Get access

Reduction of misclassification rates of obesity by body mass index using dual-energy X-ray absorptiometry scans to improve subsequent prediction of per cent fat mass in a Caucasian population


Dr S Pedersen, LMC Endocrinology Centre, Suite 102, 5940 Macleod Trail SW, Calgary, AB, Canada T2H 2G4. E-mail:


What is already known about this subject

  • • Body mass index (BMI) is not accurate in the classification of excess body fat, failing to identify as many as half of individuals with excess per cent fat mass.
  • • Normal-weight obesity, which goes undiagnosed when BMI is the only measure of adiposity utilized, has been shown to be associated with an increased risk of cardiovascular comorbidities and mortality.
  • • Dual-energy X-ray absorptiometry (DXA) is an accurate and relatively inexpensive method for indirect assessment of body composition.

What this study adds

  • • The formulae developed allow the clinician to utilize information from one baseline DXA scan to calculate a patient's per cent fat mass with a future change in weight, thus allowing the clinician to more accurately determine whether and when an individual patient should be classified as obese and thus be managed appropriately.
  • • The formulae developed enable the clinician to calculate a patient-specific BMI treatment goal, below which the patient would no longer meet the per cent fat mass criteria for obesity.


Recognition is increasing for the errors of body mass index (BMI) in classification of excess body fat. Dual-energy X-ray absorptiometry (DXA) is accurate to assess body fat mass per cent (%FM), but is underutilized clinically. We examined the prevalence of obesity misclassification by BMI in comparison to body %FM by DXA scanning, and whether there is a time-stable individual relation between the %FM and the BMI in patients scanned several times. We aimed to develop a formula where, based on a single DXA scan, %FM could be predicted following a change in weight, and a patient-specific BMI threshold could be calculated (BMIT), above which the patient would be obese by %FM criteria. Data were collected from individuals who had a DXA scan as part of a nutritional research study at the University of Copenhagen. BMI incorrectly classified 48/329 (14.6%) of men and 52/589 (8.8%) of women. The majority of men with BMI 25–27 kg m−2 and women with BMI 24–26 kg m−2 were misclassified. Using multiple scan data (189 men, 311 women) and calculating the patient-specific constant C = (1 − %FM/100)3/2 × BMI from baseline BMI and %FM, misclassification rates were halved for both genders when a personal threshold, BMIT, was used ([BMIT = C/(0.75)3/2] for men and [BMIT = C/(0.65)3/2] for women). We conclude that simple formulae allow evaluation of fatness of individual patients more accurately than BMI, with the use of one baseline DXA scan, and enable the establishment of patient-specific obesity treatment targets in clinical practice.