Assessment of body composition in spinal cord injury: A scoping review

The objective of this scoping review was to map the evidence on measurement properties of body composition tools to assess whole-body and regional fat and fat-free mass in adults with SCI, and to identify research gaps in order to set future research priorities. Electronic databases of PubMed, EMBASE and the Cochrane library were searched up to April 2020. Included studies employed assessments related to whole-body or regional fat and/or fat-free mass and provided data to quantify measurement properties that involved adults with SCI. All searches and data extractions were conducted by two independent reviewers. The scoping review was designed and conducted together with an expert panel (n = 8) that represented research, clinical, nutritional and lived SCI experience. The panel collaboratively determined the scope and design of the review and interpreted its findings. Additionally, the expert panel reached out to their professional networks to gain further stakeholder feedback via interactive practitioner surveys and workshops with people with SCI. The research gaps identified by the review, together with discussions among the expert panel including consideration of the survey and workshop feedback, informed the formulation of future research priorities. A total of 42 eligible articles were identified (1,011 males and 143 females). The only tool supported by studies showing both acceptable test-retest reliability and convergent validity was whole-body dual-energy x-ray absorptiometry (DXA). The survey/workshop participants considered the measurement burden of DXA acceptable as long as it was reliable, valid and would do no harm (e.g. radiation, skin damage). Practitioners considered cost and accessibility of DXA major barriers in applied settings. The survey/workshop participants expressed a preference towards simple tools if they could be confident in their reliability and validity. This review suggests that future research should prioritize reliability and validity studies on: (1) DXA as a surrogate ‘gold standard’ tool to assess whole-body composition, regional fat and fat-free mass; and (2) skinfold thickness and waist circumference as practical low-cost tools to assess regional fat mass in persons with SCI, and (3) females to explore potential sex differences of body composition assessment tools. Registration review protocol: CRD42018090187 (PROSPERO).

Introduction Individuals who suffer from neurologic trauma, such as spinal cord injury (SCI), undergo significant changes in body composition that increase the risk for secondary health conditions, including increased fat mass, decreased lean mass, and reduced bone density [1][2][3][4][5]. For example, the atrophy of metabolically active tissues and reduced activity levels result in increased risk of pressure ulcers, decreased energy expenditure, and a high risk of excess fat mass deposition under the skin as well as in viscera, liver and muscle [2,[6][7][8][9]. Reliable and valid body composition assessment tools are required to monitor profiles of fat and fat-free mass, and assess the effects of interventions (e.g., nutrition, exercise) to help improve these profiles [10,11]. Understanding the reliability and validity of body composition assessment tools when applied for individuals with SCI will enable clinicians, researchers or other practitioners to make an informed decision regarding its use.
Body composition assessment tools range from simple to complex with all having limitations and some degree of measurement error [10,[12][13][14]. Examples include body mass index (BMI), waist circumference, skinfold thicknesses, bioelectrical impedance analysis (BIA), air displacement plethysmography (ADP), hydrostatic weighing, dual-energy X-ray absorptiometry (DXA), computed tomography (CT) and magnetic resonance imaging (MRI). Reliability and validity of these methods have often only be assessed in able-bodied populations, notwithstanding that many of these tools incorporate various assumptions that may or may not be met in people with a disability such as those with SCI, resulting in many interpretation difficulties [14].
As an example, a recent systematic review could only establish "low to moderate" confidence in the evidence showing that aerobic exercise can improve body composition in adults with chronic SCI [15], due to imprecision in the evidence. This imprecision may in part be explained by different body composition assessment tools employed, which may or may not be reliable and valid tools for adults with SCI. However, this cannot be confirmed due to a lack of SCI-specific information on the measurement properties of the employed tools.
We undertook a scoping review to address the lack of clarity on SCI-specific evidence for reliable and valid assessment of fat and fat-free mass, and help establish research priorities for SCI-specific body composition assessment. Scoping reviews, a type of knowledge synthesis, follow a systematic approach to map evidence on a topic and identify main concepts, theories, sources, and knowledge gaps [16][17][18]. We worked with an expert panel (including research, clinical, nutritional and lived SCI experience) on the design and interpretation of the review to ensure relevance of the results and help with identification of future research priorities [16,19]. The review's objectives were to: (1) map the evidence on measurement properties of tools to assess whole-body and regional fat and fat-free mass in adults with SCI; and (2) identify research gaps in order to set future research priorities. studies, (3) study selection, (4) charting the data, (5) collating, summarizing and reporting the results, and 6) stakeholder consultation [20,21]. The scoping review followed the relevant aspects of the Preferred Reporting Items for Systematic Review and Meta-Analysis Extension for Scoping Review Protocols (PRISMA-ScR) guidelines [18]. The review protocol was registered at the International Prospective Register of Systematic Reviews (PROSPERO) under the identification number CRD42018090187.
The scoping review was designed and conducted together with an expert panel (n = 8, Table 1), which included experts with lived SCI experience (paraplegia), a clinical (e.g. rehabilitation doctor) or other practitioner background (e.g. nutritionist), and/or a research background (i.e. expertise on SCI, body composition assessment, exercise physiology and/or performance nutrition). The panel collaboratively determined the scope and design of the review, and interpreted the findings [19]. This included three roundtable discussion meetings (in-person and online), and further exchange via email. Additionally, the expert panel reached out to their professional networks to gain further stakeholder feedback via interactive practitioner surveys and workshops with people with SCI. The goal of gaining this additional feedback was to benefit from a wider range of views that complemented those of the expert panel, without claiming this feedback represented all the views of the SCI population and various clinicians/practitioners working with people with SCI. The research gaps identified by the review, together with discussions among the expert panel including consideration of the survey and workshop feedback, informed the formulation of future research priorities.

