Techniques for Undertaking Dual-Energy X-Ray Absorptiometry Whole-Body Scans to Estimate Body Composition in Tall and/or Broad Subjects

Transcription

1 International Journal of Sport Nutrition and Exercise Metabolism, 2012, 22, Human Kinetics, Inc. ORIGINAL RESEARCH Techniques for Undertaking Dual-Energy X-Ray Absorptiometry Whole-Body Scans to Estimate Body Composition in Tall and/or Broad Subjects Alisa Nana, Gary J. Slater, Will G. Hopkins, and Louise M. Burke Dual-energy X-ray absorptiometry (DXA) is becoming a popular tool to measure body composition, owing to its ease of operation and comprehensive analysis. However, some people, especially athletes, are taller and/ or broader than the active scanning area of the DXA bed and must be scanned in sections. The aim of this study was to investigate the reliability of DXA measures of whole-body composition summed from 2 or 3 partial scans. Physically active young adults (15 women, 15 men) underwent 1 whole-body and 4 partial DXA scans in a single testing session under standardized conditions. The partial scanning areas were head, whole body from the bottom of the chin down, and right and left sides of the body. Body-composition estimates from whole body were compared with estimates from summed partial scans to simulate different techniques to accommodate tall and/or broad subjects relative to the whole-body scan. Magnitudes of differences in the estimates were assessed by standardization. In simulating tall subjects, summation of partial scans that included the head scan overestimated whole-body composition by ~3 kg of lean mass and ~1 kg of fat mass, with substantial technical error of measurement. In simulating broad subjects, summation of right and left body scans produced no substantial differences in body composition than those of the whole-body scan. Summing partial DXA scans provides accurate body-composition estimates for broad subjects, but other strategies are needed to accommodate tall subjects. Keywords: DEXA, lean mass, body fat, BIG Dual-energy X-ray absorptiometry (DXA) is becoming a popular tool to measure body composition in athletes, owing to its ease of operation and comprehensive analysis of body composition (Nana, Slater, Hopkins, & Burke, 2012). One of the limitations of DXA, however, is the size of the active scanning area. Since DXA machines have been developed primarily to measure bone-mineral content (BMC), their design reflects both the profile of targeted groups (e.g., elderly women) and the focus on specific bone regions. Typical dimensions of the active scanning area of most commercial DXA machines are cm wide by cm long. The actual scanning area, which must accommodate an empty space at the commencement of the scan to initiate readings, is smaller than this (Genton, Hans, Kyle, & Pichard, 2002). In practice, it is difficult to scan an individual >~190 cm tall or with a supine body width including the separation of their arms >~58 cm. This is clearly problematic for sports in which height is a favorable characteristic; rowing, volleyball, Nana and Burke are with AIS Sports Nutrition, Australian Institute of Sport, Canberra, ACT, Australia. Slater is with the School of Health and Sports Sciences, University of the Sunshine Coast, Maroochydore, QLD, Australia. Hopkins is with Sport and Recreation, Auckland University of Technology, Auckland, New Zealand. and basketball feature athletes > cm tall (Norton, Olds, Olive, & Craig, 1996). Similarly, the trunk breadth and large muscle mass favorable for other sports (Olds, 2001; e.g., bodybuilding, rowing, rugby) exceed the width of the active scanning area of the DXA machine. The inability to measure whole-body composition of tall and/or large individuals using current standardized DXA techniques represents a limitation to clinical practice, where the increasing access to DXA is otherwise allowing it to become the preferred choice for physique assessment. It may also bias the sampling processes in research where DXA is used to monitor physique traits, characterize a subject population, or monitor the effects of an intervention to alter physique. This may become an even greater problem if trends toward increasing body size in general and athletic populations continue. Several techniques have been suggested to overcome the limitations of the small scanning area associated with DXA. For example, tall subjects can be scanned with the exclusion of the head or feet or scanned with bent knees to allow both the head and feet to be included in the scan (Evans, Prior, & Modlesky, 2005; Silva et al., 2004). Another alternative is to sum two partial scans, with preference given to dividing the body at the neck rather than at the hip (Evans et al., 2005). Another option is mummy wrapping, a technique in which a large or broad subject is tightly wrapped with a bed sheet so that 313

2 314 Nana et al. the whole body falls within the active scanning area. To date, comparisons between these different scanning options have not been thoroughly investigated, particularly when measurement of individuals who are both tall and broad is considered. Therefore, the aim of this study was to investigate techniques to estimate whole-body composition in active people that could be applied to individuals with tall and/ or broad physiques and that is both reliable and practical. Since additional DXA scanning to accommodate large subjects requires more testing time, increased radiation exposure, and increased technician manipulation and analysis, the solution should include the fewest scans possible to achieve results with acceptable levels of reliability. We excluded the bent-knees and mummywrapping techniques since they are both impractical and prevent accurate measurements of regional distribution of body composition, information that is of interest in the evaluation of athletic populations. Subjects Methods