In this week's 'Scrubbing Up', nutrition expert Dr Margaret Ashwell OBE argues that using BMI alone could be missing some patients at risk of the diseases associated with obesity.
BMI is calculated by dividing someone's weight in kilograms by the square of their height in metres.
But because BMI is based on weight, people who are unusually muscular can have BMI values which would make them overweight or obese. Examples might include professional rowers and boxers.
Different types of fat
There's a second, even more important, problem. BMI is a good proxy for total body fat, but it cannot distinguish between different types of fat distribution.
Excess fat that is found deep down in the region of the stomach gives someone a large waist circumference and an 'apple' shape. This is often associated with risk factors for serious conditions such as heart disease, raised blood pressure and diabetes.
Excess fat that is found under the skin, around the bottom, hips and thighs is usually accompanied by a smaller waist circumference and a 'pear' shape. This is generally accepted to be less harmful to health.
So a measurement that can distinguish between 'apples' and 'pears' would be a more effective way of screening for these diseases.
I have argued since 1996 that we should assess risk based on waist-to-height ratio (WHtR); saying that "Your waist circumference should not be more than half your height (WHtR 0.5)".
This is the point at which some action to decrease your waistline should be considered.
My colleagues and I recently published a review of 78 studies in 14 different countries, including Caucasian, Asian and Central American subjects, which has confirmed that WHtR is a better predictor of cardiometabolic risk than BMI and that WHtR 0.5 is a suitable boundary value.
Simple risk assessment
The WHtR measurement offers the exciting possibility that it could be used to assess risk for adults in several ethnic groups. This is because the BMI boundary values were derived from Caucasian populations.
They might not be appropriate for Asians, who tend to carry more fat centrally so their risk seems to increase at a BMI lower than 25 (between 20 and 25 is considered a healthy BMI).
WHtR might also be a simple way to assess health risk in children.
And of course waist circumference and height can be measured with a simple tape measure in any units: inches, centimetres etc, unlike the BMI which requires weighing scales and must be expressed in metric units.
The public health message is simple - "keep your waist circumference to less than half your height".
Missing the risks
So what would be the difference in screening for central obesity using WHtR, instead of screening for total obesity using BMI?
In 2009 we published a paper in which we applied the WHtR boundary value of 0.5 to nearly 2,000 British adults in the nationally representative National Diet and Nutrition Survey.
We found that one in three 'non-overweight' men (judged by a BMI of 25 or less) and more than one in seven 'non-overweight' women had a WHtR value greater than 0.5.
So these people may, unknowingly, be at increased risk of serious conditions such as heart disease and diabetes, due to their central obesity.
These are the 'non-overweight apples'.
On the other hand, one in six women and one in twenty men who were told they were overweight by BMI assessment, would actually have relatively low cardiometabolic risk because their WHtR is 0.5 or less.
These are the 'overweight pears'.
Most interesting of all, the 'non-overweight apples' had significantly higher levels of cardiometabolic risk factors (blood pressure and non-HDL cholesterol) than the 'overweight' pears'.
It is a real worry that using BMI alone for screening could miss people who are at risk from central obesity and might also be alarming those whose risk is not as great as it appears from their BMI.