Why BMI is not a good benchmark if you do significant weight training while losing weight

I am fat/obese. But I've always 'carried my weight well" (people were often surprised to find out exactly how heavy I was). In high school my weight was 78 kg and my height was 1.78 m, giving me a BMI of 24.6, which was at the top end of 'healthy' (18.5-25.0). No surprise there, as although I did OK on fitness tests at the time, I didn't play a lot of sports (just a bit of swimming and Judo) and couldn't do a chin-up to save my life.

And by the time I started this current stint of diet/lifestyle change my weight had slowly increased over the decades to reach a personal maximum of 111.8 kg (a BMI of 35.3) - which was just entering the morbid obesity range (if you have any obesity-related health conditions such as high blood pressure or diabetes). As I'd been on blood pressure medication for several years and my recent blood sugar test results had moved from the high end of 'normal' and into that grey area where type 2 diabetes is a distinct possibility (but can be managed by a combination of regular physical exercise, healthy eating and weight reduction) it was definitely time to do something permanent about my poor diet and lack of exercise...

So, based on the generic BMI guidelines, my initial 'goal' for weight loss was to get down into the 'healthy' range of BMI 18.5-25.0, which for my height would mean a weight between 59 kg (!!) and 79 kg. I had started out assuming I would stick to my CR diet and two days of fasting each week until I got under 80 kg, and then taper off the fasting to stabilize my weight in the 72-76 kg range. However, as my body type seems to be naturally mesomorph (if I lose the excess body fat and do modest amounts of weight training) I suspect that by the time I get down to around 80 kg my body fat will have already reduced to the lower end of what is considered a health body fat percentage (8%-19%) for men.

My home scales purport to give a reading of body fat%, but the reading fluctuates wildly from day to day, so only the 21-day moving average gives a reasonably consistent reading (this doesn't mean that it is very accurate though!). Plotting body fat % vs. BMI gives an interesting result, especially when compared to some published data:

The blue dots are my data points, whereas the black circles are female data and black crosses are male data from a study of Sri Lankan men and women. Admittedly I don't belong to that particular racial group, which may well explain why my data points are outliers (assuming my body fat data is even accurate), but it could also be due to the know inaccuracy of BMI as a predictor of body fat in those that have a muscular build (athletes or body builders for example).

One published formula for estimating body fat percentage from BMI data for caucasians is:

BF% = -38 + 2.2 x BMI (+/-4.8)
(for 50-60 year olds: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3183503 table 4)

which for my current BMI of 30.3 predicts a BF% = 28.7 +/- 4.8  i.e. 23.9% - 33.5%

This puts my current body fat reading of around 20% well below what is expected from my BMI.

Either my scales are ridiculously inaccurate, or my weight training is making my body fat reduce at a greater rate than would be achieved by calorie restriction alone.

Therefore, as I continue to lose weight and do weight training I'll start tracking body fat using abdominal skin fold measurements (if I can find the calipers I bought several years ago!). There are several different formulae available for converting skin fold measurements to estimated body fat, I'll probably try the various ones available at http://www.linear-software.com/online.html and see how the results compare. Once I start getting skin fold measurements I'll be able to tell if I'm really shifting my muscle:fat ratio (as I suspect), or if my bathroom scales are just giving me consistently inaccurate body fat readings...

Subscribe to Enough Wealth. Copyright 2006-2019