**Bioimpedance Vector References Need to Be Period-Specific for Assessing Body Composition and Cellular Health in Elite Soccer Players: A Brief Report**

**Tindaro Bongiovanni 1,2, Gabriele Mascherini 3,\* , Federico Genovesi <sup>4</sup> , Giulio Pasta <sup>5</sup> , Fedon Marcello Iaia <sup>2</sup> , Athos Trecroci <sup>2</sup> , Marco Ventimiglia <sup>6</sup> , Giampietro Alberti <sup>2</sup> and Francesco Campa <sup>7</sup>**


Received: 27 August 2020; Accepted: 29 September 2020; Published: 1 October 2020

**Abstract:** Purpose: Bioimpedance data through bioimpedance vector analysis (BIVA) is used to evaluate cellular function and body fluid content. This study aimed to (i) identify whether BIVA patters differ according to the competitive period and (ii) provide specific references for assessing bioelectric properties at the start of the season in male elite soccer players. Methods: The study included 131 male soccer players (age: 25.1 ± 4.7 yr, height: 183.4 ± 6.1 cm, weight: 79.3 ± 6.6) registered in the first Italian soccer division (Serie A). Bioimpedance analysis was performed just before the start of the competitive season and BIVA was applied. In order to verify the need for period-specific references, bioelectrical values measured at the start of the season were compared to the reference values for the male elite soccer player population. Results: The results of the two-sample Hotelling T<sup>2</sup> tests showed that in the bivariate interpretation of the raw bioimpedance parameters (resistance (R) and reactance (Xc)) the bioelectric properties significantly (T<sup>2</sup> <sup>=</sup> 15.3, F <sup>=</sup> 7.6, *<sup>p</sup>* <sup>≤</sup> 0.001, Mahalanobis D = 0.45) differ between the two phases of the competition analyzed. In particular, the mean impedance vector is more displaced to the left into the R-Xc graph at the beginning of the season than in the first half of the championship. Conclusions: For an accurate evaluation of body composition and cellular health, the tolerance ellipses displayed by BIVA approach into the R-Xc graph must be period-specific. This study provides new specific tolerance ellipses (R/H: 246 ± 32.1, Xc/H: 34.3 ± 5.1, r: 0.7) for performing BIVA at the beginning of the competitive season in male elite soccer players.

**Keywords:** BIVA; phase angle; R-Xc graph; tolerance ellipses

#### **1. Introduction**

Body composition analysis is currently one of the most studied evaluations in sport, mainly for the relationship between physical characteristics and sports performance [1]. In sports, excess fat mass reduces endurance performance, while an increase in lean mass, especially muscle mass, is associated

with an increase in power and strength [2]. Furthermore, the assessment of localized body composition allows the identification of differences in muscle mass and strength between areas of the body and may allow a reduction in the risk of injury (evaluation of contralateral limbs, agonist-antagonists) [3].

Body composition assessment should also be considered in sports involving weight categories, where athletes benefit from being placed in a lower weight category, in these cases any weight loss must therefore be monitored closely. Excessive training coupled with calorie restrictions can lead to excessive, unnecessary and dangerous weight loss. This weight loss in both women and men decreases performance, bone mineral density, muscle mass and is detrimental to health [4,5].

Bioelectrical impedance vector analysis (BIVA) is a method widely used to evaluate body composition and cellular health in athletes, as well as in the general population [6–9]. This method considers the raw bioelectrical parameters (resistance and reactance) standardized for the height of the subjects as a vector within a graph. Resistance (R) is the opposition to the flow of an injected alternating current, at any current frequency, through intra- and extra-cellular ionic solutions, while reactance (Xc) represents the dielectric or capacitive component of cell membranes and organelles, and tissue interfaces [10].

BIVA allows for the monitoring of vector changes over time or the comparison of the vector position within the R-Xc graph on specific population tolerance ellipses [11–13]. Given the ease and repeatability of this method, several references for athletes have recently been proposed, including those for soccer players [14], volleyball players [15], and cyclists [16], while also considering the competitive level of the athlete.

In soccer, Levi Micheli et al. [14] were the first to demonstrate how athletes need to be assessed on specific tolerance ellipses, showing bioelectric values that were far different than those of the normal healthy population. Subsequently, Mascherini et al. [17] suggested how bioimpedance vectors show displacements over the season, reflecting the changes that occur in the body composition and physical condition of the players. This was later confirmed by Campa et al. [18] who analyzed the bioelectrical changes comparing BIVA to results obtained by Dual X-ray Absorptiometry (DXA) and dilution techniques over a season in athletes, also showing that these vector changes occur in many other sports.

During the different phases of competition, the one which precedes the start of the season is among the most important periods in which to evaluate the athlete's physical condition and the body composition adjustments that are sought during the pre-season. Considering the vector changes that occur over the season, the bioelectrical references used in the BIVA assessment must be specific for the competitive period in which the athlete is tested. Therefore, the purpose of this study was to show how BIVA references provided in different phases of the season differ in male elite soccer players, also providing new references for assessing body composition in the start-of-the season period.

#### **2. Materials and Methods**

#### *2.1. Design and Participants*

A total of 131 male professional soccer players (age: 25.1 ± 4.7 yr; height: 183.4 ± 6.1 cm; weight: 79.3 ± 6.6 Kg) were recruited and participated in this observational study.

The inclusion criteria were: (1) players registered and participating in the first (Serie A) Italian National division; (2) non-injured at the time of the assessment. After having been informed about the aims and the procedures of the research, all athletes gave their written informed consent. The project was approved by the Bioethics Committee of the University of Milan (approval code: 1052019) and was conducted in accordance with the guidelines of the declaration of Helsinki.

#### *2.2. Procedures*

All measurements were performed in resting and fasting conditions at the facilities of the teams in the last week of August at 8.30 a.m. Generally, this period corresponds to the end of the preparation for the competitive season; therefore, it coincides with the start of the season. Body height was recorded to the nearest 0.1 cm with a stadiometer (SECA® 240, Hamburg, Germany) and weight was measured to the nearest 0.1 Kg with a calibrated weight scales (SECA® 877, Hamburg, Germany).

Whole-body impedance was obtained using a bioimpedance analyzer (BIA 101 Anniversary Edition, Akern, Florence, Italy). The device emits an alternating sinusoidal electric current of 400 mA at an operating single frequency of 50 kHz (±0.1%). Subjects were positioned with a leg opening of 45◦ with respect to the midline of the body, and with the upper limbs positioned 30◦ away from the trunk. The bioelectric phase angle (PhA) was calculated as the arctangent of Xc/R × 180/π. BIVA was carried out using the classic methods, e.g., normalizing R (ohm) and Xc (ohm) for height in meters [6,8].

#### *2.3. Statistical Analyses*

The two-sample Hotelling T<sup>2</sup> test was used to compare the differences in the mean impedance vectors between the bioimpedance data measured on the athletes of this study and the reference bioelectric values proposed by Levi Micheli et al. [14] The 50, 75, and 95% tolerance ellipses were generated using the BIVA software [19]. Statistical significance was predetermined as *p* < 0.05. Data were analyzed with IBM SPSS Statistics, version 24.0 (IBM Corp., Armonk, NY, USA).

#### **3. Results**

Table 1 shows anthropometric and bioelectrical characteristics of the soccer player.


**Table 1.** Descriptive statistics for the soccer players according to playing position.

Abbreviations: BMI, body mass index; R/H, resistance standardized for height; Xc/H, reactance standardized for height; PhA, phase angle.

The results of the two-sample Hotelling's T<sup>2</sup> test showed separate 95% confidence ellipses indicating a significant difference (T<sup>2</sup> <sup>=</sup> 15.3, F <sup>=</sup> 7.6, *<sup>p</sup>* <sup>≤</sup> 0.001, Mahalanobis D <sup>=</sup> 0.45) between the BIVA patters measured in this study and those proposed by Levi Micheli et al. [14] as a reference for the male elite soccer players population (Figure 1a).

The new reference ellipses and the single bioimpedance vectors measured in the soccer players at the start of the season are shown in Figure 1b.

**Figure 1.** Mean impedance vectors with the 95% confidence ellipses for the soccer players measured at the start and at the first half of the competitive season [10] (Panel **A**). Scattergrams of the individual impedance vectors plotted on the new tolerance ellipses (Panel **B**).

#### **4. Discussion**

The aim of this study was to show the importance of evaluating bioelectric properties using BIVA references that are suitable for the competitive period in which the assessment is carried out. The results of this study, which provide bioelectrical impedance data for 131 elite players, showed how the tolerance ellipses created on the basis of measurements during the different phases of the competition differ significantly for elite soccer players.

The bioimpedance data reported in the present study are comparable to previous values reported during the start-of-the season period in elite soccer players [20–22]. In comparison with the elite Italian male soccer population investigated by Levi Micheli et al. [14], the elite soccer players measured in this study showed a significant vector shift to the left on the minor axis of the tolerance ellipses. This could indicate a greater cell mass, which is a consequence of the effects sought in the preparation phase (training and controlled diet) typically, designed to increase endurance level and increase strength [23]. In fact, in a previous study, Mascherini et al. [17] suggested that the shortening of the vector was associated with changes in hydration status and increases in body cell mass. In this study, the preparation phase could have increased the intracellular/extracellular water (ICW/ECW) ratio as can be seen from a higher PhA than that measured by Levi Micheli et al. [14] (8.0 ± 0.5◦ vs. 7.7 ± 0.6◦ ). Indeed, PhA is positively associated with the ICW/ECW ratio in athletes [18,24]. Bioelectric data reflect the content of body fluids and the cellular health of the athlete and during the season, which change in response to training load and physical condition over the season [25]. In fact, the new tolerance ellipses proposed in this study differ significantly from those generated in the study by Levi Micheli et al. [14], in which bioimpedance measurements were collected in the first half of the competitive period. Furthermore, Micheli Levi et al. [14] reported that BIA data was collected over 5 months, from October to January 2009–2010, a period of time that may have generated vector changes in the athletes themselves. Our hypothesis is that the increase in workload (training) and official matches from August to October (about 6–8 matches played) or from August to January (16–17 matches played) could lead to fatigue and increased muscle turnover, as well as reduced muscle function which could result in a shift to the right of the biompedance vector. In fact, during the season, the reduction of the PhA could indicate a decreased muscle function as shown by Norman et al. [26] However, since we have not performed any muscle function tests, this hypothesis will have to be further investigated in future studies.

