Objectives: (1) To develop a method of manipulating bioelectrical impedance (BIA) that gives indices of lean and fat adjusted for body size, using a large normative cohort of children. (2) To assess the discriminant validity of the method in a group of children likely to have abnormal body composition.
Design: Two prospective cohort studies.
Setting: Normative data: Avon Longitudinal Study of Parents and Children (ALSPAC), population based cohort; proof of concept study: tertiary feeding clinic and special needs schools.
Subjects: Normative data: 7576 children measured aged between 7.25 and 8.25 (mean 7.5) (s.d.=0.2) years; proof of concept study: 29 children with either major neurodisability or receiving artificial feeding, or both, mean age 7.6 (s.d.=2) years.
Measures: Leg-to-leg (Z (T)) and arm-to-leg (Z (B)) BIA, weight and height. Total body water (TBW) was estimated from the resistance index (RI=height(2)/Z), and fat-free mass was linearly related to TBW. Fat mass was obtained by subtracting fat-free mass from total weight. Fat-free mass was log-transformed and the reciprocal transform was taken for fat mass to satisfy parametric model assumptions. Lean and fat mass were then adjusted for height and age using multiple linear regression models. The resulting standardized residuals gave the lean index and fat index, respectively.
Results: In the normative cohort, the lean index was higher and fat index lower in boys. The lean index rose steeply to the middle of the normal range of body mass index (BMI) and then slowly for higher BMI values, whereas the fat index rose linearly through and above the normal range. In the proof of concept study, the children as a group had low lean indices (mean (s.d.) -1.5 (1.7)) with average fat indices (+0.21 (2.0)) despite relatively low BMI standard deviation scores (-0.60 (2.3)), but for any given BMI, individual children had extremely wide ranges of fat indices. The lean index proved more stable and repeatable than BMI.
Conclusions: This clinical method of handling BIA reveals important variations in nutritional status that would not be detected using anthropometry alone. BIA used in this way would allow more accurate assessment of energy sufficiency in children with neurodisability and may provide a more valid identification of children at risk of underweight or obesity in field and clinical settings.