Sporting population studies.

These types of studies (Table 3) consist of single measure, cross-sectional protocols aiming to characterise sporting group samples in terms of bioelectrical data. As observed by Koury et al. [43], athletes exhibit similar trends of PA variation with age to those of the general population of the same sex and age, with a positive correlation (r = 0.63, p = 0.0004) in adolescents and a negative correlation (r = -0.27, p = 0.009) in adults. Vectors shifted to the left and with greater PA were found in both adolescent and adult athletes compared to the corresponding reference populations, which is consistent with the results reported by other studies for soccer players [41] and synchronised swimmers [57], suggesting that these differences are due to sport-specific adaptations [41]. In comparison with adolescent athletes, the mean vector of adult athletes also showed a shift to the left. Both shifts to the left indicate increased BCM and fluid content, and might reflect a better cell functioning [41].

Regarding the vector position on the RXc graph, the trend is to be outside the 50% tolerance ellipse of the respective reference population in both adolescent and adult athletes. According to this, Piccoli et al. [42] also found the mean impedance vector of bodybuilders almost completely outside the 95% tolerance ellipse of the reference population. This reflects a specific body composition and suggests that specific tolerance ellipses are needed for sport populations [36, 41, 57]. To date, only two studies [41, 57] have characterised sport-specific populations. The relationship between the new specific tolerance ellipses (for each sport, gender, age and race) and the hydration status, body composition and sport performance level should be analysed, in order to represent significant hydration changes (that compromise health or performance) or target zones of impedance vectors for athletes. Nevertheless, it is possible that a new approach is required for the exercise and sports field, beyond the current BIVA point graph, based on 50–95% tolerance ellipses and quadrants related to clinical outputs. With regard to the hydration assessment, it should be noted that fluid overload (overhydration) is not common in healthy athletes. Therefore, the analysis of the hydration status should be related to euhydration and physiological dehydration processes. In this way, as mentioned in Heavens et al. [68] regarding the identification of dehydration with single and serial measurements according to the tolerance ellipses of the reference population, the limits for “normal hydration” (individuals positioned within the 50% tolerance ellipses, according to the literature [18, 69]) should be reviewed, since subjects experiencing high levels of fluid loss can still be identified as euhydrated [68]. Other studies related to sport and exercise [47, 49] identified some individuals as euhydrated after significant BM decreases. Moreover, as shown in Table 5, the majority of the studies analysed identify the athletes outside the 50% tolerance ellipse. This is probably due to a range of “normal hydration” comprised by the ellipses wider than a hydration status/change considered as “dehydration” through other methodologies [68]. Nevertheless, the conclusions of Heavens et al. [68] should be confirmed with the appropriate methodology, since the study was not performed with a phase-sensitive device, and therefore, they could not obtain the real value of Xc. Therefore, although directional changes in vector values from serial measurements seem to be consistent with fluid loss, the current BIVA point graph is not a valid method to detect dehydration in individual athletes. Research investigating different levels of dehydration and their relationship with the new specific tolerance ellipses is needed in order to identify the limit of “normal hydration”. Furthermore, different types of dehydration can be experienced in sport: a) hypertonic dehydration (i.e. primarily a loss of water) is a common type of dehydration developed after exercise in which heavy sweating occurs; b) hypotonic dehydration (i.e. primarily a loss of electrolyte) and c) isotonic dehydration (i.e. equal losses of electrolytes and water), both may be developed by athletes competing in aesthetic-type sports and in weight classification sports in which fasting, vomiting and diuretic use are common behaviours [65]. Thus, research is needed related to the sensitivity of “classic” BIVA to each type of dehydration, as well as the behaviour of each one with regard to the tolerance ellipses. On the other hand, it should be investigated the relationship between the new specific tolerance ellipses and different sport performance levels. Maybe different sectors of the tolerance ellipses identify target zones for the athletes. With regard to the body composition assessment and in accordance with “classic BIVA”, athletes have been identified in the upper left quadrant of the reference population and obese individuals in the lower left quadrant. This would generally imply greater R/h and Xc/h values of the athletes. Nevertheless, as mentioned in the literature [22, 70], according to the electro-physical assumptions, FFM is characterised by a greater conductivity in comparison with the poorly hydrated adipose tissue, not justifying the relative shortness of vectors of obese individuals with respect to the athletes, unless contemplating their generally greater FM, fluid overload and body size. Furthermore, the vector position of athletes regarding the tolerance ellipses of the general reference population is controversial [4]. As mentioned by Buffa et al. [4], athletic individuals are not always plotted in the “athlete” quadrant of the reference population and their vectors often overlap the “obesity” area. This controversy can be observed in Table 5. From the nineteen investigations analysed, six studies did not report vectors distribution with regard to the reference population and only four found the majority or all the vectors of athletes positioned in the “athlete” area [47, 50, 54, 56]. Comparable vector position of athletes and obese individuals would imply similar values of R/h and Xc/h. The already mentioned factors FM and fluid overload could compensate the bioelectrical values between both individuals, not being “classic” BIVA (50 kHz) able to detect the differences (e.g. discriminating fluids distribution between compartments, with greater ICW content in athletes). Moreover, as mentioned in the literature [22, 70], “classic BIVA” would be characterised by a limited sensitivity in assessing the features of body composition (i.e. FM and FFM) due to the no consideration of the effect of cross-sectional areas of the body which interferes with bioelectrical values as well as lengths, according to the basic conductor theory (impedance is proportional to the conductor length and inversely related to its cross-sectional area) [71]. This effect of cross-sectional areas is particularly relevant in sport sciences because athletes of different disciplines generally differ in their body shape. To overcome this limitation of “classic” BIVA, a relatively new procedure (“specific” BIVA) has been developed [22, 70]. This method proposes a correction of bioelectrical values for body geometry and it has proven to be effective in identifying the relative proportion of FM in adults and elderly [22, 70]. Although the inclusion of anthropometric measurements can make these plots more sample-specific and perhaps less generalizable than “classic” BIVA, this adaptation may be an advance when comparing athletes with different body composition (in terms of FM and FFM). Therefore, it should be further investigated in the sports field.

