Objective: To model the development of the tri-ponderal mass index (TMI, kg/m3) throughout childhood and adolescence and to compare the utility of the TMI with that of the body mass index (BMI, kg/m2) to predict cardiometabolic risk in a population-based sample of Canadian children and youth.
Methods: We used data from the Canadian Health Measures Survey to model TMI from 6 to 19 years of age. Percentile curves were developed using the LMS method. Logistic regression was used to predict abnormal levels of cardiometabolic markers; predictive accuracy was assessed using the area under the ROC curve (AUC).
Results: Mean TMI was relatively stable from ages 6 to 19 years for both sexes, but variability increased with age. There was no notable difference in AUC values for prediction models based on BMI z-score compared with TMI for any of the outcomes. For both BMI z-score and TMI, prediction accuracy was good for homeostasis model assessment insulin resistance and having ≥3 abnormal tests (AUC>0.80), fair for C-reactive protein and poor for the remainder of the outcomes.
Conclusions: The use of a single sex-specific TMI cut-off for overweight or obesity is hampered by the increasing variability of the measure with age. Weight-for-height indices likely have only limited ability to predict cardiometabolic marker levels, and changing the scaling power of height is unlikely to improve predictive accuracy.
Keywords: epidemiology; growth; metabolic; obesity; statistics.
© Author(s) (or their employer(s)) 2019. No commercial re-use. See rights and permissions. Published by BMJ.