Body composition in childhood and adolescence shows great changes in small periods of time. Obesity is an excess of body fat mass and, in addition to being an increasingly prevalent pathology [1], it is associated with important comorbidities and a decrease in quality of life in adulthood [2].
Therefore, to estimate fat mass in a simple, acceptable and easy way, on a daily basis in a pediatric consultation, a digital calculator has been developed with reliable and reproducible predictive models, with simple anthropometric variables.
Using mathematical techniques, the anthropometric models that best predict the composition of fat mass have been obtained, analyzing a wide range of anthropometric values, obtained from a representative sample of the population aged 11 to 17 years in Vigo and, compared with the values of body composition estimated by bioimpedanciometry.
The predictive models used are generalized additive models (gam models) with variable binary response: Risk of overweight and/or obesity yes/no. To measure the response variable, bioimperdiagnostic techniques were used, applying the following thresholds: Percentages above 32,38% in men with 11 years old, percentages above 30,66% in men with 12 years old, percentages above 28,24% in men with 13 years old, percentages above 25,62% in men with 14 years old, percentages above 23,31% in men with 15 years old, percentages above 21,82% in men with 16 years old and percentages above 21,65% in men with 17; Percentages above 32,25% in women with 11 years old, percentages above 32,07% in women with 12 years old, percentages above 31,97% in women with 13 years old, percentages above 32,02% in women with 14 years old, percentages above 32,26% in women with 15 years old, percentages above 32,76% in women with 16 years old, percentages above 33,57% in women with 17 years old [3, 4].
The first of them is obtained including: sex, weight, height and arm and leg perimeters
The second model is obtained including: sex, age and body mass index (BMI).
The sensitivity, specificity, positive and negative predictive values, true positives and negatives, false positives and negatives, accuracy and the negative and positive likelyhood of both models, are shown in (Table 1).
Figures 1 and 2 show the ROC curves associated with the models, a representation of sensitivity versus specificity.
The area under the ROC curve (AUC) takes values between 0 and 1 and measures the model´s ability to classify cases well; the closer to 1, the better it predicts. For these representations, the pROC package of the statistical software R [5] was used
1. Organización Mundial de la Salud. Obesidad y sobrepeso [Internet]. 09/06/2021 [citado 14 de septiembre de 2022]. Disponible en: https://www.who.int/es/news-room/fact-sheets/detail/obesity-and-overweight
2. GBD 2015 Obesity Collaborators. Health Effects of Overweight and Obesity in 195 Countries over 25 Years. N Engl J Med. 2017;377(1):13-27.
3. Sociedad Española para el Estudio de la Obesidad (SEEDO). Consenso SEEDO’2000 para la evaluación del sobrepeso y la obesidad y el establecimiento de criterios de intervención terapéutica. Med Clínica. 2000;115(15):587-97. doi: 10.1016/S0025-7753(00)71632-0.
4. Jensen NSO, Camargo TFB, Bergamaschi DP. Comparison of methods to measure body- fat in 7-to-10-year-old children: a systematic review. Public Health. (2016);133:3–13. doi: 10.1016/j.puhe.2015.11.025.
5. Robin X, Turck N, Hainard A, et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics, 2011;12:77. doi: 10.1186/147121051277.