Koury et al. [43] observed that the distance between the confidence ellipses of adolescent and adult athletes was lower than between the ellipses among their respective reference populations, either considering all sport modalities or only paired modalities. The authors speculated that the intense training reduced the differences between adolescent and adult individuals, although this is still to be elucidated. In their study, vector and PA differences were due to differences in R/h, significantly lower in adult athletes than in adolescent athletes, with no differences in Xc/h. Similar to these findings, Micheli et al. [41] reported that in soccer players of higher competitive level, vectors shifted to the left due to a decrease in R/h, with no difference in Xc/h compared to those in lower soccer divisions. This shift to the left was also found between elite and high-level players. As suggested by Micheli et al. [41], these results reflect different ICW content (adult > adolescent; higher > lower sport levels), since the ECW/TBW ratio is inversely related to PA [72], and it could be due to the hypertrophy of muscle fibres. Furthermore, despite similar training loads among players of the highest level, differences may be due to different individual responses to the training load, or they could also be an indicator of better training and/or recovery strategies in elite teams [41]. Carrasco-Marginet et al. [57] also reported a shift to the left, with no difference in Xc/h, in young synchronised swimmers of higher competitive level. Nevertheless, since higher-level swimmers were older than the lower-level ones, it should be investigated whether the differences were due to biological maturation, to specific training or a combination of both.

As noted, a greater PA accompanying a vector shifted to the left has been observed in adult athletes compared to the healthy reference population [36, 41–43, 46, 49, 50]. This is due to i) a decreased R/h as a result of a different body composition, probably due, among other factors, to a greater muscle mass, muscle glycogen reserves and plasma volume [73, 74], and ii) an increased Xc/h, probably due to an increase in the size and number of muscle cells (hypertrophy and hyperplasia, respectively), although the last one is still a controversial topic [75]. However, since a decreased R/h is also related to greater FM [33], further research is needed in order to clarify the reason for this behaviour. Furthermore, Xc/h is not only conditioned by the cell size, but also by the thickness and composition of the cell membranes and also by the distance between them, due to their relationship with membrane capacitance (Cm) [76]. In this way, lower Xc/h values have been documented in bodybuilders (the best model of extreme muscle hypertrophy) compared to healthy active people and with no differences with the healthy reference population [42]. However, vectors shifted to the left with lower PA have been reported in competitive children in comparison with healthy control groups due to significantly lower Xc/h values in absence of differences in R/h [53]. The authors suggested that it could be due to an increase in the size of the section of the limbs or to a greater ‘sufferance’ in cell membranes maybe due to bad response to the workloads (over-training). Therefore, the interpretation of Xc/h in these cases remains unresolved.

Nescolarde et al. [36] reported differences in both whole-body and localised mean Z vectors of soccer and basketball players, attributed to the different body structure between both disciplines. Soccer players presented a whole-body vector shifted to the right on the BIVA graph compared to basketball players, due to greater R/h and Xc/h. Regarding the localised vectors, soccer players showed a shift to the left of quadriceps and hamstrings vectors, due to a decrease in R/h and an increase in Xc/h. On the other hand, gastrocnemius vectors of soccer players showed a shift to the right, due to an increase in R/h and Xc/h. The muscle groups in lower-limbs were found to be symmetrical in athletes and this could be used to detect changes in hydration and/or muscular structure.

