ORIGINAL ARTICLE. SA Jebb 1, M Siervo 1, PR Murgatroyd 2, S Evans 1, G Frühbeck 3 and AM Prentice 4. Introduction

Transcription

1 (2006) 1 7 & 2006 Nature Publishing Group All rights reserved /06 $ ORIGINAL ARTICLE Validity of the leg-to-leg bioimpedance to estimate changes in body fat during weight loss and regain in overweight women: a comparison with multi-compartment models SA Jebb 1, M Siervo 1, PR Murgatroyd 2, S Evans 1, G Frühbeck 3 and AM Prentice 4 1 MRC Human Nutrition Research, Elsie Widdowson Laboratory, Fulbourn, Cambridge, UK; 2 Wellcome Trust Clinical Research Facility, Addenbrooke s Hospital, University of Cambridge, Cambridge, UK; 3 Department of Endocrinology, Clínica Universitaria de Navarra, Pamplona, Spain and 4 MRC International Nutrition Group, London School of Hygiene and Tropical Medicine, London, UK Objectives: To investigate changes in body composition and the validity of the leg-to-leg bioimpedance (LTL) method to measure body fat during active weight loss (WL) and weight regain (WR). Design: Longitudinal, 12-week weight loss intervention ( MJ/day) and subsequent follow-up at 1 year. Subjects: Fifty-eight adult women aged between 24 and 65 years (mean age: years) and with a body mass index (BMI) X25 kg/m 2 (mean BMI: kg/m 2, range ¼ kg/m 2 ) participated in the study. Measurements: Fat mass (FM) was measured at baseline, 12 weeks, 24 weeks and 52 weeks using three- and four-compartment (4-C) models, air displacement plethysmography (ADP), deuterium dilution total body water (TBW), dual-energy X-ray absorptiometry (DXA), skinfold thickness (SFT), tetrapolar bioelectrical impedance analysis (T-BIA) and LTL. Results: At the end of the weight loss programme, subjects lost kg weight (Po0.001) and kg fat (Po0.001) but after 1 year they had regained kg of weight and kg of fat. The 4-C model showed that FM and TBW accounted for 76.2 and 23.6% of the loss in body mass and 81.8 and 17.7% of the tissue accrued during weight regain, respectively. The estimate of body fat change by LTL relative to multi-compartment models (WL bias72s.d. ¼ kg; WR bias72s.d. ¼ kg) was similar to ADP, DXA and TBW in both phases but it was better than T-BIA (WL bias72s.d. ¼ kg; WR bias72s.d. ¼ kg) and skinfold thickness (WL bias72s.d. ¼ kg; WR bias72s.d. ¼ kg). Conclusions: Weight loss and regain were associated with minimal changes in lean tissue as measured using multicompartment models. The LTL system is a useful method to measure body composition changes during clinical weight management programmes. advance online publication, 24 October 2006; doi: /sj.ijo Keywords: leg-to-leg bioimpedance; weight loss; weight gain; body composition; multi-compartment models; fat mass; humans Introduction Obesity, usually defined on the basis of body mass index (BMI) or waist circumference, is linked to an increased risk of metabolic diseases but this relationship is mediated by excess Correspondence: Dr SA Jebb, MRC Human Nutrition Research, Head of Nutrition and Health Research MRC, Elsie Widdowson Laboratory, Fulbourn Road, Cambridge, Cambridgeshire CB1 9NL, UK. susan.jebb@mrc-hnr.cam.ac.uk Received 2 March 2006; revised 2 August 2006; accepted 13 September 2006 body fat. The precise assessment of fat mass (FM) and its distribution is critical to study the association between adiposity and risk of metabolic diseases (diabetes, hypertension and hyperlipidaemia) and generate public health guidelines. 1,2 Several body composition methods are available which differ in terms of their theoretical basis and scientific assumptions, as well as cost, complexity and subject acceptability. Body composition methods such as dual energy X-ray absorptiometry (DXA), air displacement plethysmography (ADP) and stable isotope dilution (ID)