Identification of the research question
Guided by international standards for reporting and developing clinical practice guidelines [22], the expert panel specified the target population (i.e., adults with SCI, in order to focus on Table 1. The expert panel (n = 8) that collaboratively determined the scope and design of the review, and interpreted its findings. a population in most need of guidance) and outcomes (i.e., fat and fat-free mass, considering already existing guidance on bone mass [23-25]). The following research objectives and outputs for this scoping review were specified: 1. Map the evidence on measurement properties (i.e., reliability, validity, responsiveness) of tools to assess whole-body and regional fat and fat-free mass in adults with SCI, and identify gaps in the evidence; and 2. Prioritize future research directions based on the identified gaps and expert panel discussion including the views of clinicians, researchers, other practitioners and people with SCI.

Search strategy
PubMed, EMBASE (OVID) and the Cochrane library were searched for eligible studies up to April 1 st , 2020. The search strategy was developed for PubMed and modified for the other databases (S1 File). Keywords were a combination representing three concepts: SCI (e.g., "spinal cord lesion", "tetraplegia", "paraplegia", etc.); body composition assessment methods (e.g., "DXA", "bioelectrical impedance", "waist circumference", etc.); and body composition outcomes related to fat and/or fat-free mass (e.g., "body weight", "adipose tissue", "lean mass", etc.). Reference librarians verified the search strategy and the language was restricted to English, with case studies (N<3), unpublished studies, reviews and conference abstracts being excluded.
Eligible studies included an adult sample (�16 years) where at least 50% of the participants were reported as traumatic or non-traumatic SCI (excluding multiple sclerosis and spina bifida), in line with a previous review (15). The focus for this search was on body composition assessment tools related to fat or fat-free mass (excluding mineral-only or water-only measures and measures of muscle morphology such as muscle fibre size and number of muscle fibres). Any body composition assessment method was eligible, including but not limited to: BMI, waist circumference, skinfold thicknesses, BIA, ADP, hydrostatic weighing, DXA, CT, MRI, and isotope dilution. Eligible study designs included statistics and/or individual data to quantify one or more of the following measurement properties: • . Indirect refers to a method that assesses body composition using one estimate: e.g., an estimate of body fat % from the attenuation of two low-energy x-ray beams (DXA). Doubly indirect refers to a method that assesses body composition using two estimates: e.g., an estimate of body fat % from body density that was estimated from skinfold thicknesses.
• Responsiveness: The ability of the method to detect change over time when compared to the criterion and/or the convergent measure [26, 28].

Study selection
After duplicate removal, authors JTdeZ, AL and one other reviewer (see Acknowledgements) conducted title/abstract scanning and full-text screening (Fig 1). Each record and full text were judged independently by two reviewers (JTdeZ and AL); differences between the two reviewers were discussed and if necessary adjudicated by the first author JvdS.

