This retrospective study involved 174 Italian children and adolescents with obesity (52.3% males, mean age 10.8, range age 7–15 years) consecutively admitted for a first visit to the Obesity Unity, Santobono Pausilipon Children Hospital in Naples, Italy, from April 2017 to December 2017. To be included in the study the following criteria were considered: age 4–15 years, diagnosis of primary obesity, no previous weight loss treatment. Overweight, participation to competitive sports, assumption of medications that may affect body weight, or genetic, endocrine and iatrogenic causes of obesity were exclusion criteria. The institutional review board approved the clinical protocol and written informed consent for all procedures was obtained from all the children and/or their parents or legal guardians before the enrolment.
Anthropometry, pubertal and lifestyle assessment
Anthropometric measurements were assessed by the same investigator, specifically trained. Standing height was determined by a Harpenden Stadiometer (Holtain Limited, Crymych, Dyfed, UK). Body weight was measured to the nearest 0.1 kg, by using standard equipment. BMI was calculated as weight divided for height2 (kg/m2). The WHO standards for age-and sex-specific BMI percentiles were used for calculating the BMI standard deviation score (SDS). Obesity was defined by BMI Z-score > 2 [14]. Pubertal stage was assessed by a paediatrician. Prepubertal stage was defined by Tanner Stage I of breast development in girls and testicular volume in boys [15].
Lifestyle was assessed by questionnaire and included information about participation in competitive or recreational sports in the previous 6 months (yes/not) and the weekly hours of training.
Bioelectrical impedance analysis
BIA on the whole body (BIA 101 RJL, Akern, Florence Italy) was performed by the same investigator in standardized conditions: ambient temperature between 23–25 °C, fast > 3 h, empty bladder, and supine position for 10 min. Participants were asked to lie down with their legs and arms slightly abducted to ensure no contact between body segments. The measuring electrodes were placed on the anterior surface of the wrist and ankle, and the injecting electrodes on the dorsal surface of the hand and the foot, respectively [16].
Body composition calculations
FM was calculated according to the height-and-weight-based prediction model, as suggested by Hudda et al. [11] and to the BIA prediction model, as suggested by Horlick et al. [17].
According to Hudda et al. [11] the following equation was used:
$$\mathrm{FM}\;(\mathrm{kg})\;=\;\mathrm{Weight}\;-\;\exp\;\lbrack0.3073\;\times\;\mathrm{height}^2\;-\;10.0155\;\times\;\mathrm{weight}^{-1}\;+\;0.004571\;\times\;\mathrm{weight}\;+\;0.01408\;\times\;\mathrm{BA}\;-\;0.06509\;\times\;\mathrm{SA}\;-\;0.02624\;\times\;\mathrm{AO}\;-\;0.01745\;\times\;\mathrm{other}\;-\;0.9180\;\times\;\ln(\mathrm{age})\;+\;0.6488\;\times\;\mathrm{age}^{0.5}\;+\;0.04723\;\times\;\mathrm{sex}\;+\;2.8055\rbrack$$
exp = exponential function; ln = natural logarithmic transformation; score 1 if child is of black (BA), south Asian (SA), other Asian (AO), or other (other) ethnic origins and score = 0 if not; sex = 1 for male and sex = 0 for female.
For the BIA estimation model, the following equation, previously validated in children aged 4–18 years by Horlick et al. [17], was used:
$$\mathrm{FM}-\mathrm{BIA}(\mathrm{kg})\;=\;\mathrm{weight}\;-\;\lbrack3.474\;+\;0.459\;\mathrm x\;\mathrm{height}^2\;/\mathrm R\;+\;0.064\;\mathrm x\;\mathrm{weight}\rbrack\;/\;\lbrack0.769\;-\;0.009\;\mathrm x\;\mathrm{age}\;-\;0.016\;\mathrm x\;\mathrm{sex}\rbrack$$
R = electrical resistance; sex = 1 for male and sex = 0 for female.
BF% was calculated for both equations as FM/weight × 100.
Statistical analyses
Statistical Package for Social Sciences (version 26.0, SPSS, Inc., Chicago, Illinois) was used for statistical analyses. The Shapiro–Wilk test was performed to assess the normal distribution of the variables. Data not normally distributed were logarithmically transformed before analyses; for clarity of interpretation, results are expressed as untransformed values. Continuous variables were described as mean ± SD. Unpaired and paired student’s T-tests were used to compare means between boys and girls and within the same sex. Differences between the two estimation methods were normally distributed. Thus, Bland–Altman statistics [18] was used to compare the two models of FM and BF%. General linear model was performed to exclude any interactions between mean differences of the selected equations and gender or gender and age. Unadjusted and adjusted linear regression models were performed between the differences of the two methods and their corresponding averages. Partial correlation adjusted for gender, age and prepubertal stage was performed to compare the association between weekly hours of sports participation or BMI Z-score and FM or BF% estimated by the two different equations.