The prediction of abdominal visceral fat level from body composition and anthropometry: ROC analysis

To evaluate the merit of body mass index (BMI), % body fat, waist circumference and waist-to-hip ratio as predictors of abdominal visceral fat (AVF) level.

Cross-sectional measurements obtained from 458 female and 331 male subjects of French Canadian descent with an age range from 18-72 y.

AVF level was assessed by computed tomography. BMI was calculated as weight (in kg) divided by stature2 (in m), body density was derived from underwater weighing and % body fat was computed from the estimate of body density with the Siri equation. Waist-to-hip ratio was calculated as waist circumference divided by hip circumference. Receiver operating characteristic (ROC) curves were used to identify the optimal cut-off points.

In younger women (< 40 y, n = 258), waist-to-hip ratio was the poorest predictor of AVF level with areas under the ROC curves (Az) ranging from 0.684-0.716, sensitivity (Sen) from 63.3-68.8% and specificity (Spe) from 64.0-67.5%, whereas the Az, Sen and Spe for other predictors ranged from 0.924-0.983, 87.0-96.8, and 83.4-92.7, respectively. The same trend was observed in older (> or = 40 y, n = 200) women, although differences between waist-to-hip ratio and other predictors were less pronounced. In older men, waist circumference was the best overall predictor (Az from 0.88-0.92), whereas BMI showed the lowest Az values (0.831-0.875, P < or = 0.001 vs waist circumference). In younger men, BMI had the smallest Az (P < 0.007 vs others) with the lowest AVF cut-off point (100 cm2). However, with higher AVF cut-offs the differences were not significant.

Waist circumference is the best overall predictor of abdominal visceral obesity, whereas in women waist-to-hip ratio is a poor indicator of AVF and its use as a surrogate measure of visceral fat should be avoided.