Thursday, January 3, 2013

BMI, Overweight, and Longevity

     A recent article in the Journal of the American Medical Association (JAMA, 2013;309: 71-82) created quite a stir when it concluded that adults who were overweight but not obese had a 6% lower all-cause mortality rate. Nutritionists rushed to say that this did not mean that you should gain weight, although if the study had shown the reverse I am sure they would have unanimously recommended weight loss. This was not the first study to reach this conclusion, so I thought I should review it here, along with examining exactly what the BMI purports to measure since all obesity labels were based on the BMI (Basal Metabolic Index) results.
     The BMI was first defined by Quetelet in the 1800's as a surrogate number for estimating the amount of body fat. It is defined as mass(kg)/height (m) x height, i.e. the mass divided by the square of one's height. Since your mass increases with your total body volume, it scales with the cube of your height, so it is immediately obvious that only dividing by the square of the height will overestimate the total body fat of taller individuals. Similarly, muscular athletes have very little total body fat, but since muscle has mass, the BMI will  overestimate their total body fat as well. It is trivial to note that since most elderly patients have some degree of osteoporosis and therefore lighter bones, the BMI will mistakenly classify some of them as underweight.
I should also mention here that in 1998 the U.S. lowered its cutoff for the lower limit of overweight from a BMI of 27.8 to 25.0, in accordance with the WHO standards, thereby immediately reclassifying millions of Americans as overweight who the day before the change had "normal" weight. In addition, doctors in the US do not agree on the healthy lower limit of the BMI  for adult women, and the WHO does not have the classification of "underweight". And the 0.5 to 1.0 inches in height that you lose with age increases your BMI without you actually gaining any weight.

     The consensus in the US is that a BMI of 18.5 to 25 is normal weight, 25 to 30 is overweight, and greater than 30 is obese. Using these definitions, there have already been two published studies, one of diabetic patients and one of patients with ASCVD, and both demonstrated that overweight patients had a lower death rate than both normal weight and obese patients. No one has an explanation for these results, nor have I even seen theories that attempt to explain them. The JAMA study looked at 97 studies of all-cause mortality, encompassing 270,000 deaths and over 2,800,000 subjects, so it is fairly comprehensive.
We are left with the paradox that being overweight by BMI standards statistically increases your risk for high blood pressure, diabetes, ASCVD, stroke and some cancers, yet overall it is protective of mortality.

     In a previous blog I listed the 11 countries in order of longevity. I will repeat the list here adding the mean BMI of each country. Except for Japan, #1 and with the lowest BMI, there is no clear pattern:
Japan---22, Switzerland---25, Australia---26, Italy 23.5, Israel---25, Iceland---26, Spain---24.5, France---23.5, Canada---25.5, Singapore---22, and New Zealand---26.6.

     So once again we are left with "a beautiful theory ruined by an ugly fact" to quote a Nobel scientist. Perhaps we should re-normalize the BMI values and re-define normal weight. Or perhaps we should have a separate classification of "healthy BMI" and "unhealthy BMI". Or, more likely, there is a confounding fact of which we are unaware which would explain the results, but so far there are no candidates. It does make me feel better with my BMI of 27 because at 6'2" I doubt that I could get down to 185 pounds, even if I wanted to. I think we underestimate the genetic effects on us of our parents' diseases, especially parents of the same sex.


  1. I just want to re-emphasize that the BMI is a surrogate number to guesstimate the total body fat without actually measuring it by immersing the body in a pool of water. The number says nothing about the distribution (apple-shaped vs. pear-shaped) distribution of total body fat, or about yellow vs. brown fat, or about the muscle-to-fat ratio. And again, the taller or the more muscular you are, the more the BMI overestimates your total body fat, so that the IBW calculated from the BMI and height in tall people is too low, and the taller you are compared to average body height, the greater the error in the calculated Ideal Body Weight.

  2. I, too, have a BMI of 27, so your blog post provides some comfort from knowing that my weight is apparently not deleterious to my health. As for those BMI statistics from other countries, I suspect that if one were to remove people under the age of 50 from the calculations, the average BMI's for each country would increase.

    I walk regularly, jog and lift weights so my BMI may be misleading. Are there any good statistics for longevity and body fat percentage, or overall health benefits from various ranges of body fat percentages?

    Roy K

  3. There is almost no data on body fat percentages, which would require detailed measurements of the body. As I have indicated, the BMI is a surrogate measurement of body fat, and it overestimates it for taller and for more muscular people.