Data charting (extraction) and synthesis
Data charting was conducted by one reviewer and verified by another (JTdeZ and VG-T). Data charted from the eligible studies included participant demographics, study design and assessment tools used, and statistics/results. Each study was coded for showing "acceptable" reliability and/or validity or not, based on a minimum ICC of 0.70, Pearson's r of 0.80, or equivalent [28,29]. If a study's reported ICC or r (or equivalent) was lower than 0.70 or 0.80, respectively, then "not acceptable" was used, while "inconclusive" denoted that a study's reporting of statistical outcomes was incomplete. Maps of the evidence were created using the data charted from the eligible studies (Figs 2  and 3). It provides a visual overview of the tools used in studies on reliability and/or validity, recognising that any measurement tool should be both reliable (e.g., test-retest reliability) and valid (e.g., at least convergent validity if not criterion validity) in order to recommend its use in research or clinical practice [30].

Stakeholder consultation
As described above under "Protocol and design", the expert panel used a convenience sample approach (i.e. reaching out to their professional networks) to benefit from a wider range of professional and SCI views that complemented those of the expert panel. Authors CB and JvdS hosted two small group workshops that included five people with lived SCI experience (three men and two women with paraplegia [n = 3] or tetraplegia [n = 2] for longer than one year) at Spaulding Rehabilitation Hospital (Boston, MA, USA). These participants drew on their own experience living with SCI as well as that of others, given that participants were selected for their connections to a large peer network. All participants had undergone body composition measurements as a part of their rehabilitation process in either a home, clinical, or sports setting. During the workshop, participants discussed with the hosts their views on the importance of measuring body composition, their preferences of and experiences with different tools, and barriers they had experienced while using the tools. These views, preferences and experiences were captured and synthesised qualitatively (see S2 File).
Authors VG-T, TG-P and JvdS facilitated two interactive surveys with 15 clinicians and other practitioners working in the field of body composition and SCI, with competencies as nutritional, elite sport, sport science, and/or clinical research practitioners. The surveys were conducted in group sessions at Loughborough University (UK) and Spaulding Rehabilitation Hospital (USA). On-line and live surveys combined with group discussions were used to gain the views, preferences, and experienced/perceived barriers with different body composition assessment tools. Responses were synthesised qualitatively (see S3 File).
For the surveys and workshops, after being informed about the project and anticipated use of their data, all participants provided verbal informed consent. Ethical clearance was obtained Number of studies with "acceptable" convergent validity (e.g. reported ICC > 0.70), that were "not acceptable" (e.g. reported ICC < 0.70), or "inconclusive" (e.g. reported statistics incomplete) for each of the most commonly evaluated assessment tools (wholebody and regional studies combined). https://doi.org/10.1371/journal.pone.0251142.g003

PLOS ONE
Assessment of body composition in spinal cord injury: A scoping review from the Loughborough University ethics advisory committee (human participants subcommittee; UK).

Results
Out of 13,670 identified titles, a total of 42 eligible studies [8, were identified: the bibliographic database search provided 37 articles that met the eligibility criteria, and an additional five articles were identified after scanning reference lists of articles included (Fig 1). Common reasons for exclusion were insufficient data or statistics to evaluate measurement properties, conference abstracts or commentaries only, and insufficient or unclear SCI sampling (e.g. <50% or not reporting the proportion of people with SCI as part of the total sample). Data charting details for each of the 42 eligible studies are provided in Table 2.

Study designs: Measurement properties
The 42 articles that met the eligibility criteria included five studies on test-retest reliability, four studies on intra-rater or inter-rater reliability, 37 studies on convergent validity, and one study on criterion validity (Figs 1 and 2). Studies on other measurement properties, such as responsiveness, were not identified.

Participant characteristics
As detailed in Table 2, a total number of 1,154 participants were included across the 42 articles, comprised of men (n = 1011) and women (n = 143) aged 16 to 71 years with cervical, thoracic or lumber lesion levels (C1 to L5), complete or incomplete SCI (American Spinal Injury Association Impairment Scale [AIS] A, B, C and D), mostly with a relatively long-standing SCI (all studies included participants ranging from 1 to 36 years post-injury ['chronic' SCI], except for three studies [38,56,60] including participants ranging from 3 days to 4 months post-injury ['acute' SCI]). All studies included SCI-only samples, except for one study that included a sample of at least 50% people with SCI mixed with participants with spinal bifida [32]. In three articles, participants were described as wheelchair athletes [33,41,54], while the other articles provided no or limited information about physical activity levels or sports background. Fig 2 shows the 42 studies represented as those evaluating reliability on the left and the same assessment tool showing validity across from it on the right; each assessment tool bubble indicates whether they were whole-body or regional measures. Fig 3 displays in more detail the number of studies reporting acceptable convergent validity for each of the most commonly evaluated assessment tools (i.e., DXA, conductance [e.g., BIA], skinfold thickness, waist circumference, and BMI). Details on each study are provided in Table 2.

