Estimating Fitness Bias in Body Mass Index of Middle School Students

Abstract

Since muscle is more dense than fat, athletes tend to have greater mass and BMI than similarly sized non-athletes. Comparing direct adiposity measures and BMI confirms that BMI is a biased proxy for adiposity for elite athletes. A similar bias should exist for non-elite athletes as well as fit individuals.

This paper provides a methodology for indirectly estimating the size of the fitness bias in BMI using median physical activity performances. Approximately 30% of females and 33% of males are fit using this definition.

Using data from 9 062 students, regressions suggest 3.1, 95% CI [0.9, 5.3], of a female’s BMI percentile of 85, and 3.6% of her weight, CI [1.6%, 5.6%], is due to being fit, but 5.6, CI [3.3, 7.9], of a male’s BMI percentile of 85, and 5.9% of his weight, CI [3.9%, 7.9%], is due to being fit. These increases in weight are smaller than, but consistent with, the bias of more than 20% for elite athletes.

Strong performance on individual physical activities decreases BMI percentile and BMI, but doing well on multiple physical activities has the reverse effect. This provides evidence of a fitness bias. BMI report cards should include the caveat that BMI may overstate the adiposity status of fit children.