British Journal of Nutrition

Full Papers

Innovative Techniques

Computed tomography-based validation of abdominal adiposity measurements from ultrasonography, dual-energy X-ray absorptiometry and anthropometry

Anna M. I. Gradmarka1, Anders Rydha2, Frida Renströma1, Emanuella De Lucia-Rolfea3, Alison Sleigha4, Peter Nordströma5, Soren Bragea3 and Paul W. Franksa1 c1

a1 Genetic Epidemiology and Clinical Research Group, Department of Public Health and Clinical Medicine, Division of Medicine, Umeå University Hospital, Umeå 90 187, Sweden

a2 Department of Radiology, Umeå University Hospital, Umeå, Sweden

a3 MRC Epidemiology Unit, Institute of Metabolic Science, Cambridge, UK

a4 Wolfson Brain Imaging Centre, Addenbrooke's Hospital, Cambridge, UK

a5 Department of Surgery and Perioperative Sciences, Umeå University Hospital, Umeå, Sweden


Large-scale aetiological studies of obesity and its pathological consequences require accurate measurements of adipose mass, distribution and subtype. Here, we compared the validity of three abdominal obesity assessment methods (dual-energy X-ray absorptiometry (DXA), ultrasound and anthropometry) against the gold-standard method of computed tomography (CT) in twenty-nine non-diseased middle-aged men (BMI 26·5 (sd 3·1) kg/m2) and women (BMI 25·5 (sd 3·2) kg/m2). Assessments of adipose mass (kg) and distribution (total subcutaneous (TSAT), superficial subcutaneous (SSAT), deep subcutaneous (DSAT) and visceral (VAT)) were obtained. Spearman's correlations were performed adjusted for age and sex. VAT area that was assessed using ultrasound (r 0·79; P < 0·0001) and waist circumference (r 0·85; P < 0·0001) correlated highly with VAT from CT, as did BMI (r 0·67; P < 0·0001) and DXA (r 0·70; P < 0·0001). DXA (r 0·72; P = 0·0004), BMI (r 0·71; P = 0·0003), waist circumference (r 0·86; P < 0·0001) and ultrasound (r 0·52; P = 0·015) were less strongly correlated with CT TSAT. None of the comparison measures of DSAT was strongly correlated with CT DSAT (all r approximately 0·50; P < 0·02). BMI (r 0·76; P < 0·0001), waist circumference (r 0·65; P = 0·002) and DXA (r 0·75; P < 0·0001) were all fairly strongly correlated with the CT measure of SSAT, whereas ultrasound yielded a weaker yet statistically significant correlation (r 0·48; P = 0·03). Compared with CT, visceral and subcutaneous adiposity can be assessed with reasonable validity using waist circumference and BMI, respectively. Ultrasound or DXA does not generally provide substantially better measures of these traits. Highly valid assessments of DSAT do not appear to be possible with surrogate measures. These findings may help guide the selection of measures for epidemiological studies of obesity.

(Received October 19 2009)

(Revised February 04 2010)

(Accepted February 09 2010)

(Online publication April 07 2010)