Conclusion: MRI and CT are the most accurate available methods measuring body fat and has the advantage of distinguishing between visceral and subcutaneous fat. However, these and the 3 or 4 compartment methods are primarily used at the research level. To measure very obese persons the 3 compartment method is the only reliable method available. DXA is accurate but very few sites are available and it is limited to use in lean or mildly overweight persons less than around 6 foot tall. The moist accurate method available for use at the office or home is the waist and hip circumference. After circumference, the BIA and skinfold methods are acceptable.
There are several methods of measuring and predicting body fat. The most accurate are research level and are extremely complex and costly. So, how do we know how accurate a method is? When a researcher sets out to determine the accuracy of a particular method of measuring body fat he/she compares the study method to some reference method (criterion). The researcher then determines the difference in prediction between the study method and the criterion. But, just because the study method gives different results than the selected criterion, does that mean the study method is not accurate? No. Every method has a bias to either overpredict or underpredict true body fat. Depending on the criterion method selected you can have different results for the study method.
Underwater weighing, UWW, traditionally has been considered the gold standard and used as the criterion method for hundreds of studies to determine the validity of simpler methods such as skinfold, BIA and circumference. If UWW is chosen as the criterion method and BIA is the study method, and BIA differs from UWW by 3%, then BIA is reported to have a 3% error. However, another researcher might compare BIA to the skinfold method or DXA and report a different error. In fact, some researchers have compared UWW, BIA, DXA, skinfold, and BMI to 4C (one of the most accurate research methods) and determined that they all have about the same absolute error.
Most methods (DXA, skinfold, BIA, and NIR) overestimate the fat on skinny people and underestimate fat in obese people. The reason is that the equations derived to predict fat have been done on lean or mildly overweight, predominately white individuals. Obese people have different water hydration, a different body density and different fat distribution than the subjects used to derive the equations.
While DXA can provide reliable measures of fat-free mass (FFM), it is limited by the size of subject that can be measured. People over about 225 pounds or BMI of 30 cannot obtain reliable readings. This leaves UWW as the remaining reliable method to obtain body volume and isotope dilution to determine total body water (TBW) to measure density and FFM respectfully in the obese.
In terms of accuracy, here is how the methods stack up, most accurate first: Cadaver > NAA > CT > MRI > Dilution > DXA > anthropometric (WC > skinfold) and BIA
Neutron activation analysis (NAA) – as good as good can get and as accurate as chemical analysis on cadavers.
This procedure allows for the direct measurement of the body’s elements in the living human. NAA can identify and quantify about 70% of all known elements. A specimen is placed in a nuclear reactor, shot with radionuclides that are identified and measured via radiation detectors. The precision of this method is parts per billion. As you can see from the picture, this method does not allow extreme obesity to be measured.
Computed tomography CT – accuracy of <1% for full body scan.
The major advantage is that CT can separate visceral fat from subcutaneous fat; muscle from organs; and determine the type of bone (thickness) present. CT produces sharply delineated images of fat and its various components. CT’s main disadvantage is the radiation dose used to obtain the information. For this reason measurements are limited to the lumbar region (L4-L5) and the full body fat is then predicted from this single “slice”. Due to the opening size of the CT, extreme obesity cannot be measured.
MRI – accuracy of < 1%
Like CT, it can separate visceral fat from subcutaneous fat; muscle from organs; and determine bone densities. Because there is no radiation involved it can perform a total body scan for maximal accuracy and fat distribution. Not possible to measure extreme obesity.
Three-compartment (3C) method – %BF error of 1-2%
This method combines UWW (2C) with the dilution method (TBW 1C) to measure three compartments of the body. For the obese, BMI over 30 and/or weights over 250 pounds, this is the only reliable method of obtaining estimates of fat and fat-free mass. [there is no physical way to place the person on the table for DXA, NAA, CT or MRI]
Dilution method – TBW error < 1kg; absolute fat mass error ~1.4kg; %BF error ~ 2%.
