The current data were derived from a longitudinal study conducted at the Centre for Overweight Adolescent and Children’s Healthcare (COACH) at the Maastricht University Medical Centre (MUMC +). COACH is a specialized center for the evaluation and treatment of children with overweight and obesity, which provides family-based lifestyle intervention to fight obesity in children and adolescents [22].

All children and adolescents with overweight or obesity, who participated in the COACH lifestyle intervention, were eligible for study inclusion. Overweight and obesity were defined according to the International Obesity Task Force (IOTF) criteria based on body mass index (BMI) [23]. A total of 854 children and adolescents, under 18 years of age, were identified as participants. Participants were classified according to sex, weight status, and developmental stage stratified by age as “early childhood” between 2 and 6 years old, “middle childhood” between 7 and 12 years old, and “adolescence” between 13 and 18 years old to screen youngster differences.

The study was conducted according to the Declaration of Helsinki and was approved by the Medical Ethical Committee of the MUMC + (METC 13–4-130, registered at ClinicalTrial.gov as NCT02091544). A signed informed consent from all necessary parties was obtained before inclusion in this study.

Body weight was measured in underwear using calibrated electric scales (Seca© 877, Seca, Hamburg, Germany) to the nearest 0.1 kg. Standing height was measured using a rigid wall-based digital stadiometer (De Grood Metaaltechniek, Nijmegen, Netherlands) following standardized protocols.

Body mass index was calculated (BMI = weight [kg]/height [m]2). To correct for changes in BMI during childhood, age- and sex-specific BMI z-scores were extracted from the Growth Analyzer software (Growth Analyzer VE, Rotterdam, Netherlands) embedded in the electronic patient file. Children were classified as overweight: + 2 SD up to age 5, + 1 SD thereafter and obesity: + 3 SD up to 5 years, + 2 SD based on criteria of the IOTF, as described elsewhere [23].

Body circumferences were measured in standing position, using a non-elastic tape, while neck circumference was determined at the mid-thyroid level [22] by trained staff. Waist circumference (WC) was measured at the midpoint between the top of the iliac crest and the lower margin of the last palpable rib. Hip circumference was determined at the level of the maximum circumference of the gluteus. Thigh circumference was measured at the midpoint between the hip and knee, while the leg was bent in a 90° angle at the knee [22]. Body fat distribution was based on visual inspection by a clinician and subsequently classified as normal, pear, or apple-shaped body fat distribution. Waist-to-height ratio (WHtR) was calculated as a marker of adiposity distribution.

Blood pressure was measured about 20 times during a period of 1.5 h approximately to mitigate “white coat” interferences, in a sitting position using the Mobil-O-Graph equipment following the instructions of the supplier (IEM GmbH), where appropriate cuff size for the circumference of the upper arm was used as described for children [24]. Mean systolic blood pressure (SBP), diastolic blood pressure (DBP), and z-scores were calculated according to reference values related to height and sex [25]. Furthermore, blood biochemical markers were analyzed using validated standard operating procedures.

Lifestyle factors were collected through several questionnaires within a structured interview performed by a clinician. The questions conducted during the interview within a structured survey were as follows: How many hours a day do you sleep at night on a weekday?/How many glasses of sugary drinks do you consume in a day?/How many hours of screen time (TV, tablet, computer, etc.) do you watch in a day?/Do you do any physical exercise? Physical activity, dietary habits, and lifestyle factors were evaluated as published previously [21], which were ran under appropriate regression models.

Different indirect hepatic markers have been calculated, considering the necessary criteria and applying accepted formulas [26]. The equations are shown in Supplementary Table 1. Hepatic steatosis index (HSI) was chosen as a proxy marker for hepatic steatosis in this young population for further analyses.

A descriptive analysis concerning anthropometrics, biochemical, and hepatic markers across sex-, weight-, and age-specific groups was performed. The normality of the variables was screened using the Shapiro–Wilk test. Descriptive statistics were given as median and interquartile ranks (IQR), and differences were assessed by t-test or the Mann–Whitney test when non-normal distribution. Categorical variables were reported as percentages and compared with the chi-squared test. Some measurements and/or baseline data are missing, but apparently, these lacking data did not jeopardize outcomes when comparing “per protocol” and “intention-to-treat” approaches.

Lifestyle factors related to dietary and physical activity habits were chosen to construct the first linear regression model, which were adjusted for age, sleep time, sweet drink consumption, screen time, and insulin. This linear regression was not adjusted for variables such as sex and transaminases to avoid collinearity since these markers are within the equation to calculate HSI. The variance inflation factor (VIF) analysis for testing collinearity between independent variables ensured variable independence. Multiple linear regression models were used to predict liver damage, with HSI as proxy for liver disease. The variables used in the regression models were age, as it is a wide group; screen time, as a sedentary behavior variable; hours of sleep; consumption of sugar-sweetened beverages, as a dietary variable; insulin; as a proxy for health/glycemic status; and BMI and WC, as an anthropometric variable. Model 1 investigated the association between HSI and demographic characteristics and lifestyle factors including age (years), screen time (h/day), sleep time (h/day), sweet drinks consumption (glasses/day), and serum insulin levels (mU/L). Model 2 evaluated the association between HSI and age, sleep time, BMI z-score, and an interaction term between sleep time and BMI z-score. Mediation by sleep time in the relationship between liver damage (HSI) and body fat distribution in children and adolescents with overweight and obesity was further assessed using structural equation modeling following the Zhao et al. approach [27].

All p-values presented are two-tailed and were considered statistically significant at p < 0.05. Data were analyzed using STATA version 12.1 (StataCorp, College Station, TX, USA).