The reference ellipses proposed in the literature for athletes are population-specific. In addition to those for soccer players proposed by Levi Micheli [14], Campa and Toselli [15] measured male volleyball players in the second half of the in-season and showed specific BIA vector distribution in elite players in comparison to lower levels athletes. Subsequently, Giorgi et al. [16] provided bioelectrical impedance data of male road cyclists of varying performance levels, measured at the time of their optimal performance level and identified the 50, 75 and 95% tolerance ellipses for the road cyclists population, as well as for the high-performance road cyclists. In addition to these, there are also ellipses for healthy athletes built on more than 1000 male and 440 female athletes during the off-season period, therefore suitable for evaluating BIVA in the first phase of the competitive season [12].

The authors are also aware of the limitations of the study. Firstly, the subjects come from the same territory; therefore, the results obtained are not generalizable to all the soccer players around the world: a larger sample size is required even in different countries. The second is that no division by ethnicity of the players has been made in order to obtain a sample as large as possible: currently an international data collection is active that will allow us to investigate both these two limitations.

A strength of this study is in the specific time period in which the measurements were collected, not only in regard to the competitive level of the athletes, but above all for the time span in which BIA assessments were performed. In fact, BIA measurements were collected within a week, just before the start of the season, a period of time too short to generate vector changes between the players.

For the reasons mentioned above, future studies conducted with the aim of providing BIVA references for athletes should carry out the measurements according to the competitive phase for which they want to provide the new references. This is very significative given that vector changes occur during the different phases of the season in athletes, and bioelectrical values must be as informative and specific as possible, in order to obtain accurate monitoring of the body composition and physical condition of the athlete. This study demonstrates the importance of evaluating athletes on period-specific BIVA references, providing new tolerance ellipses for assessing body composition and cellular health before the start of the competitive season in elite soccer players.

#### **5. Conclusions**

Through BIVA, it is possible to evaluate body composition and the state of physical condition in the different phases of the competition in elite soccer players. This study provides specific BIVA references for the start of the season period, through which the physical condition achieved after the preparation micro cycle in soccer can be assessed.

**Author Contributions:** Conceptualization, F.C. and G.M.; methodology, T.B.; software, F.C.; validation, F.M.I., G.P.; formal analysis, F.C.; investigation, T.B.; data curation, F.G.; writing—original draft preparation, F.C. and G.M.; supervision, A.T. and M.V.; project administration, G.A. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors are grateful to all the soccer players who took part in this study.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Body Fat Assessment in International Elite Soccer Referees**

#### **Cristian Petri <sup>1</sup> , Francesco Campa <sup>2</sup> , Vitor Hugo Teixeira 3,4, Pascal Izzicupo <sup>5</sup> , Giorgio Galanti <sup>1</sup> , Angelo Pizzi <sup>6</sup> , Georgian Badicu 7,\* and Gabriele Mascherini <sup>1</sup>**


Received: 14 April 2020; Accepted: 5 June 2020; Published: 6 June 2020

**Abstract:** Soccer referees are a specific group in the sports population that are receiving increasing attention from sports scientists. A lower fat mass percentage (FM%) is a useful parameter to monitor fitness status and aerobic performance, while being able to evaluate it with a simple and quick field-based method can allow a regular assessment. The aim of this study was to provide a specific profile for referees based on morphological and body composition features while comparing the accuracy of different skinfold-based equations in estimating FM% in a cohort of soccer referees. Forty-three elite international soccer referees (age 38.8±3.6 years), who participated in the 2018 Russian World Cup, underwent body composition assessments with skinfold thickness and dual-energy X-ray absorptiometry (DXA). Six equations used to derive FM% from skinfold thickness were compared with DXA measurements. The percentage of body fat estimated using DXA was 18.2 ± 4.1%, whereas skinfold-based FM% assessed from the six formulas ranged between 11.0% ± 1.7% to 15.6% ± 2.4%. Among the six equations considered, the Faulkner's formula showed the highest correlation with FM% estimated by DXA (r = 0.77; R<sup>2</sup> = 0.59 *p* < 0.001). Additionally, a new skinfold-based equation was developed: FM% = 8.386 + (0.478 × iliac crest skinfold) + (0.395 × abdominal skinfold, r = 0.78; R <sup>2</sup> = 0.61; standard error of the estimate (SEE) = 2.62 %; *p* < 0.001). Due to these findings, national and international federations will now be able to perform regular body composition assessments using skinfold measurements.

**Keywords:** anthropometry; body composition; DXA; equation; fat mass; soccer; somatotype; skinfolds

#### **1. Introduction**

No matter the competition level, there is no official soccer game without the presence of the 23rd key element: the referee. There are more than 840,000 registered referees who arbitrate soccer games each week verifying and enforcing the rules of the game [1]. However, scientific literature on referees in relation to physiological demands, body composition, and nutrition-related aspects [2–4] is limited when compared to what is available on players.

Not only do elite-class soccer referees need to perform their best in perceptual-cognitive abilities and decision-making tasks [5], but they must also achieve an elevated aerobic performance similar to a midfield soccer player [6]. Although there are significant differences between games, referees cover on average a distance between 9 and 13 km per game, depending on the level of competition [3,7–9]. Due to the high-intensity matches in recent years [10] and as a consequence of the increase in physical demand there is a trend towards a decrease in body mass index (BMI) and body fat levels in elite referees [11]. In fact, recent studies on elite referees (FIFA) show lower levels of BMI and body fat than previously reported in Premier League referees [12,13].

In soccer players, body fat is generally estimated by skinfold thickness assessment, bioelectrical impedance analysis, and dual-energy X-ray absorptiometry (DXA). DXA is widely accepted as a practical method for assessing fat mass in athletes [14–16] and is becoming more popular in elite sports as an analysis used by both practitioners and researchers [17]. Anthropometric assessment is another popular method used to predict body fat in athletes [18–20]. The measurement of skinfolds is widely adopted to monitor changes in body composition due to training and/or dietary interventions. Skinfold thickness measurements have long been utilized to predict body fat and over the years multiple equations have been developed for this purpose [21–26]. However, to the best of the authors' knowledge, there are no studies comparing results obtained by skinfold-based measurements and DXA in soccer referees.

This study was designed to present body composition features of soccer referees called for the Russian World Cup in 2018 with two main aims. The first purpose was to compare different equations to estimate fat mass from skinfold assessment with data derived from DXA. The second aim was to develop a prediction equation for estimating the percentage of fat mass (FM%) based on anthropometric variables specific for this cohort.

#### **2. Materials and Methods**

#### *2.1. Study Population*

Using a cross-sectional design, 43 elite international soccer referees (age 38.8 ± 3.6 years) from 6 confederations were enrolled in the study during a seminar held in the Federal Technical Center of Coverciano (Italy) in April 2018, during the competitive season. They were classified as elite because they were either registered at the maximum level in the athletic federation or have received financial support for their dedication to training and games. The study was designed and conducted in accordance with the Helsinki Declaration. The ethics committee of the Italian football association approved this study and all the participants signed written informed consent prior to their inclusion in the study (approval code: 03032018).

#### *2.2. Procedures*

Participants underwent body composition assessments early in the morning, in an overnight-fasted state, and at least 12 h postexercise, with no long trips during the previous day. Further, the consumption of alcohol and stimulant beverages were not allowed for at least 15 h prior to testing.

The methodology used for the assessment of body composition was in accordance with our previous studies [26,27], using the integration of anthropometry, skinfold thickness, and DXA. Anthropometric measurements were taken following the protocol of The International Society for the Advancement of Kinanthropometry (ISAK) [28] by the same researcher (an ISAK level anthropometrist), whose technical error was 5% and 1.5% for skinfolds and all other measurements, respectively. Height (m) and body weight (kg) were measured to the nearest 0.1 cm and 0.01 kg, respectively, using a high-precision mechanical scale (SECA, Basel, Switzerland). BMI was calculated using the formula body mass/height<sup>2</sup> (kg/m<sup>2</sup> ). Biceps girth, waist girth, and hip girth (cm) were measured with a narrow, metallic, and inextensible measuring tape (Lufkin® model W606PM, London, UK; precision = 1 mm). Skinfolds were measured with a skinfold caliper (Holtain Ltd., Crymych, UK; precision = 0.2 mm) at nine anatomical sites (triceps, subscapular, biceps, iliac crest, supraspinal, pectoral, abdominal, thigh, and calf). Humerus and femur breadths were measured to the nearest 0.1 cm with a sliding caliper (GMP, Zürich, Switzerland). Somatotype was calculated according to the Heath-Carter method [29].

Body density was calculated from the Siri equation [30] adapted for age [31], which was then used to estimate FM%. The following six skinfold-based equations were used to estimate FM%:


Additionally, the average FM% measured by all of these equations was considered. Fat-Free Mass (FFM) was calculated subtracting fat mass from body weight.

A DXA scanner (Hologic QDR Series, Delphi A model, Bedford, MA, USA) with Hologic APEX 13.3:3 software version, was used to estimate FM%. The instrument was calibrated with phantoms as per the manufacturer's guidelines each day prior to measurements. Participants assumed a stationary supine position on the scanning table. All scanning and analyses were performed by the same technician to ensure consistency and in accordance with standardized testing protocols recognized as best practice [17,32,33].