We recruited 30 subjects (15 men and 15 women) from a local pool of physically active individuals who would represent the range of physiques found in athletic subpopulations who fit the scanning area. Specifically, subjects were excluded from the study if they were more than 190 cm tall (i.e., taller than the active scanning area of the DXA machine) and broader than the width of the scanning area when positioned according to our protocol. Subjects were required to be engaged in a structured training program of at least 4 hr/wk. All subjects signed a consent form approved by the human ethics committee of the Australian Institute of Sport and RMIT s human research ethics committee before participating in this study. Subject characteristics are described in Table 1. Study Overview Each subject underwent one whole-body (WHOLE) and four partial DXA scans in a single testing session (over ~30 min) in a random positioning order under standardized conditions of resting and fasting (Nana et al., 2012). Subjects were repositioned after each scan. Various combinations of the partial scans were summed to estimate total body composition; these were compared with the WHOLE for estimates of summed mass, bone mineral content, fat mass, and lean mass (Figure 1). Ideally, the technical error of measurement associated with summed partial scans would be derived from body-composition estimates of repeated measurements of summed partial scans. However, repeated partial scans could not be undertaken for this study due to the Table 1 Subject Body-Composition Characteristics, M ± SD All Men Women Age (yr) 31 ± 7 31 ± 8 30 ± 7 Height (cm) 174 ± ± ± 7 Scale body weight (kg) 70.7 ± ± ± 7.6 DXA estimates summed mass (kg) 71.1 ± ± ± 7.7 total lean (kg) 53.5 ± ± ± 5.4 total fat (kg) 14.5 ± ± ± 4.9 total BMC (kg) 3.1 ± ± ± 0.3 Note. DXA = dual-energy X-ray absorptiometry; BMC = bone mineral content. Figure 1 Positioning protocol for dual-energy X-ray absorptiometry.

3 Techniques to DXA Scan Tall and Broad Subjects 315 higher radiation exposure associated with the increased number of DXA scans. An alternative method to derive the technical error was to combine the error associated with whole-body scanning (Nana et al., 2012) with those of partial scans from the current study. See Statistical Analysis for further detail. Standardized Scanning Conditions Subjects were fasted and rested (no exercise) for at least 3 hr before the scans. They were also instructed not to drink any fluid during that period. All subjects wore light clothing consisting of plain T-shirts and shorts. All jewelry and metal objects were removed, and subjects voided their bladder before scanning. DXA Instrument Body composition was measured using a narrowed fanbeam DXA (Lunar Prodigy, GE Healthcare, Madison, WI) with analysis performed using GE Encore software (GE, Madison, WI). The DXA was calibrated with phantoms per manufacturer guidelines each day before measurement. All the scans were undertaken using the standard thickness mode, which was automatically chosen by the software. DXA Operational and Positioning Protocol All scans were performed and analyzed by one trained technician according to the AIS Whole Body DXA Scanning Protocol, which emphasizes consistency in the positioning of subjects on the scanning area of the DXA instrument (Nana et al., 2012). Specifically, for all positioning techniques, the subject s feet were placed in custom-made foam blocks to maintain a constant distance between the feet (15 cm) in each scan, and the hands were placed in custommade foam blocks so that they were in a midprone position with a standardized gap (3 cm) between the palms and trunk. These custom-made blocks were made of Styrofoam and were transparent under the DXA scan. The partial scanning areas were vertex (in the Frankfort plane) to menton (the inferior point of the mandible) for HEAD, whole body from menton down for BODY, right side of the body for RIGHT, and left side of the body for LEFT (see Figure 1). Specifically, for WHOLE, HEAD, and BODY, the technician positioned the subjects to ensure that they were centrally aligned in the scanning area. For the HEAD position, the scan was initiated per usual protocol but terminated at the superior aspect of the shoulders so that only the entire head region was measured. For BODY, the technician initiated the scanner to undertake one sweep and create an empty space before allowing the subject to reposition to the top of the scanning bed and be rescanned from the mouth down. For LEFT, the subjects were shifted to the right side of the scanning bed to intentionally allow the right side of the body (e.g., right arm and leg) to fall outside of the active scanning area, with the technician ensuring that the midline of the body (e.g., the midspine) was still within the scanning area. The opposite positioning protocol was undertaken for the RIGHT to capture the right side of the body. DXA-Analysis Protocol Body-composition estimates derived from WHOLE were compared with the sum of HEAD and BODY SCAN (simulating tall subjects), the sum of RIGHT and LEFT (simulating broad subjects), and the sum of HEAD, right body from RIGHT, and left body from LEFT (simulating tall and broad subjects). Figure 2 summarizes the various combinations of these estimates. WHOLE, BODY, LEFT, and RIGHT were analyzed automatically by the software, but regions of interest were subsequently confirmed by the technician. For BODY, special effort was made by the technician to ensure that the region of interest started at the menton down. HEAD was manually analyzed by the technician using the customized region-of-interest function of the Encore software and included the vertex to the menton. The sums of the partial scans were calculated manually by the technician (see Figure 2). Figure 2 Calculations of sums of partial scans to estimate whole-body composition.