Body composition assessment tools
Whole body composition. A three-compartment model of whole-body composition was assessed using DXA in one test-retest reliability and two convergent validity studies; all three studies reported acceptable reliability [42] or convergent validity [56,62]. Of note, DXA was used as the reference tool for most of the convergent validity studies on doubly-indirect methods (e.g., conductance, skinfolds, ADP).
A two-compartment model of whole-body composition (total body electrical conductivity) in which participants were positioned in a whole-body cylinder was assessed for its test-retest reliability and was found to be acceptable (55). While no validity studies are available for this conductance method, seven other studies have assessed convergent validity of BIA compared to DXA [34,41,54,62] or total body water [32,38,56,62]; the majority were found to be  CV for TB and regional measures between 2-6%; ICC for TB and regional measures >0.97   inconclusive or not acceptable. Another two-compartment model of whole-body composition (skinfold thickness) showed acceptable test-retest reliability [33], however only three of the eight studies assessing whole-body convergent validity of skinfold thicknesses were acceptable when compared to DXA [41,49,54,62,69], total body water (38,62), conductance [55,62], or hydrostatic weighing [33].
Other two-compartment models of whole-body composition (i.e., ADP, total body water, total body potassium) have not been assessed for their reliability; their convergent validity was reported as not acceptable (ADP) [41], not acceptable (total body water) [40] and acceptable (total body water and total body potassium) [62]. One study reported acceptable criterion validity of ADP and hydrostatic weighing when compared to a four-compartment model [36]; however this result of acceptable validity for total body volume was limited to a sample of participants with paraplegia in whom thoracic volume was not obtained and who were able to sit upright without support in the small chamber.
Lastly, the most commonly assessed whole-body composition tool for convergent validity was BMI, although no reliability studies exist. Studies compared BMI to DXA [31,37,46,57,59,60,63,69,71] or total body water [32]; none of the studies reported acceptable convergent validity. The consensus from the literature is that BMI values underestimate body fat; in other words, a healthy BMI may mask excessive adiposity in persons with SCI.
One study using ultrasound assessed inter-and intra-rater reliability of determining the soft tissue layers around the ischial tuberosity and reported some layers to be acceptable and

PLOS ONE
others not acceptable [65]; another study assessing convergent validity of ultrasound for subcutaneous adipose tissue and visceral adipose tissue was not acceptable [39]. A study using MRI to measure subcutaneous adipose tissue and visceral adipose tissue reported inconclusive inter-and intra-rater reliability and not acceptable convergent validity [44]; one other study using MRI to measure thigh/lower leg fat infiltration (i.e., muscle quality) reported acceptable inter-rater reliability [61]. Convergent validity was not acceptable in the one study using partial body potassium to assess legs' lean mass compared to DXA [68]. Lastly, two studies assessing measurement properties of analysis techniques reported inconclusive test-retest and inter-rater reliability of pQCT-derived watershed and threshold algorithms [70], and acceptable convergent validity of an MRI automated segmentation technique [52].

Stakeholder consultation
Overviews of the stakeholder consultations with persons with lived SCI and body composition assessment experience as well as with practitioners working in the field of SCI and body composition were summarised (see S2 and S3 Files for details).
The stakeholders with SCI were very clear in their feedback regarding the importance of knowing body composition for their own personal health and empowering their health-related decision making. They felt there was not enough information available on most body composition assessment tools for them to be used with confidence. The stakeholders with SCI considered the measurement burden of DXA acceptable as long as it was reliable, valid and would do no harm (e.g. radiation, skin damage). Practitioners also indicated their confidence that DXA is a reliable and valid assessment tool with acceptable measurement burden for measuring body composition. However, they considered cost and accessibility major barriers, particularly for sport performance settings.
When discussing two-compartment models including conductance, skinfolds, and ADP, opinions and experiences varied. Views on conductance (e.g., BIA) among persons with SCI and the practitioners were mixed and mostly negative. Some of the persons with SCI looked favourably upon the potential of a home-based tool, while others had experienced severe adverse events trying to use it as a standing tool; cost and accessibility were the main barriers identified by the practitioners. Persons with SCI perceived skinfold thicknesses to be a simple tool with relatively little time, effort and equipment required, but were not sure if and how this could be conducted without trained staff. Moreover, practitioners perceived skinfolds to be a reliable tool to monitor intra-individual changes over time, but not for estimating body fat, and noted caution towards skin damage and ensuring correct body positioning. ADP was described as having cost and accessibility barriers by the practitioners. The expert panel noted additional practical barriers based on their experience using ADP among athletes with SCI including participant preparation time, difficulties assessing thoracic volume, and risk of pressure injuries.
Persons with SCI and practitioners had very little confidence regarding the use of BMI and expressed concerns about invalid comparisons to able-bodied normative values. Persons with SCI expressed a preference towards simple tools, but only if they were accurate (e.g., there was little confidence in measuring waist circumference in a seated position). Other less common assessment tools, including ultrasound, MRI, pQCT, body water, and body potassium, were considered difficult to access or cost-prohibitive by the practitioners surveyed. The expert panel regarded many of these tools potentially useful for mechanistic clinical or research studies.