#### *2.3. Statistical Analysis*

Data were expressed as mean ± standard deviations (SD) and normality of distribution of the data was verified by the Kolmogorov–Smirnov test. Comparisons between FM% calculated with the different formulas and those measured by DXA were made using linear regression analysis, as well as between the sum of skinfold measurements with FM% obtained by DXA.

Given the fact that different ethnic groups participated, the effect of ethnicity on FM% was tested using the Kruskal–Wallis test. No interactions were found between ethnicity and other independent variables; therefore, we used the whole sample in the model development. The ability of the following variables (age, stature, weight, and skinfolds) in predicting FM% in the international soccer referees was assessed using stepwise regression analysis. During model development, normality of residuals and homogeneity of variance were tested. Significance at *p* ≤ 0.05 was established as the criterion for inclusion of a predictor whereas removal criteria were set at *p* ≤ 0.1. If more than one variable remained in the model, and to assess multicollinearity, a variance inflation factor (threshold as 5) was calculated for each independent variable. The data were analyzed using the statistical package IBM SPSS Statistics (version 13.0) for Windows. (SPSS Inc., Chicago, IL, USA).

#### **3. Results**

General and anthropometric characteristics and descriptive values of FM% estimated from DXA and skinfold-based equations are shown in Table 1. The referees showed an average balanced mesomorph somatotype, characterized by a dominant mesomorph component and similar values between endomorph and ectomorph components (no more than a difference of 0.5 units, Figure 1).

Correlation coefficients and level of significant differences between FM% with DXA and other practical estimates in the soccer referees are shown in Table 2. Given that no difference between ethnic groups in FM% was found (*p* = 0.241), all the values were presented together.


**Table 1.** General and anthropometric characteristics of the international-level elite referees. **Table 1.** General and anthropometric characteristics of the international-level elite referees.

Abbreviations: Σ2sk = anterior thigh, medial calf; Σ4sk-a = biceps, triceps, subscapular, iliac crest; Σ4sk-b = triceps, abdominal; anterior thigh, medial calf; Σ5sk = biceps, triceps, subscapular, iliac crest, anterior thigh; Σ6sk = triceps, subscapular, iliac crest, abdominal, anterior thigh, medial calf; Σ7sk = triceps, subscapular, iliac crest, supraspinal, abdominal, anterior thigh, medial calf; Σ9sk = biceps, triceps, subscapular, iliac crest, supraspinal, pectoral, abdominal, anterior thigh, medial calf. Σ4sk-b = triceps, abdominal; anterior thigh, medial calf; Σ5sk = biceps, triceps, subscapular, iliac crest, anterior thigh; Σ6sk= triceps, subscapular, iliac crest, abdominal, anterior thigh, medial calf; Σ7sk= triceps, subscapular, iliac crest, supraspinal, abdominal, anterior thigh, medial calf; Σ9sk = biceps, triceps, subscapular, iliac crest, supraspinal, pectoral, abdominal, anterior thigh, medial calf.

**Figure 1.** Representation of the somatotype of the International elite male soccer referees. **Figure 1.** Representation of the somatotype of the International elite male soccer referees.


**Table 2.** Correlation coefficients and level of significance between fat mass percentage (FM%) estimated by dual-energy X-ray absorptiometry (DXA) with the sum of skinfold measurements, and FM% obtained from skinfold-based equations.

All FM% values obtained by the skinfold-based equations showed large to very large positive correlations (r from 0.60 to 0.78) to those measured by DXA. The FM% estimated from all of the equations showed significant differences (*p* < 0.001) in comparison to the DXA results. The sum of skinfold measurements showed moderate to very large positive correlations with FM% obtained by DXA, except for the sum of two skinfold measurements. Relationships between DXA-derived FM% and skinfold thicknesses of different anatomical sites are shown in Table 3. The vast majority of skinfold measurements cited showed moderate to very large positive relationships with FM% (r from 0.57 to 0.71), except biceps, pectoral, anterior thigh, and medial calf.

**Table 3.** Relationships between DXA-derived FM% and different skinfolds measured in the internationallevel elite referees.


Table 4 shows the skinfold-based model for FM% generated for the international soccer referees. Only variables contributing as significant predictors using backward stepwise approach were used in the model. The final prediction model included: FM% = 8.386 + (0.478 × iliac crest skinfold) + (0.395 × abdominal skinfold, r = 0.781; R<sup>2</sup> = 0.610; SEE = 2.62 %; *p* < 0.001, Table 4; Figure 2).


**Table 4.** Developed models for FM% prediction. **Variable Coefficient R2 SEE (%)** 

*J. Funct. Morphol. Kinesiol.* **2020**, *5*, x FOR PEER REVIEW 6 of 10

Abbreviations: R<sup>2</sup> , coefficient of determination; SEE, standard error of the estimate. Abbreviations: R2, coefficient of determination; SEE, standard error of the estimate.

**Figure 2.** The scatter plot illustrates the results of the regression analysis for FM% obtained by the reference method (DXA) and predicted by the equation developed in the sample of international **Figure 2.** The scatter plot illustrates the results of the regression analysis for FM% obtained by the reference method (DXA) and predicted by the equation developed in the sample of international soccer referees.

#### soccer referees. **4. Discussion**

**4. Discussion**  This study compared six skinfold-based equations showing the different results in FM% estimation in international elite soccer referees, using DXA as a reference method. Additionally, our study is the first to have provided a specific equation for this particular sample group, as well as descriptive body composition parameters. Our results highlight that the sum of skinfold thickness measurements taken in seven (triceps, subscapular, iliac crest, supraspinal, abdominal, anterior thigh, and medial calf) or in nine (biceps, triceps, subscapular, iliac crest, supraspinal, pectoral, abdominal, anterior thigh, and medial calf) sites show a high association with the FM% estimated with DXA. Secondly, among the equations considered, that of Faulkner [22] showed the best sensitivity in assessing FM% in the international elite soccer referees. Lastly, a new equation based on the This study compared six skinfold-based equations showing the different results in FM% estimation in international elite soccer referees, using DXA as a reference method. Additionally, our study is the first to have provided a specific equation for this particular sample group, as well as descriptive body composition parameters. Our results highlight that the sum of skinfold thickness measurements taken in seven (triceps, subscapular, iliac crest, supraspinal, abdominal, anterior thigh, and medial calf) or in nine (biceps, triceps, subscapular, iliac crest, supraspinal, pectoral, abdominal, anterior thigh, and medial calf) sites show a high association with the FM% estimated with DXA. Secondly, among the equations considered, that of Faulkner [22] showed the best sensitivity in assessing FM% in the international elite soccer referees. Lastly, a new equation based on the anthropometric measurements taken on the sample group was proposed.

anthropometric measurements taken on the sample group was proposed. While excessive FM% may affect performance, body composition is an aspect of considerable interest to scientists, athletes, and coaches [34]. Typical FM% values (ranging from 5% to 19%) reported in male athletes depends on the sport, playing position, and methodology used for the assessment [35–37]. In particular, male soccer players show a percentage of fat mass ranging between While excessive FM% may affect performance, body composition is an aspect of considerable interest to scientists, athletes, and coaches [34]. Typical FM% values (ranging from 5% to 19%) reported in male athletes depends on the sport, playing position, and methodology used for the assessment [35–37]. In particular, male soccer players show a percentage of fat mass ranging between 11.7–13.7% [20,38]. Furthermore, the tested referees in this study showed a balanced mesomorph somatotype.

11.7–13.7% [20,38]. Furthermore, the tested referees in this study showed a balanced mesomorph somatotype. Similarly, high-level soccer players are characterized by a balanced mesomorph morphology but their somatotype can also be endomorphic mesomorph, ectomorphic mesomorph, mesomorph

ectomoprh, and mesomorphic ectomorph [39]; in all these cases, the dominant component is the mesomorphy, but there is a different balance between endomorphy and ectomorphy, probably due to the different roles of the game.

Our results showed substantial discrepancies in FM% prediction depending on the method plied. Therefore, care must be taken when feedback on FM% is provided to soccer referees since values are likely to be method dependent. With the exception of FM% obtained using Suarez's equation [26] or estimated by DXA, most of the data found was within the range described in the literature. As opposed to Faulkner's equation [22], the new formula suggested appears to be a simpler and faster alternative as it is specific to soccer referees. Furthermore, the use of only two skinfold sites, provides an advantage in the field-based assessment of FM%, representing a more efficient use of time.

Contrary to previous studies of elite soccer players [25,26], data collected in the present study showed that measurements of the lower body skinfolds are not accurate when predicting FM%. In this study, thigh skinfold thickness was not entered in the developed model representing strength in the new formula because it has been acknowledged that the anterior thigh skinfold is one of the least accurate sites to measure [23,24]. Furthermore, our results showed that different sums of skinfolds measured on the referees showed moderate to large correlations with DXA data, used as a criterion method. The results of the present study showed very large correlations between Σ4SKF-a, Σ4SKF-b, Σ6SKF, Σ7SKF, and Σ9SKF with FM% estimated by DXA, similar to the associations with the data found in the literature for elite soccer players [40,41]. Thus, considering the substantial differences observed between the different equations and their similar and/or lower correlations with the DXA-derived FM%, even the sum of skinfold measurements appears to be a good alternative approach in obtaining information on body fat distribution in elite soccer referees.

Some of the strengths in this study include the selection of international-level male soccer referees, all with experience in international matches. Skinfold measurement is a practical, low cost, and easily accessible alternative to more complex and expensive methods such as DXA. Consequently, the present study provides a noninvasive, cost-free, and fast tool to accurately estimate FM% in the investigated cohort. Moreover, the DXA method might not be feasible or useful when financial resources are limited or when a whole group is repeatedly measured, because up to 10 min for each person is required.

