As many of our regular readers know (and often remind each other in the comments), body mass index is not a great predictor of an individual’s risk of disease. It predicts health risk at the population level, but even then there is considerable variation within a given BMI range. Peter has talked about this at length in the past, when he summarized 3 reasons why BMI is a poor measure of your health:

**BMI does not differentiate between the Michelin Man and The Terminator (e.g. muscle and fat)****BMI does not differentiate between apples and pears****BMI does not always budge in response to lifestyle change**

He has also touched on the issue in his series on metabolically healthy obesity (e.g. people who are healthy despite a high BMI), and it was discussed at length during the recent “Is Obesity A Disease” debate here in Ottawa.

In response to this issue, a paper was published last week in PLOS ONE which reported a new measure which they call the “Body Shape Index”. The goal was to develop a method of assessing body fat distribution (which matters much more than weight or even body fat percentage) that is able to predict health risk beyond that predicted by BMI alone. My first reaction was that we already have such a measure – waist circumference. For example, this paper from Peter’s PhD suggested that waist circumference could predict diabetes risk (but not cardiovascular disease risk) even after controlling for BMI. Why not just stick with waist circumference?

From the paper:

A key limitation… of using WC as a proxy for abdominal fat distribution is that it is sensitive to body size (height and weight) as well as to fat percentage and distribution. In fact, WC is highly correlated with BMI, to the extent that differentiating the two as epidemiological risk factors can be difficult [19].

So this paper attempted to develop an index based on waist circumference which was independent of height, weight and BMI, and then see whether that index was a good predictor of health risk.

**What did they do?**

The paper used data from the NHANES study, which is a large nationally representative survey from the US. For this analysis their dataset included more than 14,000 non-pregnant adults (pregnancy obviously affects waist size for reasons unrelated to abdominal fat, so pregnant women were left out). The authors used this dataset to predict a person’s waist circumference based on their height and weight, and found they could do so relatively accurately using the following equation [note: I’m not a stats or math guru, so if anyone has read the paper and thinks I am misunderstanding the authors’ mathematical judo just let me know]:

WC ~ (BMI^{2/3})*(height^{1/2})

Taking that a step further, they defined their Body Shape Index as the ratio of a person’s actual waist circumference and the predicted circumference based on their height and weight [Update: thanks to Russell Uman for catching my error in the original writing of the equation]:

Body Shape Index= WC/[(BMI^{2/3})*(height^{1/2})]

**What did they find?**

The authors report that this new index is not correlated with BMI, height or weight, and even the correlation with WC is quite poor (which are all good things from their perspective).

What about the relationship between this new index and health? From the paper:

Death rates increased approximately exponentially with above average baseline ABSI (overall regression coefficient of per standard deviation of ABSI [95% confidence interval: –]), whereas elevated death rates were found for both high and low values of BMI and WC. (–) of the population mortality hazard was attributable to high ABSI, compared to (–) for BMI and (–) for WC. The association of death rate with ABSI held even when adjusted for other known risk factors including smoking, diabetes, blood pressure, and serum cholesterol. ABSI correlation with mortality hazard held across the range of age, sex, and BMI, and for both white and black ethnicities (but not for Mexican ethnicity), and was not weakened by excluding deaths from the first 3 yr of follow-up.

So it would appear that this new index is a better predictor of mortality (e.g. risk of death) than BMI or WC, at least in this population.

**What’s the take-home message?**

I’m not quite sure. I don’t feel I know enough about the stats used in this paper to know if the benefits of this new index are clinically significant (e.g. useful for my physician when trying to predict my health risk) or just statistically significant. It seems interesting, but I need to see some other papers on this (or hear from some stats folks) before I can really decide what to think of this new index.

I’m also not so certain that anyone is ready for a new measure in the clinical setting – my primary care provider has never even measured my waist circumference, let alone plugged it into a fancy equation. But if this new measure proves valuable, then that would be worth pushing for.

