Background: In 74 women with anorexia nervosa (body mass index [BMI] 10-17.5 kg/m(2)), a progressive disagreement between anthropometry and bioelectrical impedance analysis (BIA) estimates of fat (FM) and fat-free mass (FFM) was documented with a BMI <15 kg/m(2). Below this threshold, an abnormal body composition was detected with vector BIA independent on body weight (useful tool at the bedside). Both anthropometry and BIA are used for body composition assessment. We evaluated the agreement between their estimates of FFM and FM in extremely lean subjects with anorexia nervosa.
Methods: Observational study in 74 women (age 15-45 years; BMI 10.0-17.5 kg/m(2)) with anorexia nervosa. Anthropometry FM and FFM were estimated by skinfold thicknesses. Whole-body impedance vector components, resistance (R) and reactance (X(c)), were measured at the same time (BIA-101 analyzer, 50-kHz frequency; Akern/RJL Systems, Clinton Twp., MI). BIA estimates of FFM and FM were obtained with Sun's equations (BMI 14-39 kg/m(2)). Vector BIA was performed with the RXc graph method. The frequency of disagreement between anthropometry and BIA was evaluated as a function of increasing BMI (receiver operating characteristic curve).
Results: The correlation coefficient between methods was 0.80-0.90. The BMI cutoff value of 15 kg/m(2) predicted disagreement of FM and FFM in patients with a BMI <15 kg/m(2). In the RXc graph, the distribution of vectors from patients with disagreement was significantly shifted out of the right border of the reference tolerance ellipses.
Conclusions: Anthropometry and BIA equations should not be used in anorexic patients with a BMI <15 kg/m(2). The distribution of vectors from patients with disagreement of methods was shifted out of the right border of the reference tolerance ellipses and was superposed to vectors from patients with a BMI <15 kg/m(2). The identified impedance pattern for anorexia could be useful for feedback in individual patient refeeding (vector back into the 75% reference ellipse).