4 316 Nana et al. Statistical Analysis A reliability spreadsheet from the Sportscience Web site ( was used to derive statistics for comparing body-composition estimates of whole-body DXA scans with those obtained by addition of partial scans (Hopkins, 2000). The relevant statistics were the difference in the means and the technical error of measurement, with their confidence limits. In this study, the typical error of measurement was equivalent to the technical error of measurement. This is the variation caused by the DXA machine and/or repositioning of the subject on the scanning bed. The magnitude of the difference in the means (partial minus whole body) was interpreted after the difference was standardized by dividing it by a suitable value of standard deviation (also known as Cohen s effect size or Cohen s d statistic), as previously described (Nana et al., 2012). Ideally, the estimates of smallest worthwhile effects, or changes or differences, should be based on a functional or practical outcome of interest, such as performance, strength, metabolic function, or adherence to a weight category (Nana et al., 2012). When there are no published data on the smallest worthwhile differences in body composition, standardization is the default approach. In the current study, one third of the between-subjects SD was used for standardizing [Δmean/(1/3 SD)], because the between-subjects SDs of body composition in our study population were approximately 3 times greater than those previously found in a study with athletic populations (Stewart & Hannan, 2000). Magnitudes of standardized effects were assessed as follows: <0.2 trivial, <0.60 small, <1.20 moderate, and <2.0 large (Hopkins, Marshall, Batterham, & Hanin, 2009). The changes in the mean and technical errors were classified as substantial (i.e., reached the smallest worthwhile effects) when the standardized value reached the threshold for small ( 0.2). The technical error of measurement from the comparison of partial- and whole-body scans (error partial-whole ) was combined with the technical error for whole-body scanning (error whole-whole ) from the study of Nana et al. (2012) to provide an estimate of the error associated with repeated measurement of summed partial scans (error partial-partial ), using the following equation: 2 error = 2 error error partial-partial partial-whole 2 whole-whole This equation was derived from first principles on the assumption that partial scanning produced additional random error that would combine independently with the error from whole-body scanning to produce the error for repeated partial scanning. Confidence limits for error partial-partial were derived via the standard errors for the variances of the errors, assuming the sampling variances were normally distributed. To interpret the magnitude of the technical error associated with summed partial scans (partial-partial), we doubled the technical error before assessing it on the previously described scale (Hopkins et al., 2009; Smith & Hopkins, 2011). The effect of a linear covariate should be considered the effect of twice its standard deviation. For further clarification, see Hopkins, et al. and Smith and Hopkins. Results Differences in Estimates of Whole-Body Composition Between Summed Partial Scans and a Whole-Body Scan Results of whole-body-composition estimates for summed mass, fat mass, lean mass, and BMC are summarized in Table 2, with comparisons between the estimate from the WHOLE and various addition of partial scans (as described in Figure 2). Differences in body-composition estimates from summed partial scans of summed mass and BMC were not substantially different from those of WHOLE they were less than the smallest worthwhile effect. In the case of estimates of fat mass and lean mass, estimates derived from summing LEFT and RIGHT (to simulate broad subjects) were also not substantially different from the results of WHOLE. However, body-composition estimates of fat mass involving the use of HEAD (i.e., to simulate tall subjects and tall and broad subjects) were substantially different from the results of WHOLE. Lean-mass estimates from summed partial scans to simulate tall and broad subjects could also be substantial when taking into account the confidence limits. The addition of partial scans with HEAD (HEAD + BODY simulating tall subjects, or HEAD + right body from RIGHT + and left body from LEFT simulating tall and broad subjects) overestimated fat mass by approximately 6 7% (~900 