Discussion
The present scoping review identified a large range of body composition assessment tools and outcomes evaluated in SCI-specific research on measurement properties. Despite the common use of DXA, conductance, skinfolds, waist circumference, and BMI to measure aspects of body composition (e.g., whole-body, regional fat mass or fat-free mass), there is surprisingly little research evidence on test-retest reliability, criterion validity or responsiveness of these tools in the SCI population.
Our review found that whole-body DXA is the only tool with both test-retest reliability [42] and convergent validity [56,62] evidence to support its use in SCI practice-given that any measurement tool should be both reliable and valid in order to recommend its use in research or clinical practice [30]. However, DXA is costly and inaccessible to many in the field, causing challenges for clinicians, researchers, and other practitioners to obtain reliable and valid body composition measures [11,13].
Waist circumference and skinfold thickness measurements may hold the most promise as more practical and affordable body composition tools, as evidenced by some of the studies identified in this review [8,33,37,47,51,62,64,66,69] and a recently published study [72]. The promise was also highlighted by the views and experiences of the expert panel and the other stakeholders consulted in this study. However, further support is needed for their reliability and validity. Both tools have evidence from a single study reporting acceptable test-retest reliability [8,33]. They have been commonly researched with regards to convergent validity: 11 studies evaluating waist circumference and 12 studies evaluating skinfold thicknesses against DXA, CT, MRI, or total body water [8,35,37,38,41,44,45,49,54,55,57,59,62,64,69,71]. The majority of these reported statistics that were inconclusive or below threshold values for showing acceptable convergent validity (e.g. minimum ICC of 0.70) [28,29]. These varied results regarding convergent validity of waist circumference and skinfold thickness measures may be explained by the heterogeneity of the samples included (e.g., injury characteristics), methods of assessment (e.g., waist circumference supine vs sitting, with the latter less likely to be valid), and standardization of measurement conditions (e.g., time of day, before or after exercise, before or after bladder emptying, equipment and software type of the reference method).
Well-established standardization of measurement conditions are of great importance for precision of body composition measurement [73,74], including that of waist circumference [cf. 72] and skinfold thicknesses [cf. 75]. In addition, it is important to note that many of the body composition assessment tools require a trained professional.

Research gaps
DXA was used as a reference tool in the majority of studies (19 of the 37 studies) assessing convergent validity of doubly-indirect methods (e.g., conductance, skinfolds, ADP). It should be cautioned, then, that only one study has evaluated the test-retest reliability [42] and two the convergent validity (compared to total body water) of whole-body DXA in persons with SCI [56,62]. Four other studies have evaluated the convergent validity of regional DXA of the trunk or thigh compared to MRI [48,50,53,58]. Despite the potential of MRI [76], far less SCI-specific evidence was identified for it (e.g., no test-retest reliability studies) than for DXA, possibly due its higher costs and lesser accessibility.
Evidence-based guidelines [22] for proper scan acquisition and analysis in persons with SCI are warranted, as well as more reliability studies using both major commercial manufacturers (GE Medical Systems Inc [formerly Lunar] and Hologic Inc.) [77], and criterion validity studies using a 4-compartment model as a comparator. Such guidelines and studies should also recognize its limitations: obtaining reliable measures in persons with SCI may be challenging due to physical (e.g., contractures, spasticity, hardware in the body, urine reservoirs, obesity, etc.) and logistical (e.g., transferring, positioning) barriers introducing error of measurement [14,78].
Another important gap identified is that on waist circumference and skinfold measurements. While both waist circumference and skinfold thickness measures are low cost, easy to use, and accessible [51,59,64,66,72,75], further research is still required to identify standard techniques, sites, and their reliability, validity, and responsiveness in the SCI population, as indicated by the varying results on convergent validity identified in this review. Notwithstanding, both tools show premise as more practical, accessible and affordable body composition tools [51,64,66,72], particularly for longitudinal tracking within individual subjects [59,75]. As a caveat, this premise holds for skinfold thickness measures only (if international standards are followed and damaged skin is avoided), rather than any equations that predict percent body fat from skinfolds.
Other important gaps in the literature include the insufficient inclusion of females; only 12% of the sample in the available literature were females, which is much lower than the historical demographics of SCI reported by International Spinal Cord Society [79]. It is possible that some assessment tools may require sex-specific protocols, as highlighted by studies reporting sex dimorphism [31,54,80,81]. Further, most of the studies in the present review involved cohorts with heterogenous injury characteristics (i.e., considerable variation in times since injury, severity of injury, and level of injury); more individualized approaches are needed to identify possible discrepancies in reliability and validity of body composition assessment tools related to injury characteristics.