2 2 methods are mainly confined to research settings because of their cost and high level of expertise required. 3 5 The need for simpler methods, better suited to large-scale epidemiological studies or individual studies outside specialist centres, led to the development of predictive techniques such as those based on skinfold thickness (SFT) and more recently bioelectrical impedance (BIA). BMI is often used as an index of adiposity but in practice it makes fundamental assumptions about body composition which limit its applicability. 6 Most validation studies of these simple methods have focused on cross-sectional comparisons However, in many situations it is the change in body composition which is of primary importance. For example, the increase in body fat with age, or changes in composition occurring during weight loss interventions or as a consequence of diseases or drug treatments This study examines the validity of the leg-to-leg BIA (LTL) device against a multi-compartment model during a period of weight loss and weight regain. Previous studies have shown reasonable agreement in cross-sectional analyses in different populations and situations 10,13 16 but it is possible that the change in fatness may be measured with greater accuracy as some of the errors may cancel out. In addition this study explores the composition of the change in body composition associated with a period of very low calorie dieting (VLCD). It has been previously suggested that VLCDs are associated with excessive loss of lean tissue but most studies have used a two-compartment model. In this model acute changes in hydration will appear as fluctuations in LBM, yet this may not provide a true reflection of the changes in protein stores Methods Subjects Fifty-eight adult women aged between 24 and 65 years (mean age: years) and with a BMI X25 kg/m 2 participated in the study. Women were excluded if they had smoked or had lost 43 kg in the last 2 months or had any of the following conditions: diabetes, endocrinological diseases affecting food intake or weight regulation, secondary causes of obesity, long-term treatment with drugs affecting food intake or energy metabolism. All subjects were Caucasians and resident in the Cambridge area. The protocol was approved by the ethical committee of the MRC Dunn Nutrition Unit. Protocol Subjects took part in a 12-month study consisting of an initial weight loss phase and subsequent follow-up. The weight loss programme comprised an 8-week period of a milk-based diet providing approximately MJ/day. Volunteers were instructed to consume three pints of semiskimmed milk daily, plus one salty low-energy drink and at least two pints of other low energy fluid. They received a multivitamin and iron supplement. From weeks 9 to 12, food was gradually re-introduced into the diet and the milk ration was decreased, so that by the end of week 12 subjects were prescribed a food-based diet providing sufficient energy to match their estimated maintenance energy requirements. Subjects attended fortnightly dietician-led group sessions at which they were weighed and received basic information and advice on a low-fat, energy-reduced diet, physical activity and behavioural techniques for weight control. A more detailed description of the protocol is published elsewhere. 21 Measurements of body composition were made at weeks 0, 12 and later at 24 and 52 weeks. Anthropometry Weight was measured to the nearest 0.1 kg (Sauter E1210, Todd scales, Suffolk, UK) using a digital scale. Height was measured using a wall-mounted stadiometer to the nearest 0.1 cm (Holtain Ltd, Dyfed, Wales, UK). BMI was calculated. Measurements were made by a trained operator at baseline, 12, 24 and 52 weeks. ADP Body density was assessed with an air displacement plethysmograph (BOD POD, Life Measurement Instruments, Concord, CA, USA). Before each testing session, a calibration procedure was performed and a brief description of the procedures was presented to the subjects. Subjects entered the chamber wearing a swimming suite and swim cap, and two body volume assessments were made. Whenever the difference between the two measurements was 4150 ml, a third body volume measurement