1,000 g). Similarly, there was an ~6% (~3,000 g) overestimation in whole-body estimates of lean mass between WHOLE and the sum of HEAD and BODY (simulating tall subjects). The technical errors associated with addition of partial DXA scans (partial-partial) ranged from 0.3% to 3.5%, which were mostly substantial (i.e., greater than the smallest worthwhile effects) when doubled to interpret the magnitude. An exception was for summed mass for HEAD + BODY (simulating tall subjects). However, the lean mass for HEAD + BODY (simulating tall subjects), when taking into consideration the confidence limits or the uncertainty, could be substantial. Differences in Estimates of Head-Only Composition From a Whole-Body Scan and a Head Scan The difference in body-composition estimates of WHOLE compared with addition of partial scans involving the HEAD is thought to be due to an inherited technical error of DXA associated with scanning just the HEAD. To further highlight this technical error, bodycomposition estimates of the head region from WHOLE were compared with estimates measured by HEAD (see Figure 3[A]), and the results are presented in Table 3. All

5 Techniques to DXA Scan Tall and Broad Subjects 317 Table 2 Differences in Estimates of Whole-Body Composition Between Summed Partial Scans (to Simulate Tall, Broad, and Tall and Broad Subjects) and a Whole-Body Scan Smallest Worthwhile Effect Partial-Whole (%) TEM (%) (%) (g) Δmean ±CL TEM a whole b Whole- Summed mass Partialpartial HEAD + BODY LEFT + RIGHT * 1.25 HEAD + left body of LEFT + right body of RIGHT * 1.26 Fat mass HEAD + BODY 6.8* * 1.32 LEFT + RIGHT * 1.47 HEAD + left body of LEFT + right body of RIGHT 6.1* * 1.31 Lean mass HEAD + BODY 5.6* LEFT + RIGHT * 1.38 HEAD + left body of LEFT + right body of RIGHT * 1.38 Bone mineral content HEAD + BODY * 1.67 LEFT + RIGHT * 1.32 HEAD + left body of LEFT + right body of RIGHT * 1.50 Note. TEM = Technical error of measurement; Δmean = mean of summed partial scans minus mean of whole-body scan; CL = 90% confidence limits in ± or / form. a 90% CLs all / b TEM of whole-body composition measurements from Nana et al., (2012); 90% CLs range from / 1.24 to / *Greater than the smallest worthwhile effect. / CL the body-composition estimates (total fat, lean, BMC, and summed mass) measured by HEAD were substantially different from the head as measured from WHOLE. Specifically, HEAD substantially overestimated fat mass by ~150% (825 g) while substantially overestimating BMC by 0.9% (5 g). On the other hand, lean mass was substantially underestimated by ~25% (880 g), and summed mass underestimated by ~2% (90 g). The technical errors of measurement associated with DXA measurements of the head from HEAD were substantial for all body-composition estimates. Differences in Estimates of Body-Only Composition Between Summed Partial Scans and a Reference Scan (a Whole- Body Scan for Tall or Broad Subjects or a Body-Only Scan for Tall and Broad Subjects) Due to the large and substantial differences in DXA estimates of head from WHOLE and head from HEAD (Figure 3[A]), further analysis was undertaken in which the body from WHOLE was compared with the body from BODY (Figure 3[B]), the body from WHOLE was compared with the sum of right body from RIGHT and left body from LEFT (Figure 3[C]), and the body from BODY was compared with the sum of right body from RIGHT and left body from LEFT (Figure 3[D]). This was to further reconfirm that the technical error of the DXA technology is only limited to the HEAD and not the BODY. Comparisons of body-composition estimates of different body regions are presented in Table 4. In summary, when the head region is excluded, the differences in body-composition estimates for most measurements (mass, fat, lean, and BMC) were not substantial. Exceptions were for summed mass and BMC, where differences between body-only estimates from WHOLE or BODY and summed partial body-only scans could be substantial. Nevertheless, it can be generally concluded that there is some technical error with DXA when scanning just the HEAD. The technical error associated with DXA measurements between summed partial scans and a reference scan (partial-reference) ranged from 0.3% to 2.0%, which were mostly substantial (i.e., greater than the smallest worthwhile effects) when doubled to interpret the magnitude. An exception was for summed mass for body from WHOLE versus body from BODY and body from WHOLE versus right body from RIGHT + left body from LEFT.