However, the use of DXA as a reference method poses some limitations in the development and comparison of new equations for assessing fat mass. Most of the equations in the literature are based on the reference of hydrostatic weighing. This could predispose our equation to an overestimation of body fat. Additional limitations regarding sample size should be addressed. The sample consisted exclusively of male subjects thus, further studies involving elite female referees are needed. Furthermore, the lack of a validation sample did not allow the test of the performance of the new equation at a group level (e.g., analysis of the regression coefficients, line of identity, R<sup>2</sup> , RMSE) and at the individual level using the Bland–Altman analysis.

The practical application of this equation facilitates an increasingly accurate evaluation of the athlete. The soccer referee can be considered in all respects as an athlete, even sports research has deepened this particular population. First, the components of the physical load required during a competition [3,7,9] have been evaluated, more recently the nutritional aspects have had an increase in interest [4,12,42,43]. In this context of energy balance, the development of a new equation for the evaluation of the fat mass with a specific reference to the analyzed population allows a more accurate, reliable, and repeatable evaluation during the competitive season of the soccer referee through a field methodology.

#### **5. Conclusions**

All the equations investigated (Eston et al. [24], Yuhasz [21], Reilly et al. [25], Suarez et al. [26], and Durnin and Womersley [23]) showed positive correlations in comparison with DXA data. In particular, the equations developed by Faulkner [22] showed the best sensitivity in assessing FM% compared to DXA. Additionally, the sum of seven skinfolds, which included triceps, subscapular, iliac crest, supraspinal, abdominal, anterior thigh, and medial calf measurements showed a high correlation with FM% measured by DXA, representing an alternative approach in body composition assessment. Finally, this study provides a new formula for FM% estimation in international-level male soccer referees [FM% = 8.386 + (0.478 × iliac crest skinfold) + (0.395 × abdominal skinfold)].

**Author Contributions:** Conceptualization, C.P. and G.M. methodology, C.P., F.C., and P.I.; formal analysis, C.P., F.C., and P.I.; investigation, A.P., G.G., and C.P.; data curation, C.P., F.C., and P.I.; writing—original draft preparation, V.H.T., C.P., F.C., and G.B.; visualization, G.B., V.H.T., and G.M.; supervision, F.C., G.B., and G.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors gratefully acknowledge FIFA for their support of this research during the experiments.

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Potential Use of Wearable Sensors to Assess Cumulative Kidney Trauma in Endurance O**ff**-Road Running**

**Daniel Rojas-Valverde 1,2,\* , Rafael Timón 2 , Braulio Sánchez-Ureña <sup>3</sup> , José Pino-Ortega <sup>4</sup> , Ismael Martínez-Guardado <sup>2</sup> and Guillermo Olcina 2,\***


Received: 6 November 2020; Accepted: 11 December 2020; Published: 14 December 2020

**Abstract:** (1) Background: This study aimed to explore wearable sensors0 potential use to assess cumulative mechanical kidney trauma during endurance off-road running. (2) Methods: 18 participants (38.78 ± 10.38 years, 73.24 ± 12.6 kg, 172.17 ± 9.48 cm) ran 36 k off-road race wearing a Magnetic, Angular Rate and Gravity (MARG) sensor attached to their lower back. Impacts in g forces were recorded throughout the race using the MARG sensor. Two blood samples were collected immediately pre- and post-race: serum creatinine (sCr) and albumin (sALB). (3) Results: Sixteen impact variables were grouped using principal component analysis in four different principal components (PC) that explained 90% of the total variance. The 4th PC predicted 24% of the percentage of change (∆%) of sCr and the 3rd PC predicted the ∆% of sALB by 23%. There were pre- and post-race large changes in sCr and sALB (*p* ≤ 0.01) and 33% of participants met acute kidney injury diagnosis criteria. (4) Conclusions: The data related to impacts could better explain the cumulative mechanical kidney trauma during mountain running, opening a new range of possibilities using technology to better understand how the number and magnitude of the g-forces involved in off-road running could potentially affect kidney function.

**Keywords:** renal health; wearable devices; technology; acute kidney injury; inertial measurement units (IMU)

#### **1. Introduction**

Acute kidney injury (AKI) is a relatively uncommon condition in sports. This condition has been reported in prolonged and repetitive strenuous exercises [1]. It is understood as a transitional decrease in renal function, expressed by a reduction in glomerular filtration rate, increase in serum creatinine (sCr) and albumin (sALB), and alterations of other novel AKI-related urine and blood biomarkers during a relatively short period (1–3 days) [2].

The evidence of AKI cases in both contact and non-contact sports has been increased, but with clear different etiological backgrounds [3–5]. In contact sports like football, boxing, and rugby, AKI cases have been related to kidney contusion or trauma (grade I in American Association for Surgery of

Trauma classification) during tackles, punches, or other high-intensity actions with direct impact to the body [6,7]. On the other hand, in non-contact sports (e.g., endurance running and cycling), AKI has been related to the high number of muscle eccentric-concentric contractions leading to muscle damage [8,9].

In endurance running and mainly off-road running [8,10], some evidence has been published regarding the impact of external workload (e.g., impacts) as an additional factor that may contribute to AKI incidence, next to other known factors like dehydration, heat strain, and high metabolic activity [11]. Within this multifactorial etiology, high physical internal and external load seems to be a discernible contributing factor to the transitory decrease in renal function in endurance runners [12]. It could be due to muscle damage in response to high eccentric actions and its effect on inflammatory and hemodynamic responses that may affect the kidney [13]. New evidence has also highlighted the cumulative mechanical trauma that affects the kidney during off-road running as a potential cause of AKI [9]. Although kidneys are very well protected structures, there is relative mobility that could lead to injury even when no direct trauma occurred [14], for example, during downhill running or change of directions during training or competition.

Monitoring physical load is critical in endurance sports, such as off-road running, due to the high number of actions involved [15]. This is why non-invasive tools as wearable sensors could be an accessible option to assess potential cumulative mechanical kidney trauma, indirectly analyzing the mobility of anatomical structures near the kidneys, such as the lower back. These wearable sensors are used to monitor physical load during exercise in different parts of the body, such as the wrist, waist, and trunk [16–18]. It has also been determined that there is a relationship between the increase in serum blood factors related to kidney damage and the quantified load in the lower back [9]. Therefore, this study aimed to explore the potential use of wearable Magnetic, Angular Rate and Gravity (MARG) sensors to assess cumulative mechanical kidney trauma during off-road running.

#### **2. Materials and Methods**

#### *2.1. Design*

Participants were asked to perform three loops of a 12 km (+ascend = 600 m) circuit (total distance = 36 km and total +ascend = 1800 m), under 25◦ Celsius of temperature, and 80% of humidity (Wet Bulb globe Temperature, 3M, USA). Runners wore a MARG sensor attached to the lower back during the race, and variables of time-related impacts were extracted. Two blood samples were collected pre- and post-race to assess serum creatinine (sCr) and albumin (sALB). An analysis was made to explore a model based on impact variables that explained sCr and sALB increases between pre- and post-race.

#### *2.2. Participants*

Eighteen experienced mountain runners participated in this study (age 38.78 ± 10.38 years, weight 73.24 ± 12.6 kg, height 172.17 ± 9.48 cm). They had 4.78 ± 2.42 years of experience competing in ultra-endurance events. Participant0 s mean finish time was 4.2 ± 0.21 h. No neuromuscular, metabolic, or structural injuries were reported at least six months before the study. The participants were asked to avoid intense endurance exercise at least a week before the event.

All participants were notified of the study0 s aim, protocol details and the potential risks and rights during their participation. The study´s protocol followed all biomedical guidelines based on the Declaration of Helsinki (2013) and it was reviewed and approved by the Institutional Review Boards of Universidad Nacional (Reg. Code 2019-P005) and Universidad de Extremadura (Reg. Code 139/2020).

#### *2.3. Materials and Procedures*

Sixteen different time-related impacts (*n*/min, g forces) variables were assessed using a Magnetic, Angular Rate and Gravity (MARG) sensor (WIMU PROTM, RealTrack Systems, Almería, Spain). The devices were attached to the lower back (~L1–L3) [9] of each participant with a special spandex

dark belt adjusted with elastic straps to avoid device´s unwanted vibrations or movements (see Figure 1). The MARG´s integrate four 3-axis microelectromechanical systems accelerometers (2x ± 16 g, 1x ± 32 g, and 1x ± 400 g), gyroscope, and magnetometer. All MARG´s calibration and setting were developed following published guidelines [19,20], its reliability for neuromuscular running physical load assessment has been proven [21] and its reliability has been tested in different body parts [22]. The variables extracted were total impacts per min (ImpactsTotal/min) and fifteen progressively scaled categories of g-force magnitude, each 1 g wide (Impacts1–15 g/min). Spain). The devices were attached to the lower back (~L1–L3) [9] of each participant with a special spandex dark belt adjusted with elastic straps to avoid device´s unwanted vibrations or movements (see Figure 1). The MARG´s integrate four 3-axis microelectromechanical systems accelerometers (2x ± 16 g, 1x ± 32 g, and 1x ± 400 g), gyroscope, and magnetometer. All MARG´s calibration and setting were developed following published guidelines [19,20], its reliability for neuromuscular running physical load assessment has been proven [21] and its reliability has been tested in different body parts [22]. The variables extracted were total impacts per min (ImpactsTotal/min) and fifteen progressively scaled categories of g-force magnitude, each 1 g wide (Impacts1–15 <sup>g</sup>/min).

**Figure 1.** Inertial measurement unit attachment at runner´s lower back (L1–L3). **Figure 1.** Inertial measurement unit attachment at runner´s lower back (L1–L3).

