Objectives

To estimate and compare tri-ponderal mass index (TMI) and body mass index (BMI) at each age from childhood to young adulthood in the prediction of adulthood obesity-related outcomes.

Study design

Participants of this observational study (n = 432) were from a 20-year infancy-onset randomized atherosclerosis prevention trial. BMI and TMI were calculated using weight and height measured annually from participants between ages 2 and 20 years. Outcomes were aortic intima-media thickness (at the age of 15, 17, or 19 years), impaired fasting glucose and elevated insulin levels, homeostasis model assessment of insulin resistance index, serum lipids, and hypertension at the age of 20 years. Poisson regressions, Pearson correlation, logistic regression, and area under the curve (AUC) were used to estimate and/or compare associations and predictive utilities between BMI and TMI with all outcomes.

Results

The associations and predictive utilities of BMI and TMI with all outcomes were stronger at older ages. BMI had significantly stronger correlations than TMI with insulin (at age 16 years), systolic blood pressure (age 5-20 years), and triglycerides (age 18 years). BMI had significantly greater predictive utilities than TMI for insulin resistance (at age 14-16 years; difference in AUC = 0.018-0.024), elevated insulin levels (age 14-16 years; difference in AUC = 0.018 and 0.025), and hypertension (age 16 to 20 years; difference in AUC = 0.017-0.022) but they were similar for other outcomes.

Conclusions

TMI is not superior to BMI at any ages from childhood to young adulthood in the prediction of obesity-related outcomes in young adulthood.