6 318 Nana et al. Figure 3 Comparison between scans for estimating (A) head derived from head scan or whole scan, (B) body derived from whole scan or body scan, (C) body derived from whole scan or the addition of right body from right scan and left body from left scan, and (D) body from body scan or the addition of right body from right scan and left body from left scan. Table 3 Differences in Estimates of Head-Only Composition From a Whole-Body Scan and a Head Scan Smallest Worthwhile Effect Head From HEAD Head From WHOLE (%) % g Δmean ±CL TEM Fat * * Lean * * Bone mineral content * * Summed mass * * Note. Δmean = mean of head from HEAD minus mean of head from WHOLE; ±CL = 90% confidence limits; TEM = technical error of measurement. a 90% CLs all / *Greater than the smallest worthwhile effect Discussion DXA was originally designed to measure specific bone regions (bone mineral density of the hip and spine) of the elderly. Therefore, the dimensions of a typical DXA scanning area have become a common limitation to athletic populations for whom height and wide trunk breadth are the favorable physiques in many sports. To our knowledge, this is the first study to systematically examine different ways to undertake DXA scans on tall and/or broad individuals. The main findings were that strategies can be undertaken with DXA technology to accommodate subjects who are wider than the scanning bed. However, estimates of whole-body composition in tall or tall and broad subjects include error. Specifically, we found substantial errors that were greater than the statistically derived smallest worthwhile effects in wholebody-composition estimates when the addition of partial

7 Techniques to DXA Scan Tall and Broad Subjects 319 Table 4 Differences in Estimates of Body-Only Composition Between Summed Partial Scans (to Simulate Tall, Broad, and Tall and Broad Subjects) and a Reference Scan (a Whole-Body Scan for Tall or Broad Subjects and a Body-Only Scan for Tall and Broad Subjects) Smallest Worthwhile Effect Partial-Reference (%) (%) (g) Δmean ±CL TEM a Summed mass body from WHOLE vs body from BODY body from WHOLE vs right body from RIGHT + left body from LEFT body from BODY vs right body from RIGHT + left body from LEFT * Fat body from WHOLE vs body from BODY * body from WHOLE vs right body from RIGHT + left body from LEFT * body from BODY vs right body from RIGHT + left body from LEFT * Lean body from WHOLE vs body from BODY * body from WHOLE vs right body from RIGHT + left body from LEFT * body from BODY vs right body from RIGHT + left body from LEFT * Bone mineral content body from WHOLE vs body from BODY * body from WHOLE vs right body from RIGHT + left body from LEFT * body from BODY vs right body from RIGHT + left body from LEFT * Note. Δmean = mean of summed partial minus mean of reference scan; ±CL = 90% confidence limits; TEM = technical error of measurement a 90% confidence limits all / *Greater than the smallest worthwhile effect. scans included HEAD, with no additional error observed when simulating tall and broad subjects. Only two studies have previously investigated alternative scanning techniques to accommodate tall subjects (Evans et al., 2005; Silva et al., 2004). Silva et al. compared standard DXA scans with scans undertaken while subjects were positioned with bent knees, thus allowing tall individuals to arrange their body within the active scanning area of the DXA. With the knees bent, BMC was overestimated by 30 and 100 g, fat mass was overestimated by 1.54 and 2.20 kg, and lean mass was underestimated by 1.55 and 2.30 kg for males and females, respectively (Silva et al., 2004). However, this method could not be undertaken in our study because we found the elevation of bent knees to be greater than the scanner arm (i.e., inadequate height clearance). Therefore, we chose alternative methods to simulate scans on tall and/or broad individuals based on work from Evans et al. (2005), who modeled measurements of body composition in tall subjects by adding the estimates of two partial DXA scans and found no significant differences from whole-body-composition scans. Specifically, the study recommended dividing the body at the superior aspect of the shoulders as a preferred method over dividing the body at the proximal femurs bisecting the femoral necks (Evans et al., 2005). The width of the DXA active scanning area is also limited and will therefore be problematic for many big and broad athletes or obese subjects. Only two studies have previously investigated this problem, whereby summing half-body scans was found to accurately estimate whole-body composition in obese people (Rothney, Brychta, Schaefer, Chen, & Skarulis, 2009; Tataranni & Ravussin, 1995). However, those studies did not examine alternative scanning techniques for subjects who are both tall and broad, which is common in athletic populations. In the current study, we found that summing scans that involve the head region caused errors in estimation of whole-body composition on the order of ~3 kg lean mass and ~1 kg fat mass. These errors were considered substantial when compared with a statistically derived estimate of the smallest worthwhile effect. However, the errors may not be clinically important in the general population (e.g., 1 kg of fat mass is equal to 1.25% for an 80-kg man) or in some situations in which DXA might be used to measure body composition in athletes for example, where a general estimate of fat-free mass is needed to calculate energy availability (Loucks, 2004) or where gross profiling of physique is undertaken to differentiate characteristics between heterogeneous athletic populations. At this time, the smallest worthwhile effect for differences in physique has not been determined from