Blood serum samples were collected using 5 mL blood spray-coated silica tubes (BD Vacutainer®, Franklin Lakes, NJ, USA). After centrifugation (10 min at 2000 g), samples were stored at −20 °C. After 24 h, the samples were processed by the photometry method using an automatic biochemical analyzer (BS-200E, Mindray, China). The variable analyzed was serum creatinine (sCr, mg/dL) and serum albumin (sALB, IU/L). Acute kidney injury (sCr baseline in mg/dL \*1.5) was considered following established diagnosis criteria [23]. Two groups were made based on AKI diagnosis as follows: those participants that met AKI diagnosis (AKI) and the ones that did not (No-AKI), in order to explore differences in the number of impacts reported. Blood serum samples were collected using 5 mL blood spray-coated silica tubes (BD Vacutainer®, Franklin Lakes, NJ, USA). After centrifugation (10 min at 2000 *g*), samples were stored at −20 ◦C. After 24 h, the samples were processed by the photometry method using an automatic biochemical analyzer (BS-200E, Mindray, China). The variable analyzed was serum creatinine (sCr, mg/dL) and serum albumin (sALB, IU/L). Acute kidney injury (sCr baseline in mg/dL \*1.5) was considered following established diagnosis criteria [23]. Two groups were made based on AKI diagnosis as follows: those participants that met AKI diagnosis (AKI) and the ones that did not (No-AKI), in order to explore differences in the number of impacts reported.

Urine specific gravity (USG) was assessed as a hydration status marker. USG was confirmed and double-checked with a digital valid [24] handheld refractometer (Palm AbbeTM, Misco, Solon, OH, USA). It was classified following the hydration status ranges: well-hydrated <1.01, minimal dehydration 1.01–1.02, significant dehydration 1.02–1.03, and severe dehydration >1.03 [25]. The refractometer was cleaned with distilled water and calibrated previously. There were no reported urination problems or difficulties neither before nor after the race. Urine specific gravity (USG) was assessed as a hydration status marker. USG was confirmed and double-checked with a digital valid [24] handheld refractometer (Palm AbbeTM, Misco, Solon, OH, USA). It was classified following the hydration status ranges: well-hydrated<1.01, minimal dehydration 1.01–1.02, significant dehydration 1.02–1.03, and severe dehydration >1.03 [25]. The refractometer was cleaned with distilled water and calibrated previously. There were no reported urination problems or difficulties neither before nor after the race.

#### *2.4. Statistical Analysis 2.4. Statistical Analysis*

All sixteen impact variables were grouped using a Principal Component Analysis (PCA) following previous studies guidelines [9,26]. PCA was suitable, according to Kaiser-Meyer-Olkin (*KMO* = 0.63) values and the Barleth Sphericity test (*p* < 0.01). Eigenvalues (EV) > 1 were considered for the extraction of each Principal Component (PC). A VariMax-orthogonal rotation method was used to identify the high correlation of components. A threshold of 0.6 was set to retain loadings. The highest loading was used when a cross-loading was found between PCs. PCA procedure followed standard quality criteria [27], meeting 21 out of 21 of the quality items. All sixteen impact variables were grouped using a Principal Component Analysis (PCA) following previous studies guidelines [9,26]. PCA was suitable, according to Kaiser-Meyer-Olkin (*KMO* = 0.63) values and the Barleth Sphericity test (*p* < 0.01). Eigenvalues (EV) > 1 were considered for the extraction of each Principal Component (PC). A VariMax-orthogonal rotation method was used to identify the high correlation of components. A threshold of 0.6 was set to retain loadings. The highest loading was used when a cross-loading was found between PCs. PCA procedure followed standard quality criteria [27], meeting 21 out of 21 of the quality items.

A paired t-test was used to explore sCr and sALB changes between pre- and post-race data and the Change delta´s percentage (∆%) was calculated as follows: ((sCr post-race–sCr pre-race)/sCr pre-A paired t-test was used to explore sCr and sALB changes between pre- and post-race data and the Change delta´s percentage (∆%) was calculated as follows: ((sCr post-race–sCr pre-race)/sCr

pre-race)\*100. An unpaired t-test was performed to explore potential differences in the number of impacts between those participants who met AKI diagnosis and those who did not. USG data were analyzed using a repeated measure *t*-test. The magnitude of the differences was calculated using Cohen´s *d*. *J. Funct. Morphol. Kinesiol.* **2020**, *5*, 93 4 of 9 between those participants who met AKI diagnosis and those who did not. USG data were analyzed using a repeated measure t-test. The magnitude of the differences was calculated using Cohen´s *d*. *J. Funct. Morphol. Kinesiol.* **2020**, *5*, 93 4 of 9 between those participants who met AKI diagnosis and those who did not. USG data were analyzed using a repeated measure t-test. The magnitude of the differences was calculated using Cohen´s *d*.

Finally, a stepwise regression model (*R 2* ) was applied to resulted factor scores obtained from impact´s PCA using the ∆% of sCr and sALB as the dependent variable. This statistical technique was applied to identify which impact´s PC could predict the ∆% of sCr, and ∆% of sALB. Finally, a stepwise regression model (*R2*) was applied to resulted factor scores obtained from impact´s PCA using the ∆% of sCr and sALB as the dependent variable. This statistical technique was applied to identify which impact´s PC could predict the ∆% of sCr, and ∆% of sALB. Finally, a stepwise regression model (*R<sup>2</sup>* ) was applied to resulted factor scores obtained from impact´s PCA using the ∆% of sCr and sALB as the dependent variable. This statistical technique was applied to identify which impact´s PC could predict the ∆% of sCr, and ∆% of sALB.

All variables were presented in mean ± standard deviation. Alpha was set at *p* < 0.05 and all analyses were made using the Statistical Package for Social Science (v.22, SPSS, Chicago, IL, USA). All variables were presented in mean ± standard deviation. Alpha was set at *p* < 0.05 and all analyses were made using the Statistical Package for Social Science (v.22, SPSS, Chicago, IL, USA). All variables were presented in mean ± standard deviation. Alpha was set at *p* < 0.05 and all analyses were made using the Statistical Package for Social Science (v.22, SPSS, Chicago, IL, USA).

#### **3. Results 3. Results 3. Results**

Participants experienced a total of 170.57 ± 34.42 impacts per minute. Figure 2 shows the mean number of impacts per minute in relation to the associated magnitude of g-force (see Figure 2). Participants experienced a total of 170.57 ± 34.42 impacts per minute. Figure 2 shows the mean number of impacts per minute in relation to the associated magnitude of g-force (see Figure 2). Participants experienced a total of 170.57 ± 34.42 impacts per minute. Figure 2 shows the mean number of impacts per minute in relation to the associated magnitude of g-force (see Figure 2).

**Figure 2.** Mean values of impacts per minute associated with 15 g-force categories during off-road mountain running**. Figure 2.** Mean values of impacts per minute associated with 15 g-force categories during off-road mountain running. **Figure 2.** Mean values of impacts per minute associated with 15 g-force categories during off-road mountain running**.**

All sixteen impact-related variables were grouped in four different PC´s, explaining the 90.39% of total impacts cumulative variance. The 1st PC explained the 50.5% (EV = 8.08) of total variance, 2nd PC the 17.58% (EV = 2.81), 3rd PC the 13.05% (EV = 2.09), and 4th PC the 9.27% (EV = 1.48). Grouped variables and loadings are presented in Figure 3. All sixteen impact-related variables were grouped in four different PC´s, explaining the 90.39% of total impacts cumulative variance. The 1st PC explained the 50.5% (EV = 8.08) of total variance, 2nd PC the 17.58% (EV = 2.81), 3rd PC the 13.05% (EV = 2.09), and 4th PC the 9.27% (EV = 1.48). Grouped variables and loadings are presented in Figure 3. All sixteen impact-related variables were grouped in four different PC´s, explaining the 90.39% of total impacts cumulative variance. The 1st PC explained the 50.5% (EV = 8.08) of total variance, 2nd PC the 17.58% (EV = 2.81), 3rd PC the 13.05% (EV = 2.09), and 4th PC the 9.27% (EV = 1.48). Grouped variables and loadings are presented in Figure 3.

In follow up to the abovementioned PCA results, those participants that met AKI diagnosis criteria (33.3% of participants) registered lower number of impacts in the 1–2 g category (*t* = −2.42, *p* **Figure 3.** Principal component analysis extracted variables and loadings. \* Loadings values. **Figure 3.** Principal component analysis extracted variables and loadings. \* Loadings values.

= 0.03, *d* = −1.45, large effect size) but higher number of impacts in the 14–15 g category (*t* = −3.1, *p* =

In follow up to the abovementioned PCA results, those participants that met AKI diagnosis

In follow up to the abovementioned PCA results, those participants that met AKI diagnosis criteria (33.3% of participants) registered lower number of impacts in the 1–2 g category (*t* = −2.42, *p* = 0.03, *d* = −1.45, large effect size) but higher number of impacts in the 14–15 g category (*t* = −3.1, *p* = 0.01, *d* = −1.58, large effect size) (see Figure 4.). No differences we found in the 5–6 g or 6–7 g categories. *J. Funct. Morphol. Kinesiol.* **2020**, *5*, 93 5 of 9 0.01, *d* = −1.58, large effect size) (see Figure 4.). No differences we found in the 5–6 g or 6–7 g categories.

**Figure 4.** Differences between runners showing signs of AKI (*n* = 6) and those showing no signs of AKI (*n* = 12) regarding impacts per minute, grouped in four impact g-force categories. \* The biggest difference between the AKI and no-AKI group is that the no-AKI group managed to run "smoother," keeping impacts in the lower impact load ranges, while avoiding higher impacts loads. **Figure 4.** Differences between runners showing signs of AKI (*n* = 6) and those showing no signs of AKI (*n* = 12) regarding impacts per minute, grouped in four impact g-force categories. \* The biggest difference between the AKI and no-AKI group is that the no-AKI group managed to run "smoother," keeping impacts in the lower impact load ranges, while avoiding higher impacts loads.

