Problems with Measuring Body Composition

In previous articles on the topic of body composition, I’ve addressed the question What Does Body Composition Mean?, showed you how to do Body Composition Calculations, and taken a look at Body Composition Numbers.  In the last two articles, Measuring Body Composition Part 1 and Measuring Body Composition Part 2, I looked at different methods that, even if they don’t measure body composition exactly, will at least allow tracking of progress of some sort.

Today, I want to back up a bit and get a little more technical and look at some of the problems inherent to tracking body composition accurately.  This will lead me to finally round out this series by giving some specific recommendations on how to use the various methods to get the best measure of what’s going on.

While each of the methods that I discussed in the previous two articles have their own individual issues (some of which I discussed in those articles), the underlying issue with most methods of body composition measurement is that they have any number of inherent assumptions that, as often as not, are turning out to be false.

As well, calipers (which tend to be used most commonly for measuring body composition) have their own specific set of problems that I want to look at in some detail.

Assumptions about Body Density

As I mentioned in Measuring Body Composition Part 2, most of the methods that I’ve described don’t actually measure body fat percentage. Rather, they measure body density which is then used to estimate body fat percentage.  What does that mean?

Every tissue in your body (e.g. muscle vs. bone vs. fat cells) has a different density.  Taking you back to the horror of high-school science class, density is a measure of how much weight there is in a given volume of something.  Stuff that has more weight in a given volume has a higher density than something with less weight in that same volume.

Quick tangent: One of my major pet peeves is when folks say that muscle weighs more than fat; which doesn’t make any sense at all.  One pound of muscle weighs exactly the same as one pound of fat, the difference is that the fat is less dense and takes up more space on the body.  The graphic below shows one pound of muscle (the red stuff) versus one pound of fat (the yellow stuff).  As you can see, the fat takes up more space (it is less dense) but both weigh the same.  Ok, enough pedantry.

 

So every tissue in your body, muscle, fat, bone, organs has some density and the various body composition methods are actually making an estimate of body density.  That estimation goes into a second equation which then churns out the actual body fat percentage number.

So what’s the problem?

Well one of the key assumptions underlying most methods of body composition is about the specific densities of each of the different tissues and that that density is unchanging, and that it is identical from person to person.  All of which turn out to be more or less incorrect.

The original body density values were determined from a small sample of old white male cadavers way back when and, so far as I can tell, haven’t been updated.  To determine the values, first the cadavers were underwater weighed prior to dissection so that each tissue could be weighed and the true density determined.

The problem is that older sedentary white males are not going to have the same tissue density as a young male or female athlete but researchers have used the same values regardless.   As I mentioned in discussing underwater weighing, one of the problems there is the issue of blowing all of the air out of the lungs; this wasn’t an issue in dunking dead people.

As well, training often increases bone density and this can generate some really amusing results on body composition estimates (some equations will give negative values because of this).  There are also ethnic differences with blacks having, on average, denser bones than whites, and Asians having slightly less dense bones.  Tissue density can also change with age (e.g. bones often become less dense) and assumptions about these age related changes may be incorrect for individuals involved in heavy training.   This makes a set of assumed density values based on old white guys a little problematic.

It’s worth nothing that newer methods of body composition measurement such as DEXA allow for the tissue density values to be determined for an individual (DEXA is often used specifically to measure bone density) and avoids this particular problem.

Now, while there may not be absolutely massive differences in tissue densities between individuals or ethnic groups, my point is that the values aren’t identical or constant as many of the equations assume either.  As I noted, heavy training (especially weight training) tends to increase bone density and female bodybuilder I trained was found to have the bone density of a 20 year old despite being in her 40’s.  So the equations tended to give strange values for her.

Continuing in this vein, studies are showing that one type of lean body mass (called essential lean body mass) has a different density than another type (called inessential lean body mass). Researchers now delineate different types of subcutaneous body fat, which may have different densities as well. You’re probably starting to get the idea of the complexity of the situation and why assumptions about fixed unvarying densities for the different tissues can cause problems.

Now, as I noted above, with the exception of DEXA, pretty much all body composition share the above assumption and problems.  However, calipers, which I discussed in Measuring Body Composition Part 2 have their own individual set of problems that I want to discuss next.