Finally, I’m curious how this measure would respond to interventions. This has always been the big problem with waist-to-hip ratio and BMI – you can have a considerable reduction in visceral fat (and related health risk) without having much impact on either of these measures. If the measure doesn’t change along with health risk, then it will be of limited clinical utility. I’m assuming this new index will be better in this respect, but it will still be interesting to follow in the future.

As always, I’d love to hear what you think!

Travis

Krakauer NY, & Krakauer JC (2012). A new body shape index predicts mortality hazard independently of body mass index. PloS one, 7 (7) PMID: 22815707

**Interested in learning more about similar topics? Sign up for future updates to be delivered to your email inbox or RSS reader, check out our Facebook Page, or follow us on Twitter.**

You give the equation Body Shape Index = WC/[(weight^2/3)*(height^1/2)]

But in the original article it’s ABSI = WC/BMI^2/3 * height^1/2, which is a little different.

Sorry in advance if I’m missing something obvious!

Here’s a calculator for ABSI:

http://twtitw.firebus.com/node/279

Hopefully the math is all correct. And hopefully everyone will understand that this reasearch is too new and the sample size too limited to mean much of anything 🙂

More calculators in the comments section of the paper:

http://www.plosone.org/annotation/listThread.action?root=52123

Thanks for that, Russell! I’ve updated the post to reflect the error. This explains why the exponents changed from one line of the paper to the next – one line was using weight, while one was using BMI. I wish they would have made that jump more clear!

i confess to being a little mystified by the math. what happened to the theory put out by dr oz and others that wc should be no more than half height? and that things got progressively worse from there and there is no cutoff like 40 inches?

~~The lowest mortality hazard was for the middle quintile of both BMI and WC, although the population median was well in the WHO ‘overweight’ or ‘pre-obese’ category [39]: the 40th-60th percentile range for the sample was 25.6 -28.4 , with the exact cutoffs for the middle quintile of BMI z score varying by age and sex (Figure 1c; cf. Figure 4b). Similarly, the 40th-60th percentile range of population WC was 94–101 cm for men and 88–97 cm for women, above most suggested cut-off points for higher mortality hazard [18].~~

I’m 52, and I got this result:

BMI: 27

ABSI: 0.075

Waist: 88.9 centimeters

I’m not a scientist/statistician, but from what I gathered, that’s not bad.

I am Hispanic and my weight does cluster in the torso/belly, which has always been worrisome to me, though. I wonder why there is a Mexican deviation.

I’m a bit curious about that as well. Not sure what that means, but it’s certainly worth keeping an eye on moving forward in future studies. All these measures vary a bit for different ethnic backgrounds, so it’s good that they at least looked at the different groups available in the NHANES.

Let’s focus on the inputs (morphological measurements) and not the invention of novel equations that use the measurements as variables.

BMI (weight, height): Does not differentiate between fat and lean body mass, and therefore mostly useful for the sedentary (sadly, this is the majority of people).

Body Fat (waist, weight): There is a well-known population curve fit (adults only) that estimates body fat from these measures, commonly known as the YMCA formula.

Body Fat (waist, height, neck, hips-women only): Another well-known population curve fit, commonly known as the US Army or Navy formula. It is more accurate than the YMCA formula, which is very WC-dependent.

There are additional factors that can be plugged into these various models (formulas) to more accurately fit sub-populations, such as children, the elderly and athletes. From the studies I’ve seen there is no real ethnic factor for good vs. bad body fat percentages, although there is (somewhat) for BMI.

I am puzzled that a researcher would take some mix of these well-studied morphological measurements that are already well-correlated with body fat and then come up with some new formula and name. All one has to do is use the more-accurate (and endlessly studied) measures of body fat rather than the often-unreliable YMCA equation.

I am not a researcher, but very interested in fitness, so I researched the literature. I put this learning into an Android app but since I am not here to promote the app I won’t name it!