was performed. Finally, and after attaching a nose clip, the subject was connected to the breathing circuit via a disposable filtered tube. Tidal breathing was determined and thoracic gas volume was calculated. Final body volume was computed based on the initial body volume corrected for thoracic gas volume and a surface area artefact computed automatically. This accounted for a negative volume owing to a more compressible air induced by the isothermal conditions associated with the skin surface area. 22 Siri s 23 two-compartment formula was used to calculate percentage body fat (FM%) from body density. From FM% and body weight, total FM (kg) and total fat-free mass (FFM) (kg) were calculated. TBW Body water was measured using an ID procedure. The subjects received an oral dose of deuterium oxide (0.7 g/kg body weight) and saliva samples were collected at 4, 5 and 6 h after the dose. The concentration of deuterium in each sample was measured using isotope ratio mass spectrometry as described elsewhere 24 and the pool size calculated. The measured pool size was reduced by 4% to account for the

3 exchange of deuterium with non-aqueous hydrogen. The hydration fraction of FFM was assumed to be and FM was calculated as the difference between FFM and body weight. DXA A whole-body DXA scan was performed using a Hologic QDR-1000W scanner (Hologic Inc., Waltham, MA, USA) and analysed using an enhanced version of the software to estimate bone mineral mass (subsequently used to derive ash ), bone mineral content (BMC), fat and FFM. The duration of total body scanning time was about 20 min. Subjects were measured while wearing only a standard light cotton shirt to minimize clothing absorption. The DXA device measures the attenuation of the two energy X-ray beams crossing the tissue. This allows partitioning between bone vs soft tissue and fat vs lean tissue in pixels of the body where there is no overlaying calcified tissue. 25 SFTs SFTs were measured at four sites (triceps, biceps, subscapula and suprailiac crest) on the non-dominant side of the body using Harpenden calipers. Predicted body density was calculated using the age- and sex-specific prediction equations of Durnin and Womersley, 9 based on the sum of the SFTs at each site. The proportion of body fat was calculated from body density using the equation of Siri C The three-compartment model (3-C) assumes that the body can be divided into fat, water and fat-free dry mass (protein plus mineral) and utilizes measurements of body weight, body volume (from ADP) and total body water (TBW) from deuterium dilution space). FM (kg) ¼ BV (l) TBW (l) WT (kg) C The four-compartment model (4-C) divides the body into fat, water, protein and mineral, thereby avoiding the assumption that the ratio between mineral and protein in FFM is constant. The body composition data collected in each measurement described earlier were combined to yield an estimation of FM from a 4-C model as FM (kg) ¼ [(2.747 BV) þ (0.710 TBW)] þ [(1.460 A) þ ( BW)] where A is BMC determined by DXA (in kg). Total-body mineral mass was calculated as BMC TBW is total body water litres, A is ash (kg) and BW is body weight (kg). The protein plus carbohydrate (P þ C) compartment was derived by difference (P þ C ¼ BW TBW TMM FM). The precision of the 3-C and 4-C model to assess body fat was and kg when a 1% precision for water estimation was used, respectively. 27 The precision for estimates of TBW was based on sequential measurements of the isotopic enrichment of water in saliva samples taken at 4, 5 and 6 h after oral administration of the isotope. Precision for the measurement of water calculated from this study was 0.45 l (about 1%). 27,28 3 T-BIA Conventional tetrapolar bioelectrical impedance analysis (T- BIA) was measured using the Bodystat-1500 system (Bodystat, Douglas, Isle of Man, UK) with electrodes placed at the standard sites on the hands and feet. The measurement was made after subjects had been supine for 20 min. Body fat was calculated according to the algorithm provided by the manufacturers. A more detailed description of this method is provided elsewhere. 