Specific Caliper Problems

Despite giving values very close to that of hydrostatic weighing, calipers have their own set of problems on top of the body density issues I discussed above.

One of these is the assumption that skin thickness is the same among individuals and always constant.  While the differences tend to be small (a millimeter here or there), when you’re measuring a lot of sites, and dealing with someone who is pretty lean, a one millimeter difference can throw off the estimation.  Putting some values to it, a one millimeter difference over 10 sites turns out to be significant and can change the body fat estimate by about 1.5%.  While this is fairly irrelevant for fatter individuals, it can become relevant when folks get lean.

The next issue I sort of dealt with in the previous article and that has to do with where body fat is carried and the number of sites measured. As I mentioned in Measuring Body Composition Part 2, equations which use fewer sites (one common one is pec, abdominal, and thigh for men) can drastically under-estimate true body fat percentage if someone carries a lot of fat in an unmeasured site (upper back is a common place for males).

Often individuals are losing body fat in unmeasured places but this won’t show up with a 3 site measurement and it will look like the diet isn’t working.  As I noted in that article, taking more measurements can get around this but also requires a partner who knows what they are doing; as well, more sites gives them more chances to mess up.  So it’s always a trade-off.

Additionally, visceral fat (the stuff found in and around the gut) isn’t even measured by calipers, although methods such as DEXA (or even the waist to hip ratio) can track changes there.  By many methods, a loss of visceral fat will actually show up as a loss of lean body mass although it’s not.  Someone losing visceral fat early in a diet may think that their diet really isn’t working when it actually is.

And of course, there are also the other issues endemic to caliper technique that I mentioned in Measuring Body Composition Part 2; you need a trained operator, they have to know how to pull a proper consistent skinfold, etc.  One thing I didn’t mention is that large skinfolds (as are often seen in the female thigh) can be damn near impossible to measure accurately.

I’d note that if the BodyMextrix 2000 turns out to be a valid and accurate method, that might get around this set of problems entirely.  It won’t solve the other issues inherent to the method but at least will avoid issues related to caliper technique per se.

Problems with the Caliper Equations

In addition to the issues inherent to caliper measurements I discussed above (and if you wonder why I’m spending so much time on calipers, it’s because they tend to be the most commonly found and used method), there is another potentially bigger problem and that’s with the equations.

Essentially, there are a whole bunch (I’d imagine hundreds at this point) of caliper equations which take the measurements themselves and convert them into body fat percentage.  And the big issue here is that any single equation will only be truly valid (or even close to valid) for the group that it was originally developed in and/or for.  To understand this I have to tell you how they commonly develop the equations.

First off researchers pick the group that they want to measure.  This might be white college aged non-athlete female, or middle aged black women, or Hispanic high school girls or whatever.  I’m just picking these examples at random, don’t read anything into them.

Next the group in question will either be underwater weighted or DEXA’d and the value obtained here will be assumed to give the true body fat percentage.  Then the group will be calipered at however many sites that researchers want to look at (this can range from 3 to 10).  Then a computer is used to crank out an equation that will let the caliper measurements match up with the DEXA/underwater weighing value.

Now, the big issue comes in when you try to use an equation for one population in a different one.  So while an equation derived for the white college-aged non-athlete females may be fairly accurate in that group, it won’t be accurate at all applied to a different population.  Differences in tissue density, body fat distribution, etc. all throw a wrench into things.  Of course, this mainly just means getting ahold of the proper equation, assuming it exists.

One common way of getting around this is to develop generalized equations; basically they take a bunch of different estimation equations and mathematically put them together (don’t ask me for details) to develop a single generalized equation that gives decent average results.

Perhaps the most common set are the Jackson-Pollock generalized equations, of which there are 3, 4 and 7 site measurements.   These equations tend to show a decent correlation with ‘true’ body fat percentage for both men and women although the 3 and 4 site equations can still drastically underestimate if someone carries a lot of fat in a place that the calipers aren’t measuring.

A couple of final issues with caliper equations.  The first is that most are developed as a curve.  That is, when researchers work the math, usually you end up with something that is curved rather than being a straight line.  What that means in practice is that the equations tend to be fairly accurate in the middle range of body fats but can become very wrong very quickly at the extremes.  So once someone gets above 40% body fat (or so) or below 10%, the equations become progressively more wrong; fatter individuals will be overestimated and leaner individuals are typically underestimated.

