Can weight-related health risk be more accurately assessed by BMI, or by gender specific calculations of Percentage Body Fatness?

The problem of obesity over the last 10 years has consistently been referred to as a 'global epidemic'. The Body Mass Index (BMI) is the currently accepted measure for classifying weight-related risk, but is a crude measure that has not changed in 150 years. It is recognised as having significant limitations, largely due to its lack of distinction between fat and muscle tissue. As the health risks of obesity are linked to the fat content of the body, a more accurate method of classifying would be Percentage Body Fatness (PBF). Although skinfold thickness analysis is recognised as a valid and accurate estimate of PBF in field studies, this method is not routinely used in clinical practice. Using data collected from young adults in the United Kingdom, we compared classifications (underweight, normal weight, overweight and obese) using BMI, with classifications using estimated PBF (from skinfold thickness analysis). We identified disparity between these two methods in approximately 1/3 of participants. BMI correctly classified 66.5% of females and 62.7% of males, with different gender profiles of incorrect classification. Regression analysis was conducted using estimated PBF (by skinfold thickness analysis) as the dependent variable, with explanatory variables of age, height, weight, systolic blood pressure, frequency of vigorous exercise and grip strength. The resulting gender-specific formulae derived from this regression analysis provides a regression R(2) of around 65%, and improved correct classifications to 74% for females and 76% for males. This represents an average improvement of roughly ten percentage points over BMI (male: 7.2% points; female: 13.4% points). We hypothesise that the presented formulae provide gender-specific calculations of PBF, which result in a more accurate indicator of weight-related health risk, compared with BMI in this population. This provides a new approach to an increasingly important clinical issue. These formulae use data that can be easily, quickly and cost-effectively measured in a practice setting. If shown to be repeatable with larger and more diverse populations, the PBF formulae could provide an alternative to the BMI as the major indicator of body-composition related health risk. This would ensure resources are targeted more appropriately and efficiently.