10,26 LTL method Bioelectrical impedance from LTL was measured using the Tanita-305 body-fat analyzer (Tanita Corp., Tokyo, Japan). Subjects stood on the metal sole plates after a period of at least 10 min standing to minimize potential errors from acute shifts in fluid distribution. Body composition for all subjects was estimated using the standard prediction equations rather than those designated for athletes, regardless of the exercise habits of the participants. Measurements of body fat were calculated by prediction equations within the instrument. Statistical analysis The data were expressed as mean7s.d. Changes (D) in body weight and body composition were expressed in absolute (kg) and relative (%) values. Univariate analysis for repeated measures was used to analyse differences between phases. The measured body fat (kg) from each individual method was compared with the measurement of fat obtained from the 3-C and 4-C model using intraclass correlation (ICC) analysis and calculating the individual bias (method minus reference method) and the limits of agreements (Bland Altman method). Individual estimates of FM obtained by the LTL method were regressed (as dependent variables) against corresponding values obtained by the 3-C model during weight loss and weight regain. The analysis was carried out using Stata 8 for Windows (Stata Corporation, College Station, TX, USA) and Sigmaplot 8 for Windows (SPSS Inc., USA). Significance level was set at Po0.05. Results This study examined body composition changes during weight loss and weight regain in overweight and obese women. At the start subjects had a mean weight of

4 4 8.5 kg, height of m and BMI of kg/m 2. After the 12-week weight loss programme, subjects had lost kg (D wt ¼ %, Po0.001) of their weight, but at the end of 1 year subjects had regained kg (Po0.001), corresponding to a net weight loss of % (Table 1). All subjects lost weight during the 0 12 week period. However, 10 subjects continued to lose weight from week 12 to 52 and their data is excluded from the analysis during this phase to provide information specifically on weight regain. Weight and fat loss in these 10 subjects from 0 to 12 week were not significantly different (P40.05) from the group as a whole. During the initial 12 weeks body FM (4-C model) decreased by kg (Po0.001) but then increased in parallel with the increase in body weight. After 1 year, subjects had regained kg of the FM lost during the initial 12 weeks giving a net fat loss of kg after 1 year. The 4-C model showed that FM accounted for % (Po0.001) of loss in body mass and TBW accounted for % (Po0.001). During weight regain Table 1 Anthropometric and body composition characteristics of overweight women at baseline, after weight loss therapy (12 weeks) and after 1 year of follow-up body composition changes were fat ¼ %, Po0.001 and TBW ¼ %, Po Surprisingly, total mineral mass (TMM) appeared to increase by kg (3%; Po0.01) during the weight loss phase and decreased by (2%; Po0.01) with weight regain. The protein plus carbohydrate compartment showed comparable changes to TMM but in the opposite direction. The hydration of FFM was % at baseline and did not change after weight loss and weight regain (Table 2). Table 3 shows the change in FM by each method during the 0 12 week and week phases. The change in body fat measured by the different methods was compared to the 3-C and 4-C models. Table 4 shows that during weight loss the ICC of D FM (kg) between the 4-C model and ADP (ICC ¼ 0.86; Po0.001), TBW (ICC ¼ 0.90; Po0.001) and DXA (ICC ¼ 0.84; Po0.001) was greater than for SFT (ICC ¼ 0.39; Po0.001) and T-BIA (ICC ¼ 0.47; Po0.001), which are based on prediction equations. The equations based on the LTL system showed a similar correlation to the 4-C model (ICC ¼ 0.88; Po0.001). The correlation