Short-term vector changes (<24 h after exercise).

These types of studies (Table 1) are those which currently face more difficulties, since their validity can be easily compromised, mostly because of several factors that may affect the accuracy of the measurements despite any attempts to control them. To date, two studies have investigated the vector adaptations using this type of design.

Gatterer et al. [47] analysed the short-term bioelectrical adaptations in well-trained subjects after 1 hour of self-rated intensity cycle ergometer test in the heat (environmental chamber). They reported an increase in both R/h and Xc/h after exercise, as well as significant vector migration indicating fluid loss. Besides, they pointed out a negative relationship between changes in Xc/h and in plasma osmolality (P_osm) (r = -0.58). The authors concluded that “classic” BIVA changes mirrors water loss during exercise in the heat, and that changes in Xc/h values reflect fluid shifts between intracellular and extracellular compartments. As mentioned before, Xc is related to Cm, which is affected by the size, thickness, composition and distance between cell membranes [76]. Exercise generates processes which modify the characteristics of muscle cells (such as changes in fluid distribution). As suggested, the cell membrane becomes thinner as the cell swells and Cm increases, and the opposite happens as the cell shrinks [77], thus affecting Xc. Besides, as the cell swells, the distance to the adjacent cell membranes decreases and Cm increases (the opposite happens as the cell shrinks), also affecting Xc. Moreover, in accordance with De Lorenzo et al. [78], variations in fluid distribution would change the impedance locus and, consequently, the characteristic frequency (Fc), defined as the frequency at which Xc presents a greater value and that it is close to 50 kHz. Thus, these variations would evoke considerable changes in Xc at 50 kHz, the frequency used in BIVA [79, 80]. Nonetheless, De Lorenzo and collaborators’ hypothesis should be considered with caution because it refers to the Hanai’s model, which relays on assumptions such as spherical cell shape. Therefore, multiple factors may affect Xc values and further research should focus on this parameter in exercise.

According to Gatterer et al. [47], Carrasco-Marginet et al. [57] reported significant vector displacements along to the major axis after exercise due to significant increases in R and Xc. Furthermore, the mentioned study showed that BIVA paired graph seems to identify significant vector differences after exercise inducing mild dehydration (average loss of <1% BM) in different groups of athletes.

In opposition to both studies [47, 57], Antoni et al. [51] only found a tendency to reduction of fluids (the authors related it to an extracellular water decrease given by a significant increase in Xc) along with an increased BM in a group of men and no differences in women after approximately 10 hours of subterranean exploration (caving). Factors affecting protocols measuring pre- and post-exercise (such as dietary intake during cave activity or the skin temperature in the post measurement) could have influenced their observations. Nevertheless, despite the fact that the vector changes after fluid removal and overload (the wet–dry cycle of dialysis) as a non-physiological process is clinically well-established [69], every dehydration process induced by physical exercise is consequence of scarcely explored physiological adaptations as regard of the vector behaviour, especially at cellular level (and therefore, affecting R and Xc). In literature, Xc is an indicator of dielectric mass (membranes and tissue interfaces) in soft tissues [71]. Given the results observed in sport, it is possible that the behaviour of Xc could be due to other factors and, thus, its meaning remains to be clarified.

Long-term vector changes.

Studies investigating long-term (≥7 days) vector adaptations (Table 2), have some protocol-specific advantages in comparison with investigations focused on acute vector changes, mainly because the quality of the bioelectrical signal can be assessed independently from the acute adaptations related to exercise.

BCM and extracellular mass (ECM) have been proposed as representatives of ICW and ECW, respectively [46]. Nevertheless, it is important to note that the estimation of fluid volumes and cell mass with BIA prediction models is inappropriate when discussing changes in vector positions after interventions or treatments. Gatterer et al. [46], in their study assessing body composition using “classic” BIVA in the 2008 European Soccer Championship, found a significant lengthening of the vector within a period between 1 and 2 weeks. They attributed it to changes in BCM and ECW in both starters and non-starters after the first match with respect to baseline values, indicating body fluid loss. After the second match, only the athletes who played more (starters) showed a significant lengthening of the vector possibly due to a decrease in ECW. Therefore, they concluded that changes in body composition were mainly due to changes in ECW. However, their results should be taken with caution, since only analysis with appropriate reference methods (e.g. isotope dilution) can support them.