There were large statistical differences (*t* = −6.24, *p* < 0.01, *d* = −1.47, large effect size) between sCr pre-race (1.24 ± 0.28 mg/dL) and sCr post-race (1.74 ± 0.41 mg/dL), and large differences (*t* = −2.78, *p* = 0.01, *d* = −1.47, large effect size) in sALB pre-race (4.33 ± 1.29 IU/L) vs. post-race (5.01 ± 0.86 IU/L). The ∆% of sCr was predicted by the 4th PC in a 24% (*R<sup>2</sup>* = 0.24, *β* = 44.03, *p* < 0.01) and the ∆% of sALB by a 23% (*R <sup>2</sup>*= 0.23, *β* = 100.55, *p* = 0.04). Finally, USG as a hydration marker reported no differences between pre- and post-race measurements (1.01 ± 0.02 vs. 1.01 ± 0.01; t = 1.02, *p* = 0.07). There were large statistical differences (*t* = −6.24, *p* < 0.01, *d* = −1.47, large effect size) between sCr pre-race (1.24 ± 0.28 mg/dL) and sCr post-race (1.74 ± 0.41 mg/dL), and large differences (*t* = −2.78, *p* = 0.01, *d* = −1.47, large effect size) in sALB pre-race (4.33 ± 1.29 IU/L) vs. post-race (5.01 ± 0.86 IU/L). The ∆% of sCr was predicted by the 4th PC in a 24% (*R <sup>2</sup>* = 0.24, β = 44.03, *p* < 0.01) and the ∆% of sALB by a 23% (*R <sup>2</sup>* = 0.23, β = 100.55, *p* = 0.04). Finally, USG as a hydration marker reported no differences between pre- and post-race measurements (1.01 ± 0.02 vs. 1.01 ± 0.01; t = 1.02, *p* = 0.07).

#### **4. Discussion 4. Discussion**

Renal injury provoked by an indirect trauma has been reported in previous cases with no symptoms other than lumbar pain but with radiological findings such as subcapsular renal hematoma [14]. Some evidence suggests that urinary trauma could be present in non-contact sports such as off-road running [4,5,28]. It has been hypothesized that kidney mechanical trauma could mediate in the development of acute kidney injury after running [9]. This could be due to the kidneys′ relative mobility during some actions as a downhill run at high speeds, change of directions, falls, and other high g-forces that could affect kidney movements and shaking. This relationship needs to be explored in future studies. The results of this study suggest that the 4th PC and 3rd PC of impactrelated variables explained the ∆% of sCr and sALB between 23 to 24%. These findings indicate that the magnitude and number of impacts (g-forces) could have a potential role in the cumulative mechanical kidney trauma. Renal injury provoked by an indirect trauma has been reported in previous cases with no symptoms other than lumbar pain but with radiological findings such as subcapsular renal hematoma [14]. Some evidence suggests that urinary trauma could be present in non-contact sports such as off-road running [4,5,28]. It has been hypothesized that kidney mechanical trauma could mediate in the development of acute kidney injury after running [9]. This could be due to the kidneys0 relative mobility during some actions as a downhill run at high speeds, change of directions, falls, and other high g-forces that could affect kidney movements and shaking. This relationship needs to be explored in future studies. The results of this study suggest that the 4th PC and 3rd PC of impact-related variables explained the ∆% of sCr and sALB between 23 to 24%. These findings indicate that the magnitude and number of impacts (g-forces) could have a potential role in the cumulative mechanical kidney trauma.

Despite kidneys being well protected by abdominal and back muscles, ribs, fat, renal pedicle, and ureteropelvic junction and supporting Gerota fascia in the retroperitoneum, they are also susceptible to internal movements [14,28]. Repeated sudden accelerations and decelerations may lead to renal contusions caused by the collision of kidneys in its surrounding tissues and structures like spine and ribs. These actions could lead to renal vasculatures affections, nephron damage, consequent hematuria, and other blood markers findings [29–31]. These accelerations and deceleration could be assessed using the variable impacts as proposed in this study. The impacts Despite kidneys being well protected by abdominal and back muscles, ribs, fat, renal pedicle, and ureteropelvic junction and supporting Gerota fascia in the retroperitoneum, they are also susceptible to internal movements [14,28]. Repeated sudden accelerations and decelerations may lead to renal contusions caused by the collision of kidneys in its surrounding tissues and structures like spine and ribs. These actions could lead to renal vasculatures affections, nephron damage, consequent hematuria, and other blood markers findings [29–31]. These accelerations and deceleration could be assessed using the variable impacts as proposed in this study. The impacts between 5–7 g explained

It has been found that the ∆% of blood markers as serum creatine kinase and sCr could predict the

between 5–7 g explained the pre-post increase of sALB and the impacts of 1–2 g and 14–15 g explained

the pre-post increase of sALB and the impacts of 1–2 g and 14–15 g explained the rise in sCr. Based on the literature [9], these results may suggest that both the volume and intensity of the impacts involved during renal contusions play a special role in acute kidney injury. It has been found that the ∆% of blood markers as serum creatine kinase and sCr could predict the external workload of wearable devices placed in L1–L3 by 40% and 27%, respectively [9]. This evidence supports the idea of a new hypothesis of mechanical kidney injury during endurance off-road running based on L1–L3 external workload data [9].

The results of the present study showed that MARG sensors could be used to register the impacts and g-forces that affect the lower back, which is the kidney´s nearest external structure of the body. MARG sensors could register vertical, anterior-posterior, and mediolateral forces using the integration of accelerometer, gyroscope, and magnetometer data. The g-forces provoked by sudden accelerations and decelerations may affect the kidneys. The number and magnitude of these impacts could be monitored using MARGs attached to the kidney´s nearest external structure of the body, the lower back. Kidneys typically extend from T12 to L3 and weigh 135–150 g, so the MARG positioning should be at this level despite a slight position change due to the kidney0 s free mobility resulting from both body positions and respiration [32].

The link between the sensors0 external load and kidney trauma must be confirmed and discussed in future interventions. Previously, considering the cause of the increase in sCr may be indicative of kidney injury as well as massive muscle damage [33]. Although elevations in sCr in 33% of participants by itself should not be understood as kidney damage due to physical exercise, the rise in sALB could suggest transitory functional loss due to tubular or glomerular damage. In fact, there is evidence to suggest that proteins released into the bloodstream in high amounts (e.g., rhabdomyolysis) can overload kidney function, resulting in functional or subclinical damage reflected in an increase of sCr and sALB, respectively [34,35].

The cumulative small injuries during rough exercises as off-road mountain running might damage the kidney, resulting in AKI. Although there is no clear evidence that cumulative or subsequent AKI events contribute to future renal chronic conditions in athletes [1,36], there is enough evidence to suggest that athletes, coaches, and sports scientists should be concerned with controlling the kidney health of runners, monitoring those variables that can trigger AKI, and thus, preventing potential cases of this transitory kidney condition. Some preventive strategies have been proposed to endurance athletes such as optimal fluid and food intake, appropriate physical loading, rest, and acceptable recovery between efforts [3]. Monitoring physical load is essential and those external and internal variables that could affect not only kidney health but also general well-being should be assessed. Dehydration seems to be a factor that did not influence the AKI occurrence in this specific sample, as found in the results.

MARG units as wearable devices containing accelerometers, gyroscopes, and magnetometers allow trainers, athletes, and medical staff to monitor and control the physical external and internal loads involved during the off-road running. The information obtained would allow us to provide feedback on the kinematic behavior of the runner in an objective manner [37] and would facilitate the programming and prescription of training loads, preventing and mitigating the impact of AKI on the runner0 s health and performance.

These findings must be seen in light of some limitations. Considering that the cause of acute kidney injury is multifactorial, future studies may confirm the contribution of mechanical kidney damage in the increase of blood markers related to AKI. A global analysis of heat strain, metabolic responses, and dehydration should be made to explore the role of kidney mechanical trauma on AKI. The link of impacts assessed in the periphery of the body and mechanical trauma of hard connective tissues must be confirmed in future studies.

Also, it must be explored how much does prolonged massive g-forces impact runners during rough running (e.g., downhill, off-road, mountain) and produce kidney damage compared to similar heavy muscular exercise, but without the massive g-forces. Consequently, it should be explored if downhill running, sudden change of direction, falls, or other similar high magnitude actions produce greater damage than other running actions (e.g., uphill and flat running). Finally, there is a need to use other blood markers (e.g., Cystatin-C, NGAL, KIM-1) that allow researchers to differentiate AKI´s and extreme muscular exercise´s signs and symptoms. There is a need to review AKI0 s diagnosis criteria and its validity when applying it to sport sciences and medicine.

#### **5. Conclusions**

The results suggest that the magnitude and volume of running g-forces monitored with a MARG sensor attached to the lower back of off-road runners could predict the 24% change of serum creatinine and 23% change in serum albumin. These results must be confirmed in future research comparing similar heavy exercise with lower shock loads to the back and kidneys. Although these results may appear promising regarding the potential use of wearable devices to monitor cumulative mechanical kidney trauma in the future, greater understanding is required in the interaction of internal load (e.g., physiological responses) and external load (e.g., accelerations, impacts, decelerations) during prolonged exposure to vigorous repetitive exercise.

The results suggest that a decrease in the amount and magnitude of impacts throughout a session or between sessions can be a way to mitigate the possible collateral damage of acute kidney damage during off-road running. The foregoing considers, therefore, that the monitoring and control of training external and internal loads is essential for the prevention and recovery of AKI in off-road runners. In this sense, it is essential to provide constant feedback on running loads behavior and wearable MARG sensors could be used for these purposes.

