Should BMI be used to measure obesity in children?

Stephanie wrote in about a very common problem:

My daughter is 4 years old. She isn’t the tallest cat in town (she is about the 15th-25th percentile for height), and her BMI always ends up being in the high range (like over 85%). I worry about it. I am very health conscious for myself and my family. We live by all of the “rules.” And yet.

The family doctor doesn’t worry — been shrugging it off since day one. Maybe because both Dad and I are very lean. Maybe because, as patients of hers, she knows we are a very healthy family (regular exercise, healthy diet, no smoking, healthy pregnancy with aforementioned child). Family doc knows we have never fed our kid a drop of juice, no fast food, homemade meals, limiting screen time, healthy choices …

So I’m stumped. Why the high BMI for my daughter? I would love to hear some solid, scientific data about why this could be, as opposed to: “Meh, she’ll be fine.”

We know that obesity, in the long run, isn’t good — but we can’t even agree on what “obesity” is. BMI, or body mass index, is a single number that basically reflects weight-for-height. We figure that the more someone weighs for their height, the more likely they are to weigh “too much.” What we really need is a measure that tells us when someone’s weight is unhealthy, or likely to lead to ill health. Instead, we use that BMI number, a very poor predictor of individual health outcomes.

There are several reasons why BMI is not a great way to discriminate between healthy and unhealthy weights.

A BMI doesn’t reflect the difference between lean muscle mass and fat mass. What’s unhealthy is excess body fat, not excess body muscle. A muscular, lean individual with little body fat may have a “high” measured BMI because muscle has weight.

BMI doesn’t distinguish between kinds of body fat. We know that visceral fat — the kind in your belly, or the kind that contributes to an “apple” shape — has far more long term negative consequences for health than fat distributed in the lower body.

Criteria for “healthy” versus “unhealthy” BMI are based only on statistics, not on individual health outcomes. We’ve decided that anyone above the 85 percentile for BMI (down to age 2) is overweight, and anyone above the 95 percentile for BMI is obese. This compares a child or adult’s BMI against historical data, which assumes that people thirty years ago had a BMI distribution healthier than today. While that’s generally true for the population (obesity-related health problems are genuinely much more common now), that doesn’t mean it’s specifically true for each individual or child. In other words, relying on statistics forces us to oversimplify and generalize instead of focusing on ways to individualize our approach to maximize health.

Finally, improved diet and exercise habits improve health outcomes, even if the BMI doesn’t change. Over-focusing on BMI can lead to discouragement, preventing steps that can really improve well-being in children and adults.

So what should Stephanie’s mom do? Forget the BMI and keep up those good healthy life habits. Stay activeTurn off the TV. Eat moderate-sized portions, slowly, eating mostly plants and whole-grains. Eat as a family, and share cooking and cleaning chores together. Avoid eating out or doing take-out too often, and stay away from sweet drinks (soda and juice are equally unhealthy). Enjoy eating and playing, together as a family, and don’t worry about the numbers on the scale. The BMI is one thing, maybe a starting point to remind us to keep up healthy habits. But it’s a terrible target to use as a goal for your child’s body.

Roy Benaroch is a pediatrician who blogs at The Pediatric Insider. He is also the author of Solving Health and Behavioral Problems from Birth through Preschool: A Parent’s Guide and A Guide to Getting the Best Health Care for Your Child.

Comments are moderated before they are published. Please read the comment policy.

  • Ron Smith

    Hi, Roy.

    Good article. I’ll go further and say that BMI is not only deceptive but pigeon-holes kids based on a single parameter in a way that is itself inaccurate and destructive to children.

    Everyone would like to have some sort of pass-fail data point like the BMI to make the call. But that is just not going to happen.

    I have never, ever liked BMI. Even growth charts are something that the average Pediatrician understands well enough to do no good. Here’s the typical conversation between parent and provider.

    “Well, Mrs. Jones, little Johnny is between the 25th and the 50th percentile for weight and about the 25th percentile for height,” says Dr. Goody.

    “Is that good, Dr. Goody?” replies Mrs. Jones.

    “Yes, that’s good,” replies Dr. Goody.

    And that is the end of the conversation. If health professionals don’t understand how to interpret growth charts better than that then no wonder they clamber for something like BMI to do all the work for them!

    A better substitute is BDI, or body density index. Each type of tissue has fairly narrow density and if the body is immersed in a known volume of water which is then measured, then the volume of the body can be determined. Then it is a simple matter to calculate a total body density and create an index. I detailed this specifically in my book The Pediatric Guide for Parents (an ebook available on iTunes). It gives a composite density and not simply a crude relation between height and weight.

    Still, better than that is to clearly understand growth charts. Bear with me please and I’ll explain.

    I still use circa 1994 CDC data in my custom EMR. There are 2000 and 2010 CDC datasets as well. So why not move up to the most recent dataset?

    The datasets are based on a sample of children at a particular point in time. The datasets are much smaller than probably you might surmise. The 2000 dataset was only between 4,000 and 5,000 children.

    The census in 2000 revealed in the US that there were 15% black, 8% hispanic, and the remainder white which most likely represents European ethnogenetics. In 2010 the census revealed that there were 18% hispanic, 15% black, with the remainder of European ethnogenetic origins.

    What this means is that the growth charts of 1994, 2000, and 2010 show a shift in the stream of percentiles over time. Yet they are still parallel.

    Indeed then an individual’s growth rate (the line through their weight and height plots) is really being compared to a stream of other growth rates. So if an individual is more closely similar ethnogenetically to the particular chart sample, then the percentile reading is often closer to the 50th percentile.

    What about outliers? My Vietnamese patients are almost always well below the 5th percentile. Are they failure to thrive and does their BMI tell me anything at all?

    What about the Scandinavian family whose children are well over the 95th percentile? They clearly aren’t obese on exam.

    These two differing ethnogenetic individuals are being compared to a sampling of children that are not their ethnogenetic peers. And so they don’t fit with percentile or BMI comparisons.

    What this tells us is that parallelism is the important data to look at. All these individuals will be parallel to any of the growth charts for any data sample that I’ve mentioned, regardless of their ethnogenetic origins.

    In a time when our population is ever changing, this understanding is more important than ever. In summary, look at parallelisms and not percentiles. Teach parents these understandings. It take more time, but I promise you it will be well worth it. Forget BMI. That is an unmitigated failure. Even BDI (body density index) is not practicable.

    Then teach children to read labels and understand their BMR so that they can see how they are gaining weight and what choices are better choices for foods. It astounds them when I explain that to lose 10 pounds of fat, that they have to burn between 40,000 and 50,000 calories. At best this means strenuous activity to a negative 1,000 calories a day for almost 2 months!

    Hope that is helpful. Did I remember to say that I hate BMI? And now I’ll step down off my soapbox.

    Warmest regards,
    Ron Smith, MD
    www (adot) ronsmithmd (adot) com

  • LeoHolmMD

    Treat the patient, not the numbers.

Most Popular