and straightforward method, the pure waist circumference, apart from the far more commonly assessed BMI are measured in each visit. Furthermore, abdominal fat increases as the child grows. After vast experiences in adult (10,11) evidencing abdominal obesity association with and/or predictive capacity for cardiometabolic conditions, waist circumference percentiles (12,13) and diverse derived equations appeared in pediatric obesity preventive literature. Out of these, waist-to-height ratio (WHtR) and relative fat mass pediatric (RFMp) were selected for the present study.
WHtR (14) is associated with most health risks occurring in obese adult when this ratio is greater than 0.5 even in subjects identi ed as normal weight. The value ≥ 0.5 has also been accepted for children and adolescents for estimation of abdominal obesity. RFM derives from a large study on adults (15) that has led to (16) RFM pediatric (RFMp) which is also an estimator of fat mass percentage based on the height/waist proportion with the presumed advantage of considering genders separately and has been in close agreement to DXA measurements. Therefore, quickly facilitating an idea of the fat content not only at initial diagnosis but also at the normally long follow up of obesity.
The aim of this study is to nd out whether these clinical matched estimators can add complementary information about the (trunk) fat content to the globally recognized BMI in different nutritional statuses underweight, normal weight, overweight and obese children and especially for signaling the fat differences thresholds for gender.

Methods
Study design. Secondary analysis of exclusively initial diagnostic measures from patients attending to the Pediatric Institute for Nutrition, Growth and Metabolism Clinical Unit attached to this University Hospital. Participants: 472 (185 boys) children and adolescents aged 10.8 years (95% CI 10.1-11.1) classi ed into nutritional groups according to their BMI Zs respectively as underweight (UW, < -1SD), normal weight (NW, -1 to + 1SD), overweight (OW, + 1 to + 2SD) and obese (OB, > 2SD). Age and sex subgroups have also been studied. This baseline data came from patients attending the Unit between 2014 and 2018. All of them were followed and cared under the directions of this Unit. Children with incomplete somatic data, low school performance, atypical social status and chronic conditions were excluded (n = 27).
Intervention: According to the established rules of the Unit anthropometric measures were taken by specialized and speci c personnel for height (Harpenden stadiometer), weight (electronic scales), waist circumference (inextensible tape) and blood pressure (Dinamap, two pressure cuffs). For waist circumference (WC) we followed the recommendations of WHO (17) but with the pediatric precautions, standing position, tape horizontally located at the midpoint between lower costal margin and upper anterior iliac apophysis. The tape not too tight nor too loose and reading at the next 0.1 cm at the end of exhalation, but before recording the result checking how the centimeter readings changes (left to right) with respiratory movements. All measures have been taken in acclimatized room where children were in light underwear and barefoot, always in the morning and after a light continental breakfast. For height and WC all readings were in centimeter and centile and Z score obtained through the anthropometric program based on IOTF standards. BMI % was also in the same program after the Poskitt de nition (18). Target height: midparental height ± 6.5 cm, whether boys or girls. WHtR is a unitless ratio. RFMp was calculated according to pediatric equations ( Ethics. The applied procedures were in accordance with the standards of the institutional Ethics Committee and with the Helsinki Declarations (1964;2000). The study was approved by our Institution Ethical Committee.
Statistical analysis. A descriptive analysis has been made using mean and standard deviation, or 95% con dence interval, or median and interquartile ranges (IQR) for quantitative variables according to previously done normality test. For categorical data, absolute values and percentages were applied. Univariate study by Mann-Whitney U test for assessing gender differences in estimated fat percentages, was also applied to dependent and independent variables just enumerated below. Correlation studies for the linear relationship among the dependent and continuous covariates. A multivariate study for assessing the relationship of covariates: sex, BMI Zs, WC Zs, birth weight (BW Zs), systolic and diastolic blood pressure with dependent (WHtR and RFMp) variables. Next B(Beta) coe cient was obtained for each dependent variable in ve models, four according to every BMI group and the fth comprising all participants without any strati cation. Stata Biostatistical Program, SSS release 15, 2017, was used and P-value < 0.05 was considered as signi cant. The ndings of this study should be considered as exploratory and/ or descriptive. Table 1 exhibits the characteristics of the four groups of children plus an All group including the whole sample, which was included because of the eventual association of estimators with a wider range of BMI. The anthropometric characteristics of groups were in accordance with their BMI thresholds for under, normal, overweight and obesity.

