Multiple regression models to predict electrical bioimpedance variables through anthropometric and simple measures in...

Obesity is one of the most common nutritional and metabolic disorders worldwide, affecting all age groups. Multiple regression models can be used to predict bioelectrical impedance analysis (BIA) variables in various populations. These models incorporate a set of anthropometric and simple measures to improve the accuracy of predicting BIA variables. The goal was to develop and internally validate multiple regression models to predict BIA variables using simple clinical and anthropometric measurements in children aged 6-12 years.

This study employs an observational cross-sectional design. Body mass, height, body composition through BIA, anthropometric circumferences, and skinfolds were evaluated. Anthropometric indices were calculated. Statistical analyses were performed using R software. Non-parametric statistical tests were applied after a normality test. Pearson's correlation coefficients were calculated to assess linear associations between each pair of variables. The multiple regression models were adjusted using a stepwise method with the Akaike information criterion. The significance level was p ≤ 0.05.

This study evaluated 128 children subdivided into eutrophic (N = 64) matched for sex and age with overweight (N = 30) and obese (N = 34) groups matched by sex and age. Significant correlations were observed between BIA variables and simple clinical and anthropometric measurements. Variables with the highest correlations with simple clinical and anthropometric measurements were body fat percentage, skeletal muscle mass, total body water, and fat-free mass, respectively.

Multiple regression analysis indicated that simple clinical and anthropometric measurements could be used to estimate BIA variables in children.