Is It Time to Dump BMI for Youth? TMI May Be Better

Body Mass Index (BMI) is that essential tool we love to hate. As a marker for obesity in adults, BMI stirs never-ending debates. Despite its limitations, nothing seems likely to replace BMI for adults. But for young people, the story is very different. BMI doesn’t work so well for children and teens. And now, a new publication in JAMA Pediatrics offers persuasive evidence that TMI (tri-ponderal mass index) is a better signal of excess overweight and obesity in youth.

Based on their research, Courtney Peterson and colleagues concluded:

The tri-ponderal mass index estimates body fat levels more accurately than BMI in non-Hispanic white adolescents aged 8 to 17 years.

It is worth considering replacing BMI z scores with TMI to estimate body fat levels in adolescents.

BMI: The Standard Screening Tool for Adults

TMI goes all the way back to the early 19th century, but it never got much traction. Body mass index became the standard instead.

Body mass index is calculated from weight and height. It’s a simple formula of kilograms divided by meters squared. It works pretty well in adults for screening and epidemiology, but not for kids. That’s because kids are growing and the relationship between BMI and adiposity keeps changing as kids grow.

To overcome those problems, pediatricians use something called a BMI z-score to adjust for age and gender. But even BMI z-scores are not very accurate measures of obesity. In the present research, BMI z-scores gave the wrong answer more than 20% of the time.

TMI: Simply Better for Youth?

TMI is not radically different. It’s kilos divided by meters cubed. But that little change has a big effect. Error rates go down by 75%. And it’s way simpler than percentiles and z-scores.

Switching over from BMI to TMI will require more work. Researchers will have to show that these findings hold up in more diverse populations. But perhaps the biggest challenge will be forging consensus and bringing clarity. In a 201x review paper, Katherine Flegal and Cynthia Ogden described this need:

Terminology and measures used in studies of weight and adiposity in children and in adults can be complex and confusing. Different reports may use the same term but define it quite differently, making comparisons difficult.

Childhood obesity presents a tricky challenge. Without good measures and definitions it only gets harder. This work is overdue. Congratulations to Courtney Peterson and friends for getting it started.

Click here for the study by Peterson et al and here for the review by Flegal and Ogden. For more from UAB on this research, click here.

Eugene, photograph © Garen Dibartolomeo

Subscribe by email to follow the accumulating evidence and observations that shape our view of health, obesity, and policy.


May 17, 2017

3 Responses to “Is It Time to Dump BMI for Youth? TMI May Be Better”

  1. May 17, 2017 at 11:30 am, Pat said:

    If one wishes to compare the volumes of two similar three dimensional objects one must compare the cubes of their heights (or widths or depths any will do). The square only works for two dimensional objects.

    If the two objects being compared are of the same density, then the proportion of their mass will be the same as the proportion of their volume. That is just mathematics, and not very advanced mathematics at that.

    I can understand that in an age without calculators the square was easier to calculate, and further that for some range the results would be a good enough approximation.

    But we have calculators now. No-one calculates BMI in their head or on paper, in fact mostly they feed the height and weight into a calculator and just read out the results. So there is absolutely no need to simplify the calculation any more. and why use a method which is known to be wrong?

    Simply use a cube based index for all. Then perhaps we will find out that tall thin basketball players aren’t obese after all.

  2. May 17, 2017 at 9:23 pm, Allen Browne said:

    Many of us in pediatric weight management use BMI percentile for age and gender and % over the 95th percentile for those with obesity. Few children are body builders. This system allows us to follow their progress quite nicely when they have obesity. Anything new or being reintroduced would have to be markedly better because it would require a lot of recalculating. As you say, the key is for agreement on one system so we can easily compare results.

    • May 19, 2017 at 12:51 am, Ted said:

      Thanks, Allen, for sharing your experience and perspective on these findings.

      Pat, I appreciate your thoughts. It does occur to me that people (and children) are not objects and so they don’t scale isometrically. We organisms have funny ways of growing – not quite like cubes.