Read on Twitter
Is BMI racist? Tweet thread by a mathematician who studies weight change, BMI and body scaling https://twitter.com/saoirse_idk/status/1354515155049213960
BMI is an index that adjusts weight for height. Imagine a 280lb guy. You would have one image in your head. But if I told you that he plays for the Knicks and is 7 ft tall, the image should adjust 1/22
You can do this adjustment a number of ways: you can take weight and divide by someone’s height, or you can take weight and divide by the square root of height or…you get the idea 2/22
In 1835, a mathematician named Adolphe Quetelet discovered that the optimal adjustment is to divide weight in kg by height in meters squared. This is called a SCALING LAW. I describe how this is done in this video 3/22
In Quetelet’s time, population data was represented by averages. There was no formal measure of variation. STANDARD DEVIATION was developed by Pearson in 1893, over 20 years after Quetelet’s death 4/22
STANDARD DEVIATION is a measure of the variability in your data. This metric, which only became part of mathematics in the 1920s, is critical for identifying how well your data is represented by the average value. Stats is very new to the field of mathematics 5/22
Quetelet was ahead of his time because he recognized that there IS variability of the data and that it’s unfair to represent the data by just averages. Looking at weight alone, the data was highly variable which is why he adjusted weight for height 5/22
Quetelet derived the optimal scaling of weight to height using a sample of white/European populations. To extrapolate his findings to other races is a problem statistically. 6/22
However, our team has reproduced this optimal adjustment by height squared for different races and found that Quetelet was correct: the optimal power is squaring, independent of race 7/22 https://www.nature.com/articles/s41387-018-0068-3
See also 8/22 https://academic.oup.com/ajcn/article/100/6/1455/4576513
Quetelet was ahead of his time because he recognized that there IS variability of the data and that it’s unfair to represent the data by just averages. Looking at weight alone, the data was highly variable which is why he adjusted weight for height 9/22
He dissected a distribution of population data into thirds which was the precursor to the notion of standard deviation only formalized decades later 10/22
https://en.wikipedia.org/wiki/Standard_deviation#/media/File:Standard_deviation_diagram.svg
https://en.wikipedia.org/wiki/Standard_deviation#/media/File:Standard_deviation_diagram.svg
If one were to use weight alone, as was done before Quetelet, then the 280lb basketball player would be considered as having excess weight and be overcharged for health insurance 11/22
There is a follow-up assertion that BMI is being used to make racist decisions. What is the evidence behind this?
To answer this we need to think of BMI cut offs. 12/22
To answer this we need to think of BMI cut offs. 12/22
BMI cut-offs determine levels and degrees of underweight, normal weight, overweight: Class I obesity, Class 2 Obesity and Class 3 Obesity. 13/22
CUT OFFS are used in the medical profession to determine when to intervene. For example, at BMI=30 kg/m2 the NIH classifies the individual with Class I Obesity and a clinician may recommend weight loss. 14/22
Race matters here because the cut-offs are used to make health care decisions. Check out this podcast discussing some of these issues with cut offs, race, and maternal health 15/22 https://revealnews.org/episodes/reproducing-racism/
CUT OFFS are mathematically determined and are different by different outcomes. For example, the BMI cut-off for increased breast cancer risk is not the same BMI as the one for mortality 16/22
What we know is that African Americans have lower amounts of belly fat (visceral adipose tissue) than white ppl at the same BMI. We also know that Asian Americans have higher amounts of belly fat at the same BMI. Belly fat is correlated to heart problems 17/22
So, if a decision to intervene for heart health is made using the accepted BMI cut-offs, the intervention would be early in African Americans and late for Asian Americans 18/22
This is advantageous for African Americans, because they would be advised weight loss at an earlier stage than a white patient. Asian Americans would slip past a diagnosis because their BMI may be classified normal while they may have subclinical heart related risks 19/22
We have enough data and technology tools now for physicians to enter what they are concerned about and receive a suggested BMI cut-off for that individual based on race, sex, age. I mention this option in this video 20/22
So, is BMI racist? Health care inequities, and racially biased providers is an issue, as we’ve seen in maternal mortality rates
But not using SCALING LAW and STANDARD DEVIATION can also be detrimental because we’d quantify the population incorrectly 21/22
But not using SCALING LAW and STANDARD DEVIATION can also be detrimental because we’d quantify the population incorrectly 21/22
We must completely understand the data and statistics. “In God we trust, all others must bring data.” – W. Edwards Deming