Priorities for future research
Overall, our review suggests that the evidence base for reliable and valid assessment of SCI body composition requires further strengthening, and that it is important for practitioners, researchers and people with SCI to continue to work together with this challenge in mind. Accordingly, we formulated a set of priorities and considerations for research in this area (Table 3). These were informed by the available evidence and identified research gaps in the review, and further sensitized by the practical expertise and lived SCI experiences of the expert panel, which were further enriched by the stakeholder workshops and surveys. This ambitious research agenda can be facilitated by encouraging robust, standardized data collection through international, multi-center collaboration. Such data collection can provide sufficiently sized samples, allow sub-group analyses (e.g., according to age, sex, training status/ physical activity levels), and can substantiate ecological validity (e.g., visceral fat or waist circumference as SCI-specific cardiovascular health indicators, skinfold thickness measures as wheelchair-sport performance indicators). With this dataset, international normative values and clinical practice guidelines for body composition assessment in SCI can be established. It would also provide the foundation for internationally standardized methods and outcome measures for SCI trials such as longitudinal exercise and nutrition trials. As highlighted by others [82] and by our stakeholder consultations, any document guidance must be inclusive of describing any associated risks and accessibility concerns in individuals with SCI.
The panel recommends prioritizing research on the tools described above (i.e., DXA, waist circumference, skinfold thicknesses), rather than techniques that have theoretical flaws for SCI body composition measurement and for which the evidence is currently-at best-inconclusive. Although MRI also holds high promise for reliable and valid assessment in SCI (76), its relative inaccessibility and high cost may continue to limit its use to highly specialized clinical and research studies.

Strengths and limitations
This is the first systematic scoping review on measurement properties of various tools to assess fat and fat-free mass in adults with SCI, in accordance with international standards [16][17][18]. The collaborative work of the expert panel ensured the representation of the voices of various stakeholder groups [19,82], and facilitated setting priorities for future research [83]. The expert panel used a convenience sampling approach to reach out to their professional stakeholder networks. This further stakeholder feedback was only used to sensitize the panel's views, while remaining aware of potential bias that could occur due to this convenience sampling approach.
For feasibility reasons, the expert panel decided to limit the search for evidence to the most essential measurement properties required to recommend a measurement tool [26]. A wider search including other types of measurement properties (e.g., ecological validity) was piloted, but not considered feasible given the titles and abstracts to scan (n = 10,241), and was unlikely to change the review's conclusions. The limited reporting in the eligible studies of standardized conditions under which the measurements were conducted (e.g., positioning in the scanner, bladder voiding, position in which participants was sitting/lying, if the same investigator conducted each measurement [73,74]) surfaced during piloting of the data charting; an attempt to systematically capture these conditions as part of the data charting was deemed infeasible.

Conclusions
This scoping review has provided more clarity on what SCI-specific evidence does and does not exist for the reliable and valid use of body composition assessment tools. The review can support practitioners and researchers working with people with SCI on assessment of body composition. For the time being, the use of DXA under well-standardized measurement conditions, with awareness of the current limitations in the evidence base, is suggested to practitioners and researchers aiming to assess the effects of nutrition and/or exercise-interventions on whole-body and regional fat-and fat-free mass in adults with SCI. The available evidence and gaps identified in this review affirmed that future research should prioritise reliability and validity studies on: (1) DXA as a surrogate 'gold standard' tool to assess whole-body and regional fat and fat-free mass, (2) waist circumference and skinfold thickness measurements as practical low-cost tools to assess whole-body and regional fat mass, and (3) females to explore potential sex differences of body composition assessment tools. Such studies can provide the required evidence to develop normative values and clinical practice guidelines for assessment of body composition in SCI.