8 320 Nana et al. a functional or practical viewpoint. However, it is likely to be smaller when dealing with changes in physique in a population or individual over time or when it is due to an intervention, monitoring regional physiques, characterizing physique differences in a more homogeneous population, or determining suitability of a weight category for an athlete in a weight-classed sport. The size of the smallest worthwhile effect of differences in physique on performance outcomes or clinical issues (e.g., prevalence of injuries or illness) is currently unknown, and therefore a statistical approach via standardization is the appropriate default approach. The error associated with DXA measurement of HEAD as a region of interest (i.e., scanning the head only), which has not been reported in previous studies, may be due to technical and software limitations associated with the algorithm for determining BMC and soft tissue of the skull; specifically, it may bypass the assumptions of the whole-body fat-distribution model (Taylor, Konrad, Norman, & Harcke, 1997). In nonbone-containing pixels (arms, legs, etc.), soft tissue is simply determined by calculating the ratio of attenuation of the two photon energies, while in bone-containing pixels, the DXA assumes that soft tissue is the same as the neighboring non-bone-containing pixel. Therefore, for an isolated HEAD where the majority of the site is BMC, soft-tissue estimation would be poor because the HEAD contains negligible non-bone-containing pixels. It is suggested that estimation of soft tissues in other high-bone-containing areas such as thorax or arms may also be inaccurate (Andreoli, Scalzo, Masala, Tarantino, & Guglielmi, 2009; Lands, Hornby, Hohenkerk, & Glorieux, 1996; Roubenoff, Kehayias, Dawson-Hughes, & Heymsfield, 1993). The technical error associated with undertaking a DXA scan of just the head region in our study could potentially be attributed to the Lunar Prodigy technology and software. Differences between the results of the current study and those of Evans et al. (2005) and Silva et al. (2004), who used Hologic DXA machines, might be specific to the machine involved. Such differences include different beam technology (pencil vs. narrow fan-beam), as well as the ability to undertake an isolated DXA scan of the head region (e.g., region of interest function or equivalent). Whether this error is only limited to the Lunar Prodigy DXA scanner and not to the other scanners is unclear. A potential alternative technique for Lunar Prodigy DXA might be to dissect the body at the femoral neck, as undertaken by Evans et al.; however, this technique will need to be thoroughly investigated before implementation. Difficulties in scanning tall and/or broad subjects have several implications, especially in research. It appears that most studies involving DXA scanning of whole-body composition of athletes automatically exclude tall and/or broad subjects and thus create sampling bias. This could be important if there are fundamental differences in the physiology of individuals with larger physiques; for example, there is evidence that taller individuals have proportionally lower resting and total energy expenditure relative to body mass than their shorter counterparts (Heymsfield, Childers, Beetsch, Allison, & Pietrobelli, 2007; Heymsfield & Pietrobelli, 2010). Therefore, it is important that a satisfactory technique be developed to scan people of all physique types so that they are adequately represented in all aspects of clinical servicing and research. Even if the current protocol of summing partial scans could be used to develop a method to scan tall people with acceptable reliability, some issues could remain. First, we have assumed a constant measure of reliability across all physiques, or at least the physiques represented in our cohort. This may not be true there may be bias associated with height or width, which would require a study of larger sample size, including individuals with extreme physiques, to investigate this. Whether issues related to the validity of DXA estimates of body composition apply equally to different physiques is also of interest. It has been shown that human bodies are not geometrically similar to each other, invalidating the concept of allometric scaling according to body mass (Nevill, Stewart, Olds, & Holder, 2004). Therefore, it is likely that assumptions made about body composition do not apply equally to those with extreme physiques. There is also the potential of increased technical error associated with DXA scanning of individuals with thick bodies. This is due to the beam hardening effect the preferential loss of lower energy photons relative to high-energy photons as a result of increasing body thickness (Kohrt, 1995; Prior et al., 1997). This means that body-composition results obtained from DXA may be systematically different between a thin and an obese person (Roubenoff et al., 1993) and potentially between a lean endurance athlete and a rugby player. Prior et al. found DXA to overestimate percent body fat as body thickness increased; however, opposite results were observed in other studies (Jebb, Goldberg, Jennings, & Elia, 