Dilution is a method to determine total body water content (TBW) by administration of a tracer and measuring the tracers concentration a short time later. After collection of a body fluid (blood, urine, or saliva) to measure predose background levels, a tracer of labeled water (tritium, deuterium, or oxygen-18) is administered. After a set period of time, say 2-3 hours, another sample of body fluid is obtained. Based upon the measured dilution of the tracer and an assumed exchange rate of the tracer with nonaqueous compounds the TBW is calculated. FFM is then calculated from TBW by assuming a hydration coefficient of fat-free mass of 0.738. With this technique the smallest detectable change in TBW is about 0.8 liter.
Dual energy x-ray absorptiometry (DXA) – repeatability is 0.8% for bone; 1.7% for fat; 2.0% for body weight.
Originally designed to measure bone density, it has been expanded and is able to measure lean mass, fat mass, and bone mass. The equations used by the manufacturers unfortunately are 20 years old and do not take into account the different densities of different persons. The table is also not capable of performing a study on a person over about 225 pounds due to attenuation of the energy beam going through the thicker body. To further complicate matters there are three manufactures (Hologic, Lunar, and Norland) and the results on the same subject is different dependant on the machine and software version. This technology is not adequate to study the obese, BMI over 30, due to table size and ability of the x-ray to penetrate the body.
Underwater weighing (UWW) – total error for body fatness ~3-4% of body weight. Repeated study error on same individual is ~1.7%
Measuring body density has often been referred to as the gold standard and the most common method to determine body density is UWW. The body is weighed in the air and then fully submerged and weighed underwater. The air remaining in the lungs is measured and an estimate is made of the air remaining in the digestive tract. Measuring the volume of water displaced by the body and the difference in weight of the body in the air and in the water allows an estimate of the persons total body density.
Bioelectrical Impedance (BIA) – error of measure about 3-5%. Repeatability in a 70-kg white adult with an experienced technician is 1.5% to 3.4%.
DETAILED DISCUSSION ON BIA
It has been known that the human body conducts electricity for over 100 years. It flows through tissues that contain water. The aqueous tissues of the body are the major pathways of an electrical current, whereas body fat and bone hardly conduct current at all. BIA induces a single 50 kHz frequency and measures the body’s resistance to this current.
Bioimpedance spectroscopy (BIS) – error of measure in TBW is 3.5% to 6.9%.
Similar to BIA, this uses an electrical current to measure resistance. The difference is that it uses a spectrum of frequencies from 1-500 kHz instead of the single frequency used in BIA. All the same factors that influence BIA equally influences BIS.
Circumferences and other measurements – WC, WHR, BMI error 3-4%
DETAILED DISCUSSION AND CALCULATOR FOR BMI
DETAILED DISCUSSION AND CALCULATORS FOR WC AND WHR
Excess weight and the distribution of that weight has tremendous health consequences. Abdominal obesity, as determined by a large waist (WC) or high waist-to-hip ratio(WHR) is linked with multiple health problems. Fortunately, the waist circumference is a fair predictor of visceral fat. More importantly, WC is a good measure of fat loss. The hip circumference is a good measure of subcutaneous fat.
Skinfold thickness – approximately 3-5% error.
MORE DETAILED DISCUSSION ON SKINFOLD
At birth most fat is SAT, so skinfold measurements is an excellent measure of total body fat. With advancing age, the proportion of VAT increases and skinfold measurements are less tightly linked to total fat.
Air displacement plethysmography (BodPod) – errors +4.7% bias with SD of 4.9.
Similar to UWW except the subject does not have to be submerged underwater. This method still has all the limitations noted for UWW plus more. Several studies in children and adults show the predictions from BodPod are worse than UWW, BIA, skinfolds and other measures. One possible reason could be that the BodPod works with its own software and the equations are “dated”. Another reason are the "corrections" made to the raw data in the companies software.
Near infrared (NIR) – approximately 4% error.
When an object is exposed to infrared light it either absorbs or reflects the light according to its chemical structural properties. One company, Futrex, Inc. developed a device to analyze human composition by this method. Futrex emits light at 938 and 948 nm at the biceps. Essentially, this is a single skinfold measurement at one site.
Total-body electrical conductivity (TOBEC) – no longer in use for humans
Along the same principle as BIA and BIS, TOBEC relies on the concept that FFM is a better conductor of electrical energy than fat. TOBEC is a large coil driven by a 2.5 Hz radio frequency current. The subject is placed in the coil and the difference between the impedance of the empty coil and the coil with the subject is proportional to lean or FFM.