The Problem With Body Mass Index (BMI)
Body Mass Index (BMI) is a number I’ve seen floating around in the news lately. One example would be a recent JAMA article by Afzal and colleagues (2016). In the study, the authors wanted to see if a previously known correlation (this does not mean causation) between mortality (death) and BMI was stable over the course of 40 years. Basically BMI is a surrogate for how fat (obese) an individual is, so the connection of interest here is fat and risk of death. The takehome from the study is that the BMI with the lowest risk of death actually increased over the course of the study. Therefore, in more recent years, those that were slightly more overweight than their counterparts years ago, weren’t at quite as much risk of dying these days. I’m sure numerous unsuspecting readers took this to mean that with each passing year, being overweight is becoming less and less unhealthy. This is not at all what the study is saying and the reason is because BMI is a very flawed parameter.
Science News did a good job of summarizing the research article on their website and you can read the breakdown here. I mention this review because there are some notable lines. First is how the article mentions the ongoing debate about which BMI values are healthy (and which are not). The debate comes from the fact that BMI is calculated as your body mass divided by the square of your height (heightxheight). So, there are two values being measured: 1) how much the person weighs and 2) how tall they are. If we are using this parameter to say something about how obese a person is, why isn’t there any measure of fat in there? Well, because it’s more difficult to measure fat mass. The goal in creating BMI was to create a quick surrogate that one could use on the fly to estimate.
Credit for the development of BMI goes to…sigh…Ancel Keys (the major contributor to society’s incorrect notion that fat and cholesterol intake are the main causes of obesity and cardiovascular disease). I probably shouldn’t be too hard on him for creating it, though, because the fundamental problem here is that a “good enough” estimate (BMI) is now being treated as a very robust measure and extensive scientific studies are using it to draw conclusions. You might be wondering exactly how far off BMI actually is from reality. Unfortunately, there is no singular answer. It depends on the individual being measured. One very obvious problem is that it doesn’t handle fit/athletic individuals well at all. Let’s take 4-time CrossFit games champion Rich Froning. He’s 5’9″ weighing 89 kg. If you look at the BMI graph above you will see that it labels him as Overweight/Borderline Obese. Looking at the picture of Rich on the left, does that make any sense? Of course not. Similarly, individuals with very small frames may actually be carrying a lot of fat and BMI will classify them as normal (not obese). In any study that collects large numbers of BMI values, there will invariably be people whose body weights are high because they have a fair bit of lean mass, or low with a large percentage of fat mass. BMI does not differentiate these cases, yet they are always essential to the health and scientific claims being made from BMI data.
In the Science News review, they interview a doctor who says “[the JAMA article] is a very carefully done study.” The study certainly looks like it was carried out very well, but if the foundation isn’t strong or of good quality, it really doesn’t matter. Scientists and doctors should always be trying to improve our understanding of the natural world and to do that the tools we use must improve. It’s time to stop debating the usefulness of BMI and move on to more accurate measures.