1995; Tothill, Avenell, Love, & Reid, 1994). Manufacturers have claimed to improve their software to address the beam-hardening effect in an effort to overcome this issue (Andreoli et al., 2009; Kohrt, 1995; Laskey, 1996), but accuracy of body-composition measurement in various body thicknesses as a result of these upgrades is still questionable until further studies are conducted. In addition, there are certain practical issues that must be accepted in using a partial-scan technique to work with these athletes. The increased number of DXA scans per assessment will lead to increased overall radiation exposure per year (particularly in longitudinal monitoring over time) and could be significant for athletes who are already exposed to ionizing radiation from other diagnostic imaging techniques (Cross, Smart, & Thomson, 2003; Orchard, Read, & Anderson, 2005). The extent of increased radiation exposure from addition of partial DXA scans depends on the scanning technique (degree of overlapping) and the DXA beam configuration (i.e., pencil or fan beam). For example, it is speculated that there is a 25 50% dose increase per subject per partial scan from a narrow-fan-beam DXA machine (e.g., Lunar

9 Techniques to DXA Scan Tall and Broad Subjects 321 Prodigy; D. Leslie, personal communication, November 11, 2011). Furthermore, increasing the number of DXA scans required for a single assessment would lead to increased scanning time per subject and therefore a reduction in the number of subjects who can be scanned within a certain time frame. This number was already limited by the requirement to have subjects fasted and rested to meet the criteria for the standardized protocol (Nana et al., 2012). Even if we can accommodate tall subjects by summing partial scans (HEAD + BODY), this technique will not work for very tall subjects (e.g., >2.20 m), who are likely to have a body length (neck to feet) that is greater than the active scanning area of the DXA machine. Such very tall subjects might need to be scanned three times per assessment to capture whole-body composition, adding further practical complications and potential for error. Finally, one of the limitations of using partial scanning techniques is the loss of the ability to monitor regional body composition (e.g., differences between left and right sides). The substantial errors associated with HEAD would suggest that it be excluded when scanning a tall subject (i.e., just scanning from the neck down). For an adult athlete, DXA scanning without the head is still practical to use in longitudinal tracking of body composition, as the body-composition estimates of the head region are unlikely to change substantially in a mature adult. On the other hand, if the head region is excluded, a precise calculation of fat-free mass (i.e., to estimate energy availability or determine a minimum weight category) is not possible. The remaining possibility would be to offset this estimate of whole-body composition by adding the physique characteristics of an average head. Although this can be done, it requires a manual calculation that is separate from the information provided by the DXA machine; this further adds to the time cost of scanning protocols, as well as introducing an opportunity for manual error. In this investigation, we tackled the practical problem of estimating whole-body composition in subjects who are taller and/or broader than the scanning area of a Lunar Prodigy DXA machine. We found that the estimates of whole-body composition achieved by summing partial scans to simulate broad individuals were acceptably reliable. However, summing partial scans involving a scan of just the head region (simulating tall individuals and tall and broad individuals) caused errors in estimation of whole-body composition on the order of ~3 kg of lean mass and ~1 kg fat mass. These errors are within the range of body-composition changes observed during a season among athletic populations (Argus, Gill, Keogh, Hopkins, & Beaven, 2010; Harley, Hind, & O Hara J, 2010). In addition, it is likely that the results from our study represent the minimum technical error associated with summing partial DXA scans to estimate whole-body composition. Further work is needed to develop other standardized protocols for using DXA to estimate body composition, particularly with tall and very tall athletes. In the meantime, elucidation of the smallest worthwhile effect in measuring differences in physique traits from a functional viewpoint is needed to determine the importance of the errors involved in scanning large/tall people. Only with this information can we decide how to interpret DXA estimates of body composition in larger athletes. Acknowledgment This investigation was supported by funding from the Physique and Fuel Centre program of the Australian Institute of Sport (AIS) and RMIT University, and the AIS Sports Nutrition Discipline. References Andreoli, A., Scalzo, G., Masala, S., Tarantino, U., & Guglielmi, G. (2009). Body composition assessment by dual-energy X-ray absorptiometry (DXA). La Radiologia Medica, 114(2), PubMed doi: /s Argus, C.K., Gill, N., Keogh, J., Hopkins, W.G., & Beaven, C.M. (2010). Effects of a short-term pre-season training programme on the body composition and anaerobic performance of professional rugby union players. Journal of Sports Sciences, 28(6), PubMed doi: / Cross, T.M., Smart, R.C., & Thomson, J.E. (2003). Exposure to diagnostic ionizing radiation in sports