between the 3-C and 4-C model was unsurprisingly very high (ICC ¼ 0.99; Po0.001), reflecting the similarities in the model. The association between changes in FM measured by LTL method Baseline (n ¼ 58) 12 weeks (n ¼ 58) 52 weeks (n ¼ 48) Table 3 Changes in FM measured by different methods after weight loss and weight gain Age (years) Height (m) Weight (kg) b Body fat (kg) b Model 4c b Model 3c ADP b Deuterium dilution (TBW) b DXA b SFT T-BIA a LTL b ADP, air displacement plethysmography; DXA, dual energy X-ray absorptiometry; LTL, leg-to-leg method; Model 4c, 4-compartment model; Model 3c, 3-compartment model; SFT, skinfold thickness; TBW, total body water; T-BIA, tetrapolar bioimpedance. Mean7s.d. are shown. Univariate analysis for repeated measures was used. a Po0.05. b Po Weight loss (baseline 12 weeks) FM (kg) Weight gain a (12 52 weeks) Model 4C Model 3C ADP Deuterium dilution (TBW) DXA SFT T-BIA LTL ADP, air displacement plethysmography; DXA, dual energy X-ray absorptiometry; FM, fat mass; LTL, leg-to-leg method; Model 4c, 4-compartment model; Model 3c, 3-compartment model; SFT, skinfold thickness; TBW, total body water; T-BIA, tetrapolar bioimpedance. a Ten subjects were excluded from this analysis as they were losing weight during the follow-up phase. Number of subjects ¼ 48. Table 2 Four-compartment model: changes in body compartments and FFM hydration after weight loss and weight regain Baseline 12 weeks 52 weeks a Weight loss (baseline 12 weeks) Weight gain a (12 52 weeks) FM (kg) TBW (kg) TMM (kg) Protein+CHO (kg) Hydration FFM (%) a FM, free mass; FFM, fat-free mass; TBW, total body water; TMM, total mineral mass. a Ten subjects were excluded from this analysis as they were losing weight during the follow-up phase. Number of subjects ¼ 48.

5 Table 4 Intraclass correlation between compartment models (3-C, 4-C) and other methods of body composition during weight gain and weight loss 5 Weight loss (baseline 12 weeks) Weight gain a (12 52 weeks) ICC Model4c ICC Model3c ICC Model4c ICC Model3c Model 3c 0.99 ( ) F 0.98 ( ) F ADP 0.86 ( ) 0.87 ( ) 0.82 ( ) 0.88 ( ) Deuterium dilution (TBW) 0.90 ( ) 0.90 ( ) 0.87 ( ) 0.88 ( ) DXA 0.84 ( ) 0.84 ( ) 0.77 ( ) 0.80 ( ) SFT 0.39 ( ) 0.48 ( ) 0.59 ( ) 0.62 ( ) T-BIA 0.47 ( ) 0.47 ( ) 0.36 ( ) a 0.38 ( ) a LTL 0.88 ( ) 0.87 ( ) 0.81 ( ) 0.87 ( ) ADP, air displacement plethysmography; DXA, dual energy X-ray absorptiometry; ICC, intraclass coefficient of correlation; LTL, leg-to-leg method; Model 4c, 4-compartment model; Model 3c, 3-compartment model; SFT, skinfold thickness; TBW, total body water; T-BIA, tetrapolar bioimpedance. ICC ¼ a : Po0.01; all the other variables: Po % CI are shown in brackets. a Ten subjects were excluded from this analysis as they were losing weight during the follow-up phase. Number of subjects ¼ 48. Table 5 Bland Altman analysis to compare the validity of body composition methods to assess FM change against multi-compartment models (3-C, 4-C) during active weight loss and weight gain Weight loss (baseline 12 weeks) Weight gain (12 weeks 52 weeks) a Model 3c Model 3c Bias 72 s.d. Bias 72 s.d. ADP Dilution method (TBW) DXA SFT T-BIA LTL Figure 1 Linear regression of changes (D) in FM measured by a 3-C model and LTL during weight loss and weigh regain. & Weight gain: n ¼ 48, R 2 ¼ 0.76, Po0.001; J weight loss: n ¼ 58, R 2 ¼ 0.77, Po ADP, air displacement plethysmography; DXA, dual energy X-ray absorptiometry; FM, fat mass; LTL, leg-to-leg method; Model 4c, 4-compartment model; Model 3c, 3-compartment model; SFT, skinfold thickness; TBW, total body water; T-BIA, tetrapolar bioimpedance. a Ten subjects were excluded from this analysis as they were losing weight during the follow-up phase. Number of subjects ¼ 48. and the 3-C model was comparable in both weight loss (R 2 ¼ 0.77, Po0.001) and weight regain (R 2 ¼ 0.76, Po0.001) phases (Figure 1). The similar performance of the LTL method in the two phases was also confirmed by the lack of a significant