Any experts on body composition and/or DEXA?

Triathlon Forum
1

Hey, all:
I’ve been getting DEXA scans roughly once per year for the last 5 years. I’ve been under the impression that they are the gold standard, but the latest one has me puzzled. For it to be true, despite a “cleaner” diet, it implies I lost 5 kilos of muscle and gained 3 kilos of fat. I could believe maybe I lost a bit of muscle, but to lose that much muscle and gain fat seems unlikely.

Can any experts weigh in? Thank you!

Some background:
The first four were performed at an orthopedic clinic as part of a university hospital. A few days ago I had one at a new facility (a commercial operation), and it gave me some results I found puzzling.

I want to believe the DEXA is accurate…but the results are odd. Visual tests, pinch tests and just general comments I get all suggest I am more lean than I have ever been, and yet the latest test was high by quite a bit. I’m a numbers guy and first searched for reasons why what I see and feel might be wrong and why the numbers are right, but some research suggests that DEXAs can be inaccurate.

I also learned that a hydrostatic test is the most accurate (and that DEXAs read higher than the dunk tank).

Data/info from prior tests:
About 6 years ago I started dropping weight–about 30 pounds at a steady rate over 15 months before I had my first one (13.9% BF…but 21.0 (!) android). It was a wake-up call as “the mirror test” had me thinking I did not have much I could lose.

My 2nd scan was 6 months later, 7 pounds lighter, 9.0% overall and 14.0 android. Progress. The DEXA reported that I gained a kilo of muscle and lost 3.5kg of fat. The results made sense as I further cleaned up my diet. Exercise was about the same.

Scan #3 was 15 months later…2.5kg heavier, 10.3% BF, 16.7 android. +1.5 kilo of muscle and +1 of fat. I was a little disappointed and thought this might be a bit high, but in reality my diet was probably not as tight as scan #2.

Scan #4 was 1 year later. Down 2.5kg (same as 2nd scan) @ 8.4% BF, 12.6 android.

Scan 5, another year later, new facility. Down a further 2kg. 13.6% BF (!!!), 9.9 android. My diet is about the cleanest it has ever been. Yes, I did drop 2kg, but it was done carefully (or so I thought), just cutting out a few indulgences and portion control, nothing drastic, no major cravings, etc.

Each test was done in the morning–no breakfast, similar level of (de)hydration, similar level of exercise in the days leading up to the test (inflammation?).

I’m as lean as I’ve ever been. Power on the bike is solid–even hitting higher numbers than before. I can see a six-pack even when not flexing (that is new).

Other stuff about the new facility:
This new facility had the same GE machine & software. The operator told me it had been calibrated that morning. There was ambient light in the room (whereas the ones in the hospital setting were done in an interior room with fluorescent light, but more consistent (grasping at straws?). The new facility was a weight loss facility that seems to use DEXAs as a way to confirm changes in body composition. While there was no sales pitch, I noticed they had some money-back guarantee offers about helping clients meet goals. The skeptic in me wondered how much an operator can influence calibration (no idea how automated the process is) and whether they might have an interest in bumping things up a bit for the 1st test. That is really reaching, though. Right? :slight_smile:

Thanks so much for any help, ideas, etc.

Regardless of whether the DEXA is right or wrong, my athletic performance is solid, so I’m secure in that. On the assumption this is right, though, (and that I could have lost muscle and gained fat??) I have been thinking about a consult with someone who is really well-versed in nutrition for athletes…like an Allen Lim type of person. I’m open to ideas and suggestions there.

2

anyone?

3

Why not cross-check those results with a bod pod? The only thing that makes sense to me is that you changed facilities and the dexa scan might not be set up in the same manner as the other and gives different results accordingly. I know very little about those machines but was looking for one a couple of years ago and called a doctors office that advertised one. However, they told me theirs was set up primarily to check bone density on women for signs of osteoporosis, and that the imaging filters/software apparently are different if you use the same equipment to test for body composition. So, and I am totally guessing here, I presume that a dexa scan that was set up solely for body composition would give different results than one set up for another purpose, but used differently, or a dexa scan that is used for multiple purposes with software that perhaps makes more guesses based on ranges similar to a body composition home scale, but at a few levels higher precision.

4

Why not cross-check those results with a bod pod? The only thing that makes sense to me is that you changed facilities and the dexa scan might not be set up in the same manner as the other and gives different results accordingly. I know very little about those machines but was looking for one a couple of years ago and called a doctors office that advertised one. However, they told me theirs was set up primarily to check bone density on women for signs of osteoporosis, and that the imaging filters/software apparently are different if you use the same equipment to test for body composition. So, and I am totally guessing here, I presume that a dexa scan that was set up solely for body composition would give different results than one set up for another purpose, but used differently, or a dexa scan that is used for multiple purposes with software that perhaps makes more guesses based on ranges similar to a body composition home scale, but at a few levels higher precision.

Thanks for the reply.

IME the facilities that only look at bone density are typically running older software on the same machines. I could be wrong, but the machines I’ve encountered are all GE machines running GE software. I’ve spoken with people at the type of facility you describe.

All DEXAs seem to spit out the bone density info (that is their primary purpose); the body comp stuff is “extra” in a manner of thinking.

Accuracy, I hear, is not a strength of the bod pod. (I also live in a small town and am not aware of any facilities that have it. I’ve done the first 4 DEXA scans when I passed through a larger city 5h away, and last weekend’s one was while I happened to be in a city 8h away.)

