2023 Volume 28 Pages 25
Background: A 4-year longitudinal study was conducted to develop a model and a point system for predicting childhood obesity.
Methods: This study included 1,504 Japanese 10-year-old children who underwent health check-ups between 2011 and 2015. Multivariable logistic regression analysis was conducted using the explanatory variables overweight and lifestyle. Obesity was defined as percentage overweight (POW) ≥ 20% calculated by the following equation: (actual weight − standard weight by height and sex)/standard weight by height and sex × 100 (%). The model was validated using the Hosmer-Lemeshow test on 10-year-olds.
Results: Our prediction model for development of childhood obesity was based on seven binary variables: sex, lack of sleep, ≥2-h use of television/ games/ smartphone, hypertension, dyslipidemia, hepatic dysfunction, and being overweight. The area under the curve of the receiver operating characteristic curve was 0.803 (95% confidence interval, 0.740 to 0.866). When validated in non-obese children (n = 415), there was no significant difference between actual and predicted numbers of children with obesity (Hosmer-Lemeshow chi-square = 7.90, p = 0.18).
Conclusions: The validated prediction model and point score for obesity development were shown to be useful tools for predicting the future 4-year risk of developing obesity among 10 years-old children. The point system may be useful for reducing the occurrence of childhood obesity and promoting better health.
This article cannot obtain the latest cited-by information.