relationship (R 2 ¼ 0.02, P40.05) between the FM bias (LTL minus 3-C model) during weight loss and weight gain (data not shown). The mean bias and limits of agreement for each method, relative to the 3-C model are shown in Table 5. Overall the average bias was small for all methods with the exception of changes in FM (kg) measured by SFT during the weight loss phase. However, the limits of agreement (72s.d.) reflect the greater variability of SFT and T-BIA in measuring D FM, relative to DXA, ADP or TBW. During both weight loss and regain the limits of agreement for the LTL system were closer to the reference methods than SFT or T-BIA. Data on the agreement between the 3-C model and the LTL method during weight loss ( kg) and regain ( ) are shown in Figure 2. Discussion The present study examined the effects of weight loss and regain on body composition changes using multi-compartment models and assessed the validity of simpler twocompartment methods to measure changes in body FM during periods of negative and positive energy imbalance. The multi-compartment models (3-C, 4-C) theoretically reflect the changes in body fat content more accurately than the more common two-compartment model. We observed that the relative change in FM and TBW did not significantly differ during weight loss and weight regain. This confirms our previous observations in a small group of women over three successive loss/regain cycles 17 and it is close to theoretically appropriate values. 29 However, it is notable that the calculation of the energy imbalance, based on the loss of fat and lean tissue, implies that energy intake was higher than prescribed. We conclude that the clinical experience of subjects prescribed VLCDs does not imply any inappropriate loss of lean tissue,

6 6 Figure 2 Bland Altman analysis: FM estimate by LTL method in comparison to a 3-C model during weight loss (n ¼ 58; (a)) and weight gain (n ¼ 48; (b)). Solid lines indicate mean differences; dashed lines indicate 795% confidence interval. but investigation in subjects with full compliance to the prescription is required in order to demonstrate the true effect of a VLCD per se on body composition. The 3-C and 4-C models were highly correlated during both weight loss (ICC ¼ 0.99; Po0.001) and regain (ICC ¼ 0.98; Po0.001) phases. The measured variation in mineral mass in the 4-C model was statistically significant, but clinically small and in the opposite direction to weight change. Similar changes have been observed in other weight loss studies and the results attributed to technical artefacts related to the particular DXA instruments used. 30 The technical limitations of the DXA may preclude the valid assessment of mineral content during short-term weight change as a consequence of changes in the hydration of FFM, fat distribution, thickness of subcutaneous fat layers and positioning of the patient. 25,30 There is stronger evidence of the short-term stability of mineral mass during weight loss and we have therefore used the 3-C model as reference method for comparison with other methods to avoid any confounding owing to imprecision in the measurement of the mineral mass. In our study, the accuracy of the DXA to estimate body fat changes was lower than ADP and TBW in both phases but its performance improved with weight regain. The LTL method was more precise in estimating body fat changes than DXA and the difference was accentuated by the increase in weight. A recent comparative study has explored the validity of different methods after a weight loss of 5.6 kg. BIA (tetrapolar, leg-to-leg) underestimated percentage body fat whereas ADP overestimated percentage fat compared with DXA. However, the DXA was chosen by the authors as the reference method and a comparison of the changes (D) in FM was not performed. 