A final comment has to do with age which is often included in the caliper equations. Many have found to their chagrin (yes, chagrin) that a birthday suddenly shows them as being fatter, even if the actual caliper measurements haven’t changed.  What’s going on?

The reason has to do with some of the assumptions about tissue density that are being made; it’s usually assumed that bone density is being lost with age, and that muscle is being lost.  Thus an identical caliper measurement (say that a three site measurement gives 45 millimeters total skinfolds) will give a different body fat estimation with increasing age.  The charts that often accompany calipers will show body fat percentage as a function of age and when you move to a higher category, often the values go up.


What’s the Solution?

In previous articles I’ve gone into some detail about different methods of measuring body composition and why I think it’s important; in this article I seem to be saying that all of the methods aren’t actually that accurate and have all kinds of problems.

First and foremost, that’s not exactly what I’m saying.  My only point with today’s article is to make some of the underlying issues with the method known.  Body composition measurements aren’t perfect but no method of measuring much of anything is; that doesn’t make them useless.

Just keep in mind that sometimes some very weird values (such as negative values in lean athletes) can come up; when values fail the reality check (e.g. someone is claiming to be 1.4% body fat based on calipers), it’s time to take a step back and reconsider what’s going on.

Beyond that, unless you’re doing clinical work where absolute accuracy is required in your measurements, it’s usually good enough that the measurements be at least consistent.  That is, for any given individual, it’s far more important to get consistent measurements even if those measurements aren’t exactly accurate.

What does that mean?  Hopefully the following example will make it clear.

Let’s say that someone weighs 150 pounds, let’s say that they get on two different scales and one puts them at 153 pounds and the other at 147 pounds.  Neither scale is accurate because they both give values that are different than the actual value.

Now let’s say this person loses 5 pounds so that they really weigh 145 pounds.  Now they get on the same two scales and get the following measurements.  The first scale, that originally said 147 pounds says the person weighs 142 pounds; that is it accurately measured the 5 pound loss.  But the second, which originally said 153 pounds says that the person weighs 150 pounds; the second did not accurately measure the 5 pound loss.

Neither scale was accurate, but the first was at least consistent (it was off by 3 pounds each time); it reflected the changes properly.  The second scale wasn’t accurate or consistent (it was off by 3 pounds initially and by 5 pounds the second time around).

For measuring body fat percentage, usually it’s more important to be consistent than accurate.  Now, there are places where you want an accurate measurement (e.g. to determine actual lean body mass or what have you) but the reality is that you’re not going to get 100% accuracy almost no matter what you do (DEXA might be the lone exception here).  You can get close (calipers are usually 3-5% off of true measures) but that’s about it.  So the best you can get is consistent.

And most of the issues I’ve discussed in this article are going to be consistent for any given individual. Your bone density is unlikely to change massively over the course of a diet, neither will skin thickness for the most part.  Unless you happen to have a birthday and move up in the body fat percentage chart, that’s a non-issue too.

So assuming that you use the same method, measure at the same time of the day under similar conditions (e.g. don’t compare carb-loaded to depleted), at the same time of the month (this is critical for females due to shifts in water balance and such over the menstrual cycle), etc. you can at least get consistently comparable measures to track changes.

You may not know the exactly accurate values (unless you can afford to get DEXA done a bunch of times) but you can measure changes.  You can tell if you’re losing fat, staying the same or gaining fat.  Of course, as I’ll detail in an upcoming article, standing in front of the mirror or taking pictures every 4 weeks would tell you as much.


Summing Up

Body composition measurements, like almost everything to do with the life sciences, has its set of problems.  Assumptions about tissue density is an inherent problem in most measurements of body composition (DEXA being one of the few exceptions).

Calipers, while common, have their own set of problems and assumptions both related to the method and the equations that are used to crank out the actual body fat percentage numbers.

However, outside of clinical practice, at least consistent and comparable measures can usually be made even if complete accuracy can’t be obtained.  For most applications, this tends to be sufficient.

In an upcoming article, I’ll give some concrete recommendations on how I think people can or should use the different methods of tracking body composition changes to make sure their training and eating programs are working the way that they want them to.