medicine: Assessing and monitoring the risk. Clinical Journal of Sport Medicine, 13(3), PubMed doi: / Evans, E.M., Prior, B.M., & Modlesky, C.M. (2005). A mathematical method to estimate body composition in tall individuals using DXA. Medicine and Science in Sports and Exercise, 37(7), PubMed doi: /01. mss Genton, L., Hans, D., Kyle, U.G., & Pichard, C. (2002). Dual-energy X-ray absorptiometry and body composition: Differences between devices and comparison with reference methods. Nutrition (Burbank, Los Angeles County, Calif.), 18(1), PubMed doi: / S (01) Harley, J.A., Hind, K., & O Hara J.P. (2011). Three-compartment body composition changes in elite rugby league players during a super league season, measured by dualenergy X-ray absorptiometry. Journal of Strength and Conditioning Research, 25, Heymsfield, S.B., Childers, D., Beetsch, J., Allison, D.B., & Pietrobelli, A. (2007). Body size and human energy requirements: Reduced mass-specific resting energy expenditure in tall adults. Journal of Applied Physiology, 103(5), PubMed doi: /japplphysiol Heymsfield, S.B., & Pietrobelli, A. (2010). Body size and human energy requirements: Reduced mass-specific total energy expenditure in tall adults. American Journal of Human Biology, 22(3), PubMed

10 322 Nana et al. Hopkins, W.G. (2000). Measures of reliability in sports medicine and science. Sports Medicine (Auckland, N.Z.), 30(1), PubMed doi: / Hopkins, W.G., Marshall, S.W., Batterham, A.M., & Hanin, J. (2009). Progressive statistics for studies in sports medicine and exercise science. Medicine and Science in Sports and Exercise, 41(1), PubMed doi: / MSS.0b013e31818cb278 Jebb, S.A., Goldberg, G.R., Jennings, G., & Elia, M. (1995). Dual-energy X-ray absorptiometry measurements of body composition: Effects of depth and tissue thickness, including comparisons with direct analysis. Clinical Science, 88(3), PubMed Kohrt, W.M. (1995). Body composition by DXA: Tried and true? Medicine and Science in Sports and Exercise, 27(10), PubMed Lands, L.C., Hornby, L., Hohenkerk, J.M., & Glorieux, F.H. (1996). Accuracy of measurements of small changes in soft-tissue mass by dual-energy x-ray absorptiometry. Clinical and Investigative Medicine. Medecine Clinique et Experimentale, 19(4), PubMed Laskey, M.A. (1996). Dual-energy X-ray absorptiometry and body composition. Nutrition (Burbank, Los Angeles County, Calif.), 12(1), PubMed doi: / (95) Loucks, A.B. (2004). Energy balance and body composition in sports and exercise. Journal of Sports Sciences, 22(1), PubMed doi: / Nana, A., Slater, G.J., Hopkins, W.G., & Burke, L.M. (2012). Effects of daily activities on DXA measurements of body composition in active people. Medicine and Science in Sports and Exercise, 44(1), PubMed doi: /mss.0b013e318228b60e Nevill, A.M., Stewart, A.D., Olds, T., & Holder, R. (2004). Are adult physiques geometrically similar? The dangers of allometric scaling using body mass power laws. American Journal of Physical Anthropology, 124(2), PubMed doi: /ajpa Norton, K., Olds, T., Olive, S., & Craig, N. (1996). Anthropometry and sports performance. In K. Norton & T. Olds (Eds.), Anthropometrica (pp ). Sydney, Australia: University of New South Wales Press. Olds, T. (2001). The evolution of physique in male rugby union players in the twentieth century. Journal of Sports Sciences, 19(4), PubMed doi: / Orchard, J.W., Read, J.W., & Anderson, I.J. (2005). The use of diagnostic imaging in sports medicine. The Medical Journal of Australia, 183(9), PubMed Prior, B.M., Cureton, K.J., Modlesky, C.M., Evans, E.M., Sloniger, M.A., Saunders, M., & Lewis, R.O. (1997). In vivo validation of whole body composition estimates from dual-energy X-ray absorptiometry. Journal of Applied Physiology, 83(2), PubMed Rothney, M.P., Brychta, R.J., Schaefer, E.V., Chen, K.Y., & Skarulis, M.C. (2009). Body composition measured by dual-energy X-ray absorptiometry half-body scans in obese adults. Obesity (Silver Spring, Md.), 17(6), PubMed Roubenoff, R., Kehayias, J.J., Dawson-Hughes, B., & Heymsfield, S.B. (1993). Use of dual-energy x-ray absorptiometry in body-composition studies: Not yet a gold standard. The American Journal of Clinical Nutrition, 58(5), PubMed Silva, A.M., Baptista, F., Minderico, C.S., Rodrigues, A.R., Pietrobelli, A., Teixeira, P.J.,... Sardinha, L.B. (2004). Calibration models to measure body composition in taller subjects using DXA. International Journal of Body Composition Research, 2(4), Smith, T.B., & Hopkins, W.G. (2011). Variability and predictability of finals times of elite rowers. Medicine and Science in Sports and Exercise, 43(11), PubMed doi: /mss.0b013e31821d3f8e Stewart, A.D., & Hannan, J. (2000). Sub-regional tissue morphometry in male athletes and controls using dual X-ray absorptiometry (DXA). International Journal of Sport Nutrition and Exercise Metabolism, 10(2), PubMed Tataranni, P.A., & Ravussin, E. (1995). Use of dual-energy X-ray absorptiometry in obese individuals. The American Journal of Clinical Nutrition, 62(4), PubMed Taylor, A., Konrad, P.T., Norman, M.E., & Harcke, H.T. (1997). Total body bone mineral density in young children: Influence of head bone mineral density. Journal of Bone and Mineral Research, 12(4), PubMed doi: / jbmr Tothill, P., Avenell, A., Love, J., & Reid, D.M. (1994). Comparisons between Hologic, Lunar and Norland dual-energy X-ray absorptiometers and other techniques used for whole-body soft tissue measurements. European Journal of Clinical Nutrition, 48(11), PubMed