[A comparison of different generalized models to study the factors associated with body mass index and circumference...

Introduction: the adiposity excess has become a public health problem; therefore, many authors have studied this phenomenon from different standpoints. However, studies that have evaluated adiposity indicators sometimes do not report the assessment of model assumptions, which may have introduced potential bias in the interpretations. Objective: to compare the metrics of generalized models to estimate factors associated with body mass index and waist circumference in Colombian adults for the year 2015. Methodology: it was an analytical cross-sectional study. A sample of 72,815 adults (aged 18 to 65 years) was considered. Weight, height, and waist circumference were measured, with the first two used to calculate BMI. To estimate the association, five regression models were developed for each marker. The model fit was assessed using GAIC, and residuals were checked using worm-plot graphs. Results: for both adiposity markers, the generalized additive models for location, scale, and shape model showed the best fit according to GAIC and the worm-plot graph. Conclusions: the use of generalized additive models for location, scale, and shape for studying body mass index and waist circumference provides better model fit and more effectively meets underlying assumptions. The estimator magnitudes across models maintain the same directionality; however, their magnitudes show significant variations.