The association of early postnatal growth with diseases in adults such as hypertension, type 2 diabetes and CHD has generated interest in studying postnatal growth. Bioelectrical impedance analysis (BIA) is a useful measure to estimate total body water (TBW) and fat-free mass (FFM). We evaluated three published equations (Fjeld et al. (Pediatr Res (1990) 27, 98–102), Bocage (MSc Thesis (1988) University of West Indies) and Kushner et al. (Am J Clin Nutr (1992) 56, 835–839) to measure TBW and derived FFM based on BIA, using 2H2O dilution as a reference method for suitability in infants in India. In a cross-sectional study in seventy-eight apparently healthy infants aged 6–24 months from the urban poor attending an immunisation clinic at a hospital in Kolkata, we measured their length to the nearest 0·1 cm, weight to the nearest 10 g, resistance at 50 kHz using BIA and TBW using 2H2O dilution. TBW was derived using three published BIA-based equations and compared with TBW using 2H2O dilution. Based on the BIA equations of Fjeld et al., Bocage and Kushner et al., the mean TBW values were 2·46 % (P < 0·001), 4·62 % (P < 0·001) and 9·50 % (P < 0·001) lower than the reference 2H2O method, respectively. All three published BIA-based equations consistently underestimated the TBW and FFM and appeared inadequate for studying infants in India. The equation described by Fjeld et al. gave the smallest deviation from the reference method and may be used for field studies. New equations based on population-specific data are desirable for a more precise measure of TBW.

The contribution of gluconeogenesis to glucose production is estimated from the enrichment of the H bound to C-5 of glucose relative to either that bound to C-2 of glucose or the enrichment in body water on ingesting 2H2O in the fasted state. Contributions of all gluconeogenic substrates are included in the estimate and the limitation of an uncertain precursor enrichment removed. The half-life of 2H2O in body water precludes a repeat study for many weeks. Glycogen cycling could result in underestimation, but there is evidence that glycogen cycling does not occur in liver in the fasted state. Gluconeogenesis has been estimated by mass-isotopomer-distribution analyses, usually by administering 13C-labelled glycerol. Underestimates emphasize the major limitation of the method, i.e. the need to assume a single enrichment of the precursor pool. Estimates of gluconeogenesis from isotopomer distribution in arterial-blood glucose and lactate on infusing [U-13C6] glucose are unreliable, as a proportion of the glucose is formed from glycerol and from amino acids not converted to glucose via pyruvate. Loss of label in the Krebs cycle and relying on enrichment of arterial-blood lactate as a measure of hepatic pyruvate further add to the uncertainty. Estimates of the rate of gluconeogenesis by NMR are obtained by subtraction of the rate of glycogenolysis determined by NMR from the rate of glucose production. Estimates are then the mean rate for the period over which glycogen contents are measured. Technical considerations can limit the accuracy of analyses and result in overestimates.