Two separate experiments were undertaken. In the hydration study, changes in body composition were induced by intravenous infusion of PN in underweight patients with short bowel syndrome. In the lard study, changes were induced by placing packets of lard on lean healthy volunteers.
All participants were weighed (after voiding) on a calibrated digital scale accurate within 0.1 kg (subjects wearing light clothes). The heights were measured after a maximal inhalation to the nearest 0.1 cm by using a wall-mounted stadiometer. The averages of two measurements for both height and weight were used as the criterion measurement. The body mass index was calculated as weight divided by height squared (kg/m2).
Dual energy X-ray absorptiometry
Measurements of body composition were performed with the Norland XR-36 DXA densitometer (Norland Corporation, Fort Atkinson, WIS, U.S.A.) with the subject supine. The host software was rev. 2.5.2. and the scanner software rev. 2.0.0. The theory and methodology for body composition by DXA has previously been described . Briefly, while the patient lightly dressed lay on a scan table for about 20 min., transverse scans approximately 1 cm apart were performed from top to heel. The instrument uses X-rays of two distinct energy levels that are attenuated by fat, bone and lean mass to different extent. By computerization of data inputs from approximately 11000 pixels DXA estimates body composition based on a three-compartment model, measuring TBBMC, FM, and LTM. The total mass (TM) by DXA is the sum of all three compartments, whereas the soft tissue mass (STM) by DXA includes only the LTM and FM. The fat free tissue mass (FFM) includes both the TBBMC and LTM. Among others, Hendel et al.  have reported precision errors of body composition of the Norland XR-36 densitometer. They were 2.2% for TBBMC, 2.7% for FFM, and 2.6% for FM%. In our hands the between-measurements CV%'s of TBBMC, LTM and FM, were 1.5%, 1.6%, and 3.9%, respectively .
This study comprised 11 patients (9 women and 2 men) with short bowel syndrome treated with daily supplements of PN. The participants were selected for low body weight (BMI < 22 kg/m2). The diagnoses were: Crohn's disease (n = 6), ischaemic infarction (n = 2) and other (n = 3). Patients were on average (mean ± SD) 49.5 ± 17.1 yr., 1.58 ± 0.07 m, 48.5 ± 9.5 kg, and 19.3 ± 3.1 kg/m2. The PN was in all patients provided as a 3 L 'all in one' plastic bag containing a fixed composition of protein, glucose, and electrolytes, and four patients had additional supplements of 1–2 L of saline. The infusion was given continuously over a period of 8–10 hours during the night. Before starting the infusion all patients were weighed on a scale and scanned as described below. Immediately after completing the infusion the patients were reweighed and rescanned. Due to the large volume of intravenous fluid provided with the parenteral nutrition, patients were allowed to void during the study.
The theoretical soft-tissue attenuation (RST) of an 'all in one' 3 L bag of PN was calculated using the equation RST = Σ (-fi × (μmi)L) / Σ (-fi × (μmi)H), where (fi) is the mass fraction, (μmi) is the mass attenuation coefficient of the i'th component at high (H) and low (L) photon energy levels . The theoretical RST-value for PN was calculated to 1.365, which is very close to that of normal saline (1.377). Given the calculated RST-value PN should theoretically be scanned by DXA as consisting of approximately 2% FM and 98% LTM. Such values were confirmed in vivo by scanning one subject three times before and after placing 2 bags (6.96 kg) of PN on the subject's legs. By DXA, the equivalent 6.99 kg increase in TM (TM) resulted from a 7.31 kg gain of LTM (104.5%) and a 0.32 kg loss of FM (- 4.6%), whereas the TBBMC changed only 6 g (0.1%).
For this experiment two packets of porcine lard (with a small amount of muscle-tissue attached) were constructed and enclosed in plastic wrap. The lard packet dimensions were 19.8 cm × 38.8 cm × 2.6 cm, and weighed, using a beam scale, 3.49 kg. The total fat fraction of the lard, measured in triplicate by chemical fat extraction according to the method of Folch et al. , was 52.2% (CV% = 6.4%). The participants for this study were selected for low body weight (BMI < 22 kg/m2). Eight healthy lean male volunteers, who were on average (mean ± SD) 26.4 ± 7.4 years of age, agreed to participate. Anthropometric measures of the participants were taken immediately before the study and averaged 1.81 ± 0.07 m in height, 69.0 ± 7.7 kg in weight, and 21.0 ± 0.9 kg/m2 in BMI. Without reposition, four consecutive total body DXA scans were performed on each participant. Two scans without added lard served as baseline measurements (the average values were used as the criterion measurements), and two scans were performed with the lard packets placed alternately on the abdomen centred over the lumbar vertebrae, and on the thighs at midpoint of the femur. The placements of the lard over the thighs and trunk were chosen to represent regions where the ability of DXA to correctly measure soft tissue composition is known to be good and poor, respectively.
The Ethics Committee for Medical Research in Copenhagen, Denmark, approved the study protocol and the study was conducted in accordance with the Declaration of Helsinki of 1975, as revised in 1983. Written and oral informed consent was obtained from all patients prior to inclusion.
All results are expressed as means ± standard deviation (SD) unless otherwise indicated. A paired Students t-test was used to compare paired variables. Association between variables was established by Pearson's correlation coefficients and linear regression. The CV%'s for the measurements of TBBMC or FM were calculated from the within-subjects SD's divided by their respective grand means. All statistical tests were two-tailed, and a p value of less than 0.05 was considered statistically significant. The SPSS statistical program version 10.0 (SPSS Inc., Chicago, USA) was used for all analyses.