33 Evans et al 12 used instead a 4-C model as reference method and the errors for estimates of changes in FM % by DXA, BIA and SFT were similar even though DXA overestimated decreases in FM % during weight loss. Another study compared the 4-C model with the 3-C model, underwater weighing, DXA, BIA and SFT in 32 obese women after a weight loss of 13 kg. With the exception of the 3-C model, all other methods underestimated fat loss by at least 1.6 kg. However, the water fraction of the fat-free body component was increased after weight reduction and most methods may have underestimated fat loss because of unexpected changes in hydration of the fat-free body component. 11 The present study is larger that previous analyses and uniquely has also considered the relationship between methods during weight regain. The accuracy of the TBW method was comparable to ADP, probably because there was no change in hydration of FFM during both phases (data not shown). The LTL method performed better than the other prediction methods and had a comparable performance in measuring FM changes during weight loss and weight regain. This data is comparable with previous validation studies of the LTL system. 8,10 Utter et al 15 have studied the validity of the method during negative energy balance and they proved that the LTL could accurately detect body composition changes in obese women undergoing a moderate weight loss. The measured superiority of the LTL system relative to the supine tetra-polar BIA system may be a consequence of a more robust prediction equation or technical advantages of measurements biased towards the lower body. Although, theoretically, tetrapolar systems provide a whole-body analysis, in practice the arms make a disproportionately large contribution to total impedance owing to their small cross-sectional area and this may introduce the imprecision observed. Overall, the LTL system is a simple, rapid and highly reproducible system to measure body composition. Conclusions Short-term changes in body composition during acute weight loss and subsequent regain were similar and close

7 to theoretically appropriate values. The LTL system is a simple and practical tool to measure these changes in body composition and shows good agreement with a 3-C model. The accuracy is similar to those observed for ADP and TBW. Our findings suggest that the LTL system may be a useful tool for the clinical monitoring of body composition changes during weight management programmes. Acknowledgements We thank to Mr Anthony Wright for the stable isotopes analysis, Dr Gail R Goldberg and Mrs Lorraine Cooney for the assistance with volunteers and measurements and Dr Adrian Mander for statistical advice. The study was funded by Tanita UK Ltd. References 1 Bray GA. Medical consequences of obesity. J Clin Endocrinol Metab 2004; 89: Pi-Sunyer FX. The epidemiology of central fat distribution in relation to disease. Nutr Rev 2004; 62: S120 S Ellis KJ. Human body composition: in vivo methods. Physiol Rev 2000; 80: Jebb SA, Elia M. Techniques for the measurement of body composition: a practical guide. Int J Obes Relat Metab Disord 1993; 17: Pietrobelli A, Heymsfield SB. Establishing body composition in obesity. J Endocrinol Invest 2002; 25: Prentice AM, Jebb SA. Beyond body mass index. Obes Rev 2001; 2: Sung RY, Lau P, Yu CW, Lam PK, Nelson EA. Measurement of body fat using leg-to-leg bioimpedance. Arch Dis Child 2001; 85: Xie X, Kolthoff N, Barenholt O, Nielsen SP. Validation of a leg-toleg bioimpedance analysis system in assessing body composition in postmenopausal women. Int J Obes Relat Metab Disord 1999; 23: Durnin JVGA, Womersley J. Body fat assessed from total body density and its estimation from skinfolds thickness measurement in 481 men and women aged 16 to 72 years. Br J Nutr 1974; 32: Jebb SA, Cole TJ, Doman D, Murgatroyd PR, Prentice AM. Evaluation of the novel Tanita body-fat analyser to measure body composition by comparison with a four-compartment model. Br J Nutr 2000; 83: Fogelholm MG, Sievanen HT, van Marken Lichtenbelt D, Westerterp KR. Assessment of fat mass loss during weight reduction in obese women. Metabolism 