Results
In table 2, the association of trunk fat estimators with the main independent variables appears in the four nutritional groups plus the whole sample (All group). First, adjusted coe cient of determination (aR², table rst columns) will be considered, which in the case of WHtR the aR² values explain its association with the BMIs that characterize each group, ranging from 0.88 in All group to 0.007 in the Underweight group. When considering aR² for RFMp the association in all studied groups ranged from 0.87 in All group to 0.57 in the Underweight group signaling a certain advantage for this latter estimator, although both explain nearly 88% of the outcomes. Second (multivariate study, subsequent columns), as regards B coe cient as clearly appears in the case of WHtR values, the association level to the six variables analyzed in each nutritional group showed lower levels indicating a lesser association of variables specially birth weight, systolic and diastolic blood pressure. As expected in RFMp, B coe cient had high values for waist circumference and sex in the four nutritional groups and All group, ranging from 3.77 in the Obese to 4.58 in Underweight group and 3.36 in All group, signaling the association of these greater values for girls whatever nutritional group they belonged, thus pointing out a signi cative association for trunk fat.
Standard correlation matrix: in Normal weight group the correlation of BMI and trunk fat estimators appeared, the highest being for WHtR (r = 0.63; p<0.001) and for RFMp (r= 0.58, P<0.001) These trends were maintained for Overweight and Obese groups. The correlation matrix values did not show any further remarkable issue. Table 3 shows the different degrees of association of sex with BMI, waist circumference and fat estimators in the analyzed nutritional groups. As expected, no gender difference appears in BMI values as it was used as a primary categorizer for underweight, overweight or obesity from the whole sample. WC Zs has higher and signi cant levels in girls in all groups, even in underweights (p <0.004). Concerning (trunk) fat mass estimators, WHtR was greater in girls than in boys in the All group (p<0.004) and speci cally in the overweight group (p<0.001) and in the obese group (p<0.004). RFMp showed differences in the All, Normal weight, Overweight and Obese groups in favor of female with same p-values (p < 0.001). All these sex differences occurred with no different BMIs, suggesting it is worth considering that in a bodyweight kilo fat plays a role besides other components.

Discussion
The main ndings of this study are: 1) WC, WHtR and RFMp can be used to estimate truncal fat because the used models explain that their change is associated to the analyzed variables (Table 2), with this association reasonably higher in the whole sample adding valuable information to the BMI lesser estimative capacity of body fat content. 2) The estimators exhibit differences between boys and girls in all nutritional groups whereas BMI did not, being understood that the non-statistically signi cance, in this case is not a primary outcome for equality (20). Estimators mean values and uncertainty range in each nutritional group were obtained.
Derived risk of excessive trunk fat was described in mid 1900s in the adult (21) stressing the body shape of the individuals. These risks were proven later by means of the association with T2D (22,23), hyperuricemia (24), elevation of free T and MRI assessed abdominal fat distribution (25), with heart failure mid-range ejection fraction (26), or even with the vintage metabolic syndrome, this relationship was established by means of different waist circumference derived indexes (27). It is worth referring to the conclusions of Baton Rouge (28) after analyzing these varied equations, although in favor of the Waist circumference index, considered the indexes capacity for evaluating an individual person's health risks. For gaining feasibility other waistheight indexes may be useful (29) and have already been tested in different geographic areas in children as cardio-metabolic risk factor (30). More speci cally and due to the simplicity and reliability of measures, WHtR was chosen for assessing central adiposity successfully in children in a remote South Paci c archipelago (31). With the present-day conception of pediatric obesity risks it is worth considering that these elevated trunk markers are associated with the main biochemical markers for insulin resistance, in ammatory and metabolic abnormalities (32). The clinical approach of trunk fat has led to the description of a situation out of 17000 participants with BMI < 25 kg/m², but with excessive body fat (33). In children and youngsters, the estimation of trunk fat by by-proxy methods has been slow due to the varied charts for WC despite the publications of McCarthy (34,35) facilitating Zs calculation. More recent publications (12,36) included the international centile cut-offs, but nevertheless truncal assessment has not reached the accepted level of BMI in clinical grounds. WC is still considered as a reliable measure for assessing abdominal obesity (37) especially in countries with uneven care distribution, and in others, with better conditions its evaluation is the rst or preliminary step for subsequent more precise tests (38) Waist-to-height ratio. WHtR was proposed almost simultaneously in adults in 1995 in Japan by SD Hsieh and in the UK by M Ashwell (14) demonstrating that ratios > 0.5 were strongly associated with myocardial ischemia and metabolic risk factors (T2D). This association has also been described in children and adolescents elsewhere (39,40). Other variants of this ratio (41) are not widely used. In the AVON longitudinal study (8) with nearly 3000 children followed over 8 years, ratios > 0.5 were associated in adolescents with raised fasting blood lipids, glucose, insulin and blood pressure for boys (OR 6.8; 95% CI 4.4-10.6) and for girls (OR 3.8; 95% CI 2.3-6.3) and also the stubbornness of this ratio once established, being highly speci c versus BMI. Similar results had been shown after a systematic review and meta-analysis (42). WHtR, could be considered as a simple and reliable rst step.
Relative fat mass pediatrics (RFMp). As mentioned, Woolcott and Bergman (15) derived an equation from adult height/ waist for estimation of whole-body fat percentage and later on for children and adolescents (16) always matched with DXA values. The novelty of this estimator is the sex consideration that decreases the rate of misclassi cation of relative fat mass with a more precise diagnosis of obesity/ adiposity in females. This equation has been tested in other parts of the world (43,44) in adult populations but also in adolescents (45). In our study the initial correlation to BMI as the major standard criterion for overweight and obesity classi cation, was signi cant, clearly in the whole sample and Normal weight groups, but in Overweight and Obese ones the degree of correlation slightly decreased, in agreement with the next multiple regression comment. This fact is interesting because for BMI calculation waist circumference doesn't intervene which would be more related to body or trunk fat than BMI. The normal distribution and density of RFMp in this study could give an adequate probability to upcoming data ( Figure) Multiple linear regression. The high aR² values are indicative of the appropriateness of the used estimators but as below 90% (predictive capacity), they should be considered as indicative of association mainly for female gender and for waist circumference and in a lesser extent to BMI Zs. Speci c analyses of our nutritional statuses pointed out in All group, the ones with greater fat deposit (obese) to the lowest (underweight) an association between gender and RFMp, the regression B coe cients would imply that these girls have an increment of 3.36 units in their RFMp in respect to the boys, or in the case of WC Zs each unit of increment implies an increase of 2.97 in the RFMp. This associative trend has been very similar and regular in the Normal weight, Overweight and Obese groups. All of this would indicate a greater precision than BMI Zs, basically due to the fact that BMI does not consider waist circumference, which is also manifest through its lower coe cients (Table 2). In WHtR these associations remain at a lower level but keeping their p-values, therefore the simplicity of its calculation (just a cm ratio Waist/Height) and its well proved threshold of 0.5 make it an e cacious screening tool.
The Underweight group, (7.4% of all) deserve few words for its presence here, all these children were discreetly affected (BMI Zs -1.35 SD; 95% CI -1.46 to -1.24) and with a minor reduction of target height (-0.05 SD; 95% CI -0.23 to 0.25) in 20 instances where both progenitors were measured, suggesting a light undernutrition of familiar component, furthermore their social level could be hardly considered as of lower class. The inclusion in the study was motivated in order to assess the estimators' behavior on the opposite spectrum side of overweight.
As regards sex, in all nutritional groups BMI did not show differences between genders conversely the estimators clearly did, as weight apart from fat comprises non-fat body components that veil adiposity. Furthermore, DXA studies revealed a greater proportion of fat in girls particularly when they come up to puberty (46,47). This association with DXA was already studied by us (48) in 142 overweight and obese with an age (mean, 95%CI) of 11.5 (10.3-11.8) years, with no differences between boys and girls in age neither in BMI Zs; furthermore, in this study we found a relation and differences between %trunk fat and sex, 42. (2.1-2.8) girls. p = 0.001; the regression analysis between WC and %trunk fat gave for WC a regression coe cient of β = 2.9 (P = 0.001). Thus, the need to separate sexes in pediatric obesity studies.
The prepuberal age of our girls w BMI%: relative body mass index; MLR: multiple linear regression; NG: nutritional groups; OB: obese; OW: overweight; RFMp; relative fat mass index pediatric; WC: waist circumference; WHtR: waist to height ratio. ould justify the growing values of RFMp on the Normal, Overweight and Obese groups, again this occurs with BMI Z scores no different from boys. The median of RFMp is signi cantly higher in girls than in boys, sharing the same classi catory BMI range in all groups (whatever the BMI degrees were), this is probably in agreement to higher fat content in girls at this age (16) consequently giving an idea of the abdominal fat.
Blood pressure. Only a weak relation of diastolic blood pressure to RFMp (r = 0 .206, p < 0.001) appeared in the All group. Despite well-established policies for BP measurement in the clinical area of this Unit, results are not as consistent as other clinical parameters, and probably the doubts about BP screening (49,50) not only apply to these data but lead to reconsidering these policies.
Present techniques allow the measurement of abdominal fat separately from subcutaneous fat in children (51,52), but these could not be the available method for better evaluation of the growing heavier section of double burden malnutrition in Low-and Middle-Income Countries (LMIC), so it is natural to look for by-proxy estimators. Further, they are useful even where DXA is available but cannot be justi ed in every follow-up visit. Therefore, and after rm association of estimators based on waist circumferences with trunk fat, we recommend the studied estimators to be used because of their safety, simplicity (53) and longtime known low variability of (54) and good correlation with CT and MRI (55) when facing this silent and metabolically unhealthy fat accumulation (3,57) and furthermore for gaining precision as considering gender (56) and as in the adult, if more than one is used (58) Limitations. Having a series of accurate standardized anthropometric data, we tried to check eventual advantages of WHtR and RFMp, therefore without a speci c design for child obesity studies (59). As females were predominant (286/472, 60.6%) the subsequent higher proportion in the study groups could be a confounding consequence, nor was it possible to consider the validation concept (60). In this way another weak point is the lack of precise comparative pattern as could have been DXA as just mentioned, but in this case the aim was implementing diagnosis purposes through anthropometry for nutritional deviations, some of them not requiring more complex technics with side effects.
Conclusions. WHtR is a very simple and reliable method that does not need reference growth centile charts, consequently it should be the rst step screening, while RFMp gives an idea of the body (and trunk) fat content in both genders. Both could warn us of cardio metabolic consequences already present or in the near future specially if they increase in the follow-up. At present BMI Z score is considered the widest marker for overweight and obesity (while BMI percentage is better understood by the family), hence and in order to increase clinical accuracy both estimators should be added in routine anthropometric measurements in primary health care and in speci c surveys. No nancial and non-nancial bene ts have been received or will be received from any source or party.

Funding
None was received Authors' contribution MM did the study design the study design, data analysis and article drafting. VP performed also the data analysis and the statistical study. Both authors revised thoroughly the manuscript and approved the submitted version.