**Author Contributions:** Conceptualization, D.R.-V.; methodology, D.R.-V., G.O., B.S.-U., and R.T.; software, D.R.-V., and J.P.-O.; validation, D.R.-V., G.O., B.S.-U., and R.T.; formal analysis, D.R.-V.; investigation, D.R.-V., B.S.-U., and J.P.-O.; resources, D.R.-V. and B.S.-U.; data curation, D.R.-V., J.P.-O., and I.M.-G.; writing—original draft preparation, D.R.-V., and I.M.-G.; writing—review and editing, D.R.-V., G.O., B.S.-U. and R.T.; supervision, D.R.-V., G.O., B.S.-U., and R.T.; project administration, D.R.-V., G.O., B.S.-U., and R.T.; funding acquisition, D.R.-V., G.O., B.S.-U., and R.T. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** Authors would like to express special thanks of gratitude to all participants and researchers of the CIDISAD and PROCESA laboratories for their administrative and technical support for the development of this study.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **E**ff**ect of Video Observation and Motor Imagery on Simple Reaction Time in Cadet Pilots**

**Felice Sirico 1,\* , Veronica Romano <sup>1</sup> , Anna Maria Sacco <sup>1</sup> , Immacolata Belviso <sup>1</sup> , Vittoria Didonna <sup>2</sup> , Daria Nurzynska <sup>1</sup> , Clotilde Castaldo <sup>1</sup> , Stefano Palermi <sup>1</sup> , Giuseppe Sannino <sup>1</sup> , Elisabetta Della Valle <sup>1</sup> , Stefania Montagnani <sup>1</sup> and Franca Di Meglio <sup>1</sup>**


Received: 10 October 2020; Accepted: 3 December 2020; Published: 5 December 2020

**Abstract:** Neuromotor training can improve motor performance in athletes and patients. However, few data are available about their effect on reaction time (RT). We investigated the influence of video observation/motor imagery (VO/MI) on simple RT to visual and auditory stimuli. The experimental group comprised 21 cadets who performed VO/MI training over 4 weeks. Nineteen cadets completed a sham intervention as control. The main outcome measure was RT to auditory and visual stimuli for the upper and lower limbs. The RT to auditory stimuli improved significantly post-intervention in both groups (control vs. experimental mean change for upper limbs: −40 ms vs. −40 ms, *p* = 0.0008; for lower limbs: −50 ms vs. −30 ms, *p* = 0.0174). A trend towards reduced RT to visual stimuli was observed (for upper limbs: −30 ms vs. −20 ms, *p* = 0.0876; for lower limbs: −30 ms vs. −20 ms, *p* = 0.0675). The interaction term was not significant. Only the specific VO/MI training produced a linear correlation between the improvement in the RT to auditory and visual stimuli for the upper (*r* = 0.703) and lower limbs (*r* = 0.473). In conclusion, VO/MI training does not improve RT when compared to control, but it may be useful in individuals who need to simultaneously develop a fast response to different types of stimuli.

**Keywords:** reaction time; pilots; motor imagery; video observation

#### **1. Introduction**

The time to respond to an external stimulus (reaction time) is the time lapse between the presentation of a stimulus and the onset of a voluntary response in a subject. The reaction time can be defined as the interval required to perceive the stimulus, process the information, fulfill an appropriate decision-making process, and initiate a motor task as a response [1]. Such a sequence of events adding to the reaction time is typical of real-life tasks and plays a critical role in many human activities related to sport or the professional performance of drivers, military personnel, security guards, or pilots. In neurophysiology, reaction time represents a valid indicator of an individual's sensorimotor coordination and performance [2]. Three different types of reaction time can be described, based on the relationship between stimulus and response: simple, recognition, and choice reaction time [3]. In simple reaction time studies, there is one stimulus (auditory, visual or tactile) and one response. In recognition reaction time studies, stimuli to be responded to are interspersed with

distracters that should not be followed with a response. In choice reaction time studies, several stimuli require different responses.

Reduction of reaction time is a desirable aim of intervention, both in the general population and its subsets, including athletes (e.g., swimmers or sprinters starting off the block in response to auditory stimulus or volleyball players pushing off in response to visual stimulus), patients affected by diabetes or osteoporosis (e.g., for fall prevention) [4], youths with intellectual disabilities, and patients with acoustic or visual impairment [5,6].

The reaction times of aviation pilots to auditory or visual stimuli and the skills in the execution of complex movements in response to these stimuli are of paramount importance during flight [7]. Aviation requires a combination of decision-making and kinesthetic skills. Many tasks during aircraft flying require continuous visual and auditory monitoring of the cues outside and inside the aircraft. Hence, the basic requirements of the pilot profession include fast and efficient information processing and fast and accurate reaction time [8]. Kennedy et al. [9] found that greater intra-individual variability in reaction time had an adverse impact on the ability of the pilot to maintain control of the aircraft in a flight simulator. These observations highlight the importance of the reduction of reaction time to different types of stimuli for the multitude of different tasks.

The motor imagery (MI) technique, trying to develop precise mental representations of the motor ability, led to improved performance of skilled movements [10]. Similarly, the video observation (VO) aids short-term motor skills learning [11]. These results can have the mirror neurons system as neurofunctional and neuroanatomical basis, a system able to facilitate subsequent movement executions by directly matching the observed action to the internal simulation of that action [12].

While it is known that MI and VO can exert positive effects on motor skill performance [13], few data are available regarding the effectiveness of these neuromotor training techniques in the reduction of the reaction times. Therefore, the aim of the present study was to investigate the influence of MI and VO on the simple reaction time to auditory and visual stimuli.

#### **2. Materials and Methods**

#### *2.1. Subjects*

The study protocol was approved in advance by the Ethical Committee of the University of Naples Federico II (22 March 2017, protocol number 58/17). Each subject provided written informed consent before participating. Participants were recruited on a voluntary basis among adult (age >18 years) male pilot cadets enrolled at the Italian Air Force Academy in Pozzuoli (Italy). Subjects with painful conditions during the three previous months and subjects affected by known orthopedic, rheumatologic, visual, acoustic, or neurological diseases that could interfere with the correct execution of the study protocol were excluded.

#### *2.2. Procedure*

Before randomization, all eligible subjects performed a pre-test evaluation of the imagery ability, using the revised movement imagery questionnaire (MIQ-R) [14]. The MIQ-R scores were collected at baseline and compared between groups to test for the homogeneity in imagery ability but were not considered in randomization or as an outcome measure.

The auditory and visual reaction times were measured using the Optojump device (Microgate, Bolzano, Italy), a previously validated and used tool for the measurement of reaction times [5,15–17]. This device is based on an infrared led technology and composed of transmitting and receiving parallel bars. To measure the reaction time for the lower limbs, the bars were positioned on the floor and the subject stood between the bars. The subject received the instruction to jump, lifting both feet off the floor, in response to an auditory (sound produced by the device) or a visual (appearance of a green ball on the device screen) stimulus. For the measurement of the upper-limb reaction time, the bars were positioned on a table with adjustable height. The subject stood in front of the table and

the device height was set to allow positioning of both palms flat on the table between bars with full bilateral elbow extension and wrist extension. The subject received the instruction to lift both hands in response to auditory or visual stimulus. Each subject was allowed a single practice attempt to gain confidence with the equipment. All tests were supervised by a trained physician who was blinded to the group allocation.

The main outcome measure was the reaction time expressed in milliseconds (ms). Auditory and visual reaction times for lower and upper limbs were assessed during the same session, in the following order: three trials of auditory reaction time for lower limbs, three trials of visual reaction time for lower limbs, three trials of auditory reaction time for upper limbs, and three trials of visual reaction time for upper limbs. The mean value of each triplicate measurement was calculated for statistical analysis.

Subjects were randomly assigned to the control or experimental group, using dedicated online software (https://www.sealedenvelope.com/simple-randomiser/v1/lists). Following randomization, each subject received an identical-looking USB pen drive containing a video demonstrating the motor tasks necessary to complete either sham (control group) or specific (experimental group) VO/MI intervention protocol. All cadets participated in a training session dedicated to the principles and aim of the VO and MI techniques. Subjects were informed not to discuss the protocols, observe others during VO/MI protocol execution, or share any information about the study throughout its duration.

The VO/MI intervention comprised individual, supervised and non-directed sessions, carried out regularly for 4 weeks. For both sham VO/MI (control) and specific VO/MI (experimental) groups, each task represented in the video was repeated three times in a loop. No instructions were provided in the video about the observed tasks and their subsequent imaging and execution. The video, watched on a laptop 9.7" screen, had no audio, except for the sound in the task related to the auditory stimulation reaction time assessment observed by the specific VO/MI group.

Each VO session was followed by an MI session and then the actual movement execution. All subjects were instructed to keep their eyes closed during MI. The whole routine, involving imagining the tasks observed in the video and actual movements, was performed by the cadets, always using the same equipment and at the same time of the day. Recorded tasks were performed by a male age-matched model wearing the leisure uniforms worn by all cadets at the Academy. Subjects allocated to the experimental group watched a video depicting the auditory and visual reaction time assessment, which was identical to that carried out during the baseline and end-point assessments. Hence, the experimental group performed an MI activity and then the actual movements based on VO aimed at improvement of auditory and visual reaction times. Subjects in the control group watched a video depicting activity included in an everyday physical training program (running, static bench, full push-up, standing toe touch), performed in the gym, followed by VO and MI procedures involving those activities. Thus, the VO/MI practiced by the control group was not related in any way to the end task for which the reaction time was measured.

#### *2.3. Statistical Analysis*

The main outcome of the study was to assess the change in the reaction time to auditory and visual stimulation following a neuromotor intervention comprising specific VO/MI compared with a sham neuromotor intervention. Therefore, the null hypothesis of the study was that the specific VO/MI would have no impact on the reaction times.