1997; 46: Evans EM, Saunders MJ, Spano MA, Arngrimsson SA, Lewis RD, Cureton KJ. Body-composition changes with diet and exercise in obese women: a comparison of estimates from clinical methods and a 4-component model. Am J Clin Nutr 1999; 70: Hollander FM, De Roos NM, De Vries JH, Van Berkhout FT. Assessment of nutritional status in adult patients with cystic fibrosis: whole-body bioimpedance vs body mass index, skinfolds, and leg-to-leg bioimpedance. J Am Diet Assoc 2005; 105: Tsui E, Gao X, Zinman B. Bioelectrical impedance analysis (BIA) using bipolar foot electrodes in the assessment of body composition in Type 2 diabetes mellitus. Diabet Medic 1998; 15: Utter AC, Nieman DC, Ward AN, Butterworth DE. Use of the legto-leg bioelectrical impedance method in assessing body-composition change in obese women. Am J Clin Nutr 1999; 69: Nunez C, Gallagher D, Visser M, Pi-Sunyer FX, Wang Z, Heymsfield SB. Bioimpedance analysis: evaluation of leg-to-leg system based on pressure contact footpad electrodes. Med Sci Sports Exerc 1997; 29: Jebb SA, Goldberg GR, Coward WA, Murgatroyd PR, Prentice AM. Effects of weight cycling caused by intermittent dieting on metabolic rate and body composition in obese women. Int J Obes Relat Metab Disord 1991; 15: Prentice AM, Jebb SA, Goldberg GR, Coward WA, Murgatroyd PR, Poppitt SD et al. Effects of weight cycling on body composition. Am J Clin Nutr 1992; 56: 209S 216S. 19 van der Kooy K, Leenen R, Deuremberg P, Seidell JC, Westerterp KR, Hutavst GAJ. Changes in fat free mass in obese subjects after weight loss: a comparison of body composition measures. Int J Obes Relat Metab Disord 1992; 16: Albu J, Smolowitz J, Lichtman S, Heymsfield SB, Wang J, Pierson RN et al. Composition of weight loss in severely obese women: a new look at old methods. Metabolism 1992; 10: Krebs JD, Evans S, Cooney L, Mishra GD, Fruhbeck G, Finer N et al. Changes in risk factors for cardiovascular disease with body fat loss in obese women. Diabetes Obes Metab 2002; 4: Dempster P, Aitkens S. A new air displacement method for the determination of human body composition. Med Sci Sports Exerc 1995; 27: Siri WS. Body Composition from Fluid Spaces and Density: A Combined Analysis of Methods. National Academy of Sciences: Washington DC, Hoffman DJ, Sawaya AL, Coward WA, Wright A, Martins PA, de Nascimento C et al. Energy expenditure of stunted and nonstunted boys and girls living in the shantytowns of Sao Paulo, Brazil. Am J Clin Nutr 2000; 72: Pietrobelli A, Formica C, Wang Z, Heymsfield SB. Dual energy X ray absorptiometry body composition model: review of physical concepts. Am J Physiol 1996; 271: Fuller NJ, Sawyer MB, Elia M. Comparative evaluation of body composition methods and predictions, and calculation of density and hydration fraction of fat-free mass, in obese women. Int J Obes Relat Metab Disord 1994; 18: Fuller NJ, Jebb SA, Laskey MA, Coward WA, Elia M. Fourcomponent model for the assessment of body composition in humans: comparison with alternative methods, and evaluation of the density and hydration of fat-free mass. Clin Sci (London) 1992; 82: Murgatroyd PR, Coward WA. An improved method to estimate changes in whole body fat and protein mass in man. Br J Nutr 1989; 62: Garrow JS, Stalley S, Diethelm R, Pittet Ph, Hesp R, Halliday D. A new method for measuring the body density of obese adults. Br J Nutr 1979; 42: Tothill P. Dual-energy X-ray absorptiometry measurements of total-body bone mineral during weight change. J Clin Densitom 2005; 8: Gossain VV, Rao DS, Carella MJ, Divine G, Rovner DR. Bone mineral density (BMD) in obesity effect of weight loss. J Med 1999; 30: Vestergaard P, Borglum J, Heickendorff L, Mosekilde L, Richelsen B. Artifact in bone mineral measurements during a very low calorie diet: short-term effects of growth hormone. J Clin Densitom 2000; 3: Frisard MI, Greenway FL, Delany JP. Comparison of methods to assess body composition changes during a period of weight loss. Obes Res 2005; 13: