The Pitfalls of Bodyfat “Measurement”: BIA & Skinfolds Strike Again (Free)

If you've read my series on body fat testing, you'll know that the accuracy of body fat testing methods is not as high as many people think.  While these methods do reasonably well when looking at groups of people, they don't do as well when looking at individuals.  In fact, some methods can be so far off on an individual basis, that you would be better off simply using things like waist size and the mirror to assess your progress.

Bioelectrical impedance (BIA) and skinfolds are popular techniques for assessing body composition.  I have written about BIA before, and have shown that it can be very inaccurate both as a snapshot of your body fat percentage, and for tracking change over time.  I have also discussed skinfold accuracy in the past.  Two recent studies published in the Journal of Strength and Conditioning Research provide further evidence as to the inaccuracy of BIA and skinfolds on an individual level.

The Studies

The Research Question

Study Design

  • BIA study
    • 82 adults (42 men and 40 women)
    • BIA performed using a single frequency Quantum IV device
    • Body fat % and Fat-Free Mass (FFM) estimated using four different BIA equations
    • Body composition estimates were compared to the gold standard 4-Compartment model
  • BMI study
    • 130 adults (67 men and 63 women)
    • Body fat % estimated using 5 different BMI-based equations, and a 7-site skinfold equation
    • Body composition estimates were compared to the gold standard 4-Compartment model

Results

  • BIA study
    • The average error for body fat % of the four BIA equations ranged from -0.1 to 3.7%, indicating reasonable accuracy on an average basis
    • Three of the 4 BIA equations overpredicted body fat % (i.e., the average % body fat was higher compared to the 4C model)
    • The average error for FFM ranged from -0.1 kg to -2.9 kg
    • Three of the 4 BIA equations underpredicted FFM (i.e., average FFM was lower with BIA compared to the 4C model)
    • On an individual level, body fat % errors ranged from +11.1% to -6.5% in 95% of subjects, indicating sizeable errors in individuals
    • On an individual level, FFM errors ranged from +4.3 kg to -9.1 kg in 95% of subjects, indicating sizeable errors in individuals
    • Results are discussed in more detail in the video
  • BMI study
    • Average error for body fat % for the five BMI equations ranged from -0.2% to 3.2%, indicating reasonable accuracy on an average basis (and similar to BIA)
    • Average error for skinfolds was -4.8%, indicating that skinfolds underpredicted body fat % (i.e., skinfolds showed a lower body fat % on average compared to the 4C model) and was worse than BMI
    • On an individual level for the BMI equations, body fat % errors ranged from 15.9% to -12.4% in 95% of people, indicating sizeable errors in individuals
    • On an individual level for skinfolds, body fat % error ranged from 2.5% to -12% in 95% of people
    • Results are discussed in more detail in the video

Summary

  • Average error for BIA was up to 3.7%, but error in individuals was as high as 11.1%
  • Average error for BMI was up to 3.2%, but error in individuals was as high as 15.9%
  • Average error for skinfolds was 4.8%, but error in individuals was as high as 12%

Limitations

  • Subjects on average were young (early 20's), healthy, and of normal weight
    • Results could be different for different populations
  • Results could be different for different BIA machines or skinfold equations
    • Accuracy of BIA may be improved with segmental (separates body into arms, legs, and torso), multifrequency (uses multiple frequencies to distinguish between intracellular and extracellular water) equipment
  • Study only gives us snapshot accuracy and can't tell us the accuracy for assessing change over time
  • The 4 compartment model can have "propagated error", which is error compounded due to errors in the individual techniques used to obtain the estimate (hydrostatic weighing to get body density, DEXA to get bone density, bioelectrical impedance to get body water)
    • Researchers did take steps to account for this

Practical Application

  • BIA and BMI can be reasonable methods for estimating body composition in groups of young, healthy people, and have similar error rates when looking at groups.
  • For individuals, BMI has unacceptably high error rates for estimating body composition, but skinfolds and BIA are not much better.  For example, BIA could tell you your body fat % is 20% when in could be as low as 9.9% or as high as 26.5%.  Thus, take the body fat % number you get with these techniques with a grain of salt.
  • If you are going to use skinfolds or BIA for tracking change over time, make sure the change you expect to see is larger than the error rate of the device.  Since the individual error rates are high for these methods, you should take very long periods of time between measurements.  Don't expect these methods to accurately detect a drop in your body fat percentage from 23% to 20%.  You would want anywhere from 5-10% drops or more.
  • The best way to assess progress is the mirror, and measurements such as waist circumference.  Skinfolds can also be used, but ignore the body fat percentage and just look for decreases in skinfold thickness.  Make sure you standardized the measurements as much as possible (same technician, same conditions, etc.).

If you would like to see more research reviews, video presentations, feature articles, and other content like this, become a Weightology member!

 

Leave a Reply Text

Your email address will not be published. Required fields are marked *