Would love to hear from someone who knows more about DEXA if the operator/facility has any influence on calibration (or is it fully automated?). Software seemed up-to-date to my eye.

Anything else that could be wrong with me…or with the DEXA? The more I think about it the more I think it would be impossible or unlikely for me to lose 5kg muscle and gain 3kg fat over 6 months while keeping my diet largely consistent (if anything…a bit less snacking)–but maybe I’m wrong. If it IS possible, then I’d like to figure out how to make that not happen again.

5

I’ve only ever had DEXA scans for bone density. Out of curiosity, how much are you paying for these tests?

6

I’ve only ever had DEXA scans for bone density. Out of curiosity, how much are you paying for these tests?

1st 4 were free. Last one was $49.

For bone density, it is usually $$$ due to the cost of “interpretation.” The report spit out is the same, and if you don’t need any interpretation you get the same bone density data, T-scores, Z-scores, etc.

Also nice to have those things for a cyclist.

7

I’ve only ever had DEXA scans for bone density. Out of curiosity, how much are you paying for these tests?
1st 4 were free. Last one was $49.

For bone density, it is usually $$$ due to the cost of “interpretation.” The report spit out is the same, and if you don’t need any interpretation you get the same bone density data, T-scores, Z-scores, etc.

Also nice to have those things for a cyclist.

I guess that’s not so bad - doesn’t seem terribly more expensive than having someone measure it with skinfold calipers. My bone density tests have been rather pricey .

If you had four done on one machine and the fifth done on a different machine, I would suspect the odd results have something to do with that (although it shouldn’t). If you can see your abs and your power output is better than ever, then your answer regardless of what the bf numbers are is “keep doing what you’re doing.”

8

I studied DXA quite a bit in grad school. This was in mid '90s but I don’t believe the foundation of the technology has changed, just the software. DXA works by x-raying your body and software determines the amount of fat, bone and muscle based on the different densities of the tissues and their “reflection” of the xray. You’ll note that on your scans, bone is usually black (or a dark color) and non-bone tissue is lighter in color. Software does all the calculations and different versions of software will produce different results (as will different hardware). That is why when you read a research paper a notation of the brand of scanner AND the software version are mentioned. Over time, the software algorithms are refined and tweaked. Also notice the lines on your “skeleton” of the printout. Those are usually placed there by the software (although the operator can adjust manually if needed or for research) quessing where they should be. They separate your body into compartments so that you can calculate the amount of tissue in your limbs, midsection etc. Very slight movement of those lines can impact the results.
I’d ask the DXA operator what version of software they use and see if your older reports have that printed on them to see if they match or not.
Also, ask if the operator did any manipulation of the “lines”.

9

I studied DXA quite a bit in grad school. This was in mid '90s but I don’t believe the foundation of the technology has changed, just the software. DXA works by x-raying your body and software determines the amount of fat, bone and muscle based on the different densities of the tissues and their “reflection” of the xray. You’ll note that on your scans, bone is usually black (or a dark color) and non-bone tissue is lighter in color. Software does all the calculations and different versions of software will produce different results (as will different hardware). That is why when you read a research paper a notation of the brand of scanner AND the software version are mentioned. Over time, the software algorithms are refined and tweaked. Also notice the lines on your “skeleton” of the printout. Those are usually placed there by the software (although the operator can adjust manually if needed or for research) quessing where they should be. They separate your body into compartments so that you can calculate the amount of tissue in your limbs, midsection etc. Very slight movement of those lines can impact the results.
I’d ask the DXA operator what version of software they use and see if your older reports have that printed on them to see if they match or not.
Also, ask if the operator did any manipulation of the “lines”.

thank you for the reply.

interesting point about the lines. i can see how this could affect something like android/gynoid fat.

i compared the last 2 results carefully. the lines look comparable to the naked eye between scans. i am 2 kilos lighter as measured (agrees with the scale). i notice that my arms are 0.7kg heavier (+1kg fat, slightly less muscle), legs are +1.5kg fat, -2kg muscle.

android & gynoid were no change in total tissue or fat, but 3kg more muscle (android…out of 3.9kg total tissue???) and 0.5kg more muscle gynoid (out of 10kg).

seems a bit weird to add 3kg of muscle to the mid-section.

i checked all my reports and do not see any info on the software version. they are all running Lunar Prodigy (most recent one at the new facility was Lunar Prodigy Primo and the report is clearly a bit different–but the basics (Body Composition/Region/Tissue/Fat/Lean/BMC) are there.

reason to be skeptical?

10

Well I know how much of a painstaking process we went through in a research lab to keep things calibrated and I can pretty much GUARANTEE you that no one in the field, especially in a doctor’s office, will spare the time or expense to calibrate thoroughly. I think a perfectly working system still has around 2% error, so it only goes way up from there if things aren’t maintained carefully.
I also remember something about a “fan ray” as opposed to a “pencil ray” which describes how the xray is dispursed from the machine. Settings like that, speed of scan, body position can all introduce error too.
Interestingly, a year ago I did a DXA then a max VO2 bike test. After the bike test I noticed the DXA software (thus results) was set for “female”, so I asked how that would affect the results. We immediately redid the DXA with correct setting (male) knowing that there was some fluid loss from the bike test. Despite the fluid loss and the errant setting, the results were pretty identical. Bodywater is less of a factor with this technology (as opposed to other methods like uww, skinfolds and impedance) which is why I think it has become the new “gold standard” for body comp.