Background
The measurement of waist circumference for assessing abdominal obesity has become widely accepted.
Material and Method
To evaluate the reliability of measuring waist circumference, anthropometric parameters were measured in 150 adult patients by two diabetes educators on two consecutive days. The intraobserver difference was defined as the difference in the measurements made by the same educator on different days and the interobserver difference as the difference in the measurements made by the two educators on the same days.
Results
The correlation coefficients (r values) of the measurements were >0.99 for the educators and days. The interobserver difference was statistically significant for waist circumference (96.23 cm vs. 97.08 cm, p<0.0001). Although the %Delta values (percent difference in the two means) for waist circumference proved to be 2.5-6.3 times higher and the% absolute Delta values (percent average difference) 1.5-2.8 times higher than those for body mass index, the absolute values of the differences (Delta, the difference of the two means, and absolute delta, the average difference) were small (waist circumference Delta: 0.17 cm and 0.85 cm, absolute Delta: 1.51 cm and 2.15 cm; body mass index Delta: 0.02 kg/m(2) and 0.04 kg/m(2), absolute Delta: 0.292 kg/m(2) and 0.226 kg/m(2); intraobserver and interobserver differences, respectively).
Conclusions
The intraobserver and interobserver variability for waist circumference were higher than those for body mass index. Nevertheless, the differences in repeated measurements of waist circumference were small when expressed in absolute values. The reliability of waist circumference should be considered in the clinical practice.

Keywords: Waist-Hip Ratio, Obesity - diagnosis, Observer Variation, Body Mass Index, Aged, 80 and over, Anthropometry, Adiposity, Abdominal Fat - pathology