Similarly to the results of Gatterer et al. [46], rapid loss of BM protocols within a few days before competition in boxers [50] was found to be achieved mainly by isotonic dehydration (they attributed it principally due to changes in ECW), as identified by the significant vector lengthening on the RXc point graph and the decreases in plasma and blood volume. Nevertheless, as mentioned before, their results should be further investigated with appropriate reference methods for the estimation of fluid volumes, since BIA prediction models are inappropriate to discuss changes in vector positions. According to the results of Reljic et al. [50], Piccoli et al. [49], also found a significant lengthening of the vector with isotonic dehydration at high altitude (5500 m). Nevertheless, although a subsequent hypertonic dehydration was identified by a decreased BM (-3.0 kg) and several hydration biochemical markers, the vector lengthening was not significant. The causes that explain why the vector remained unchanged after such a BM loss were not elucidated, and the authors recognised the difficulty of explaining the metabolic reasons that led to such BM reduction. In any case, emphasis should be placed on the importance of not considering body fluids quantitatively only (i.e., volume), but also regarding their qualitative composition, due to the biological adaptations generated by different types of exercise. For instance, after descent to sea level, the impedance vector underwent a significant shortening and returned close to baseline values. Lastly, significant relationships were found between changes in bioelectrical variables (R/h and Xc/h) and changes in the following hydration biomarkers along measurements performed at altitude and at sea level: BM, urine volume, P_osm, serum Na⁺, K⁺, Cl^- and glucose, and urine osmolar excretion [49].

On the other hand, two studies [55, 56] found significant shortening of the vector along three weeks of multistage road bicycle race, indicating fluid gain during the tour and attributing these results to muscle oedema, haemodilution, released water from muscle glycogen oxidation, and excess fluid intake. Although the vector shortening was not related to power output or rating of perceived exertion [55], it was negatively associated with performance during the last stages [56], suggesting the authors that increases in plasma volume and improved thermoregulatory capacity could explain these outputs. Nevertheless, their results should be taken with caution, since measurements were performed approximately two hours after exercise and this could have altered the data.

Regarding studies analysing longer-term vector adaptations, Mascherini et al. [48] analysed a soccer team across a sport season and reported a significant shortening of the vector in the pre-season associated with an improvement in endurance performance possibly due to plasma volume expansion and enhanced glycogen storage. These results are in agreement with other studies [45, 54] which also found significant bioelectrical differences in the pre-season, hypothesising that they were due to fluid expansion. Bonuccelli et al. [45] and Macherini et al. [48] found a significant lengthening of the vector in the mid-season compared to pre-season results. This could indicate a reduced body fluid volume (i.e., decreased plasma or interstitial volume) despite an increased intracellular fluid associated with an increase in BCM, and consequently in PA [41]. However, while Mascherini et al. [48] reported a significant shortening of the vector at the end of the season compared to the mid-season, Bonuccelli et al. [45] observed a significant water content decrease. Sport calendars could have led to adopt training strategies inducing different performance status and evoked opposite vector displacements.

On the other hand, regarding the age-related decreases in Xc and PA [81], improvements have been reported after six months of resistance training in elderly women [52], suggesting increased amount and quality of soft tissues. These improvements were accompanied by increases in leg strength and thigh circumference. Along with these changes, BIVA showed a significant vector migration after the training program.

With regard to children, one study [53] evaluated the body composition in participants of swimming and gymnastics along one year. The baseline measurement (T0) was performed at a period preceding races and sporting events, just as the third measurement (T2) one year later. The second measurement (T1) was executed six months after T0 in a period characterised by a softer daily training. They found a significant increase in Xc from T0 to T1, along with increased PA and ICW (derived from ECW/TBW ratio). The authors hypothesised that this was due to an improvement in the muscular trophism with higher levels of intracellular proteins and glycogen and to a lower stress from training program. After one-year follow-up, no significant differences were found in R, Xc and PA. However, again, their hypotheses should be taken with caution, since fluid estimations were calculated from BIA prediction models. Variables as the type of sport and training strategy should be taken into account when monitoring along a season, since they might influence the bioelectrical measures. Moreover, also intra-group comparisons between seasons should be analysed with caution, since inter-seasonal bioelectrical variations could be effected by factors such as biological maturation.

Injury identification and follow-up.

These studies [36, 37] consisted in single cross-sectional protocols aiming to identify bioelectrical patterns of change depending on the injury type and grade, and longitudinal protocols aiming at assessing bioimpedance vector sensitivity to monitor injuries and their recovery. R and Xc were found to be decreased in the injured muscles due to the oedema and to the disruption of the muscle structure, respectively. Furthermore, the more severe the injury was, the more R and Xc were decreased. On the other hand, a bioelectrical symmetry between muscular groups in lower-limbs was found. The follow-up of the injury identified bioelectrical patterns of changes similar to those in wound healing and an increase of R and Xc values were observed to values close to pre-injury.

Overall, localised bioimpedance vector analysis appears as an alternative method that could help to assess soft tissue injury and to monitor the injury recovery process [36, 37].