The distribution of continuous variables was assessed using the Shapiro–Wilk test and reported as mean ± SD. The MIQ-R scores were considered ordinal and reported as median and interquartile range (IQR) for visual and kinesthetic subscales. Data were analyzed by a Mixed Model ANOVA. Time was considered as within-subjects factor (baseline and post-intervention evaluation). Sham and specific interventions (group variable) were considered as the between-subjects factor. Interaction between time and group was investigated. The correlation between the ∆ Reaction Time to auditory and the ∆ Reaction Time to visual stimulus was assessed by Pearson correlation coefficients. All tests were considered significant if the p value was less than 0.05. Data analysis was performed using STATA software (StataCorp. v.12, College Station, TX, USA).

#### **3. Results**

In total, 41 males were assessed for eligibility. One cadet was excluded due to a recent orthopedic injury. Included subjects were randomly assigned to the control (*n* = 19) or intervention (*n* = 21) group. The mean age was 21.05 years (SD 0.97, range 20–23) in the control group and 20.7 years (SD 0.96, range 20–23) in the experimental group (*p* = 0.573). The MIQ-R scores were similar between groups (control group: median 20, IQR 19–21; experimental group: median 19, IQR 18–20, *p* = 0.105 for the kinesthetic subscale and control: median 19, IQR 18–19; experimental: median 17, IQR 17–19, *p* = 0.101 for the visual subscale). *J. Funct. Morphol. Kinesiol.* **2020**, *5*, x FOR PEER REVIEW 4 of 9 **3. Results** 

The mean scores for the reaction time to auditory and visual stimuli for the upper and lower limbs pre- and post-intervention are reported in Figure 1 and the results of the analysis are reported in Table 1. The reaction time to auditory stimuli for the upper and lower limbs post-intervention improved significantly in both groups (control: mean change −40 ms, SD 40, experimental: mean change −40 ms, SD 80 for upper limbs; control: mean change −50 ms, SD 140, experimental: mean change −30 ms, SD 70 for lower limbs). A trend towards reduced reaction time to visual stimuli for upper and lower limbs was also observed in both groups (control: mean change −30 ms, SD 90, experimental: mean change −20 ms, SD 100 for upper limbs; control: mean change −30 ms, SD 90, experimental: mean change −20 ms, SD 80 for lower limbs). The effect of time was significant in all groups. The group effect for auditory and visual RT was not significant in upper limbs, while it was significant in lower limbs. In all comparisons, interaction term was not significant. In total, 41 males were assessed for eligibility. One cadet was excluded due to a recent orthopedic injury. Included subjects were randomly assigned to the control (*n* = 19) or intervention (*n* = 21) group. The mean age was 21.05 years (SD 0.97, range 20–23) in the control group and 20.7 years (SD 0.96, range 20–23) in the experimental group (*p* = 0.573). The MIQ-R scores were similar between groups (control group: median 20, IQR 19–21; experimental group: median 19, IQR 18–20, *p* = 0.105 for the kinesthetic subscale and control: median 19, IQR 18–19; experimental: median 17, IQR 17–19, *p* = 0.101 for the visual subscale). The mean scores for the reaction time to auditory and visual stimuli for the upper and lower limbs pre- and post-intervention are reported in Figure 1 and the results of the analysis are reported in Table 1. The reaction time to auditory stimuli for the upper and lower limbs post-intervention improved significantly in both groups (control: mean change −40 ms, SD 40, experimental: mean change −40 ms, SD 80 for upper limbs; control: mean change −50 ms, SD 140, experimental: mean

While results showed similar improvement in the reaction times in both groups, the correlation between the reduction of the reaction times to visual and auditory stimuli differed between the groups (Figure 2). In the experimental group, reductions in the reaction times to visual and auditory stimuli were significantly correlated, with a high coefficient, for both upper (*r* = 0.703) and lower limbs (*r* = 0.473). Conversely, this correlation was not significant in the control group (*r* = 0.262 for the upper and *r* = 0.09 for the lower limbs). change −30 ms, SD 70 for lower limbs). A trend towards reduced reaction time to visual stimuli for upper and lower limbs was also observed in both groups (control: mean change −30 ms, SD 90, experimental: mean change−20 ms, SD 100 for upper limbs; control: mean change −30 ms, SD 90, experimental: mean change −20 ms, SD 80 for lower limbs). The effect of time was significant in all groups. The group effect for auditory and visual RT was not significant in upper limbs, while it was significant in lower limbs. In all comparisons, interaction term was not significant.

**Figure 1.** Reaction times to visual and auditory stimuli for upper and lower limbs in the experimental (specific VO/MI) and control (sham VO/MI) groups. Data are reported in milliseconds as a mean and 95% confidence interval. Solid circle indicates experimental group (specific intervention), hollow circle indicates control group (sham intervention). **Figure 1.** Reaction times to visual and auditory stimuli for upper and lower limbs in the experimental (specific VO/MI) and control (sham VO/MI) groups. Data are reported in milliseconds as a mean and 95% confidence interval. Solid circle indicates experimental group (specific intervention), hollow circle indicates control group (sham intervention).

Auditory,


**Table 1.** Reaction times (RT) to auditory and visual stimuli for the upper and lower limbs pre- and post- VO/MI training in the control and experimental groups. Auditory, lower 530 (80) 480 (110) −50 (140) 470 (70) 440 (60) −30 (70)

*J. Funct. Morphol. Kinesiol.* **2020**, *5*, x FOR PEER REVIEW 5 of 9

**Table 1.** Reaction times (RT) to auditory and visual stimuli for the upper and lower limbs pre- and

**Change, Mean (SD)** 

upper 430 (70) 400 (80) −30 (90) 420 (90) 400 (60) −20 (100)

**Pre, Mean (SD)** 

**Post, Mean (SD)** 

**Change, Mean (SD)** 

**Limbs RT in ms Control Group Experimental Group** 

**Post, Mean (SD)** 

post- VO/MI training in the control and experimental groups.

**Pre, Mean (SD)** 

the upper and *r* = 0.09 for the lower limbs).

**Figure 2.** Correlation between the reduction in the reaction time (ΔRT) to visual and auditory stimuli **Figure 2.** Correlation between the reduction in the reaction time (∆RT) to visual and auditory stimuli for upper and lower limbs in the experimental (specific VO/MI) and control (sham VO/MI) groups.

#### for upper and lower limbs in the experimental (specific VO/MI) and control (sham VO/MI) groups. **4. Discussion**

**4. Discussion**  Our study demonstrated that a neuromotor intervention comprising a specific VO/MI did not significantly improve reaction times to visual and auditory stimuli for upper or lower limbs when compared with controls. However, the correlation between the reduction in reaction time to visual Our study demonstrated that a neuromotor intervention comprising a specific VO/MI did not significantly improve reaction times to visual and auditory stimuli for upper or lower limbs when compared with controls. However, the correlation between the reduction in reaction time to visual and auditory stimuli was demonstrated only in the experimental group. Importantly, we observed that the reaction times for upper and lower limbs to auditory stimuli in both groups were lower than those registered for visual stimuli at baseline. Following the specific or sham VO/MI training, reductions in reaction times in the experimental and control groups were always more significant for the auditory than visual stimuli.

Specific MI/VO sessions were programmed to respect the elements of successful interventions identified by Schuster et al. [18]; these were individual, supervised and non-directed sessions, added after physical practice. Furthermore, the whole specific routine (VO followed by MI followed by movement execution) was based on the approach suggested by Holmes and Collins [19], which incorporates physical, environment, timing, task, learning, emotion, and perspective (PETTLEP) elements into imagery. Monitoring and adjusting for as many PETTLEP elements as feasible were recently proposed to optimize the intervention outcome and to maximize the functional equivalence of imaged and actual task execution, since the PETTLEP technique was associated with a greater ease and/or vividness of the MI [20]. Its effectiveness was observed in several fields where the best possible performance of movement is crucial, such as sport, surgery or music [21]. Despite those previously reported positive effects in the performance of the movements, in our study, the VO/MI technique incorporating PETTLEP elements showed no advantage in improving reaction times to auditory and visual stimuli over the control group, in which those principles were not respected. This is in contrast with findings published by Simons et al. [22], who found that brain-training interventions improve performance in the trained tasks but not in the unrelated tasks. Nevertheless, their literature review did not include studies of MI or VO but focused on studies aimed at improving cognitive skills rather than simple reaction times.

Knowing that the reaction times positively correlate with physical fitness level [23], it can be argued that mandatory participation in the physical training program, that is included in the schedule of all Air Force Academy students, could have contributed to the improved reaction times observed in both control and intervention groups. Indeed, it was observed that students who exercised regularly had shorter reaction times than those who lead a sedentary life [24]. Shorter simple reaction times were also observed in elderly diabetic patients with or without neuropathy after moderate or intense supervised exercise program compared with pre-training [25]. Similarly, reaction time improved in children and adolescents with mild intellectual disability who participated in physical fitness training programs, compared to a control group [26]. Recent studies suggest that the negative effects of stressful conditions on workplace performance and, presumably, reaction time could be overcome by physical training [27,28].

The reaction time represents a complex neuromotor skill and it can be influenced by several external and internal factors, including type of stimulus (auditory, visual, or tactile), sex, age, physical fitness, level of fatigue, distraction, alcohol, personality type, dominant limb, biological rhythm, and health [13,23,29,30]. Accordingly, the study population selected for our study was homogeneous by gender, age, instruction level, biological rhythm, professional demands, instruction, and physical fitness, owing to a common resident training program applied to the entire sample by the Air Force Academy. Although it was not possible to control for all variables able to influence reaction times, many of the considered factors were homogeneously distributed in the study population.

Admittedly, the present study has still some limitations related to the subject of the investigation. Because studies of VO/MI effects on reaction time are lacking, we calculated the sample size for our study based on similar data on motor performance published elsewhere [13] and assumed a reduction of 150 ms in the auditory and visual reaction time as clinically significant. The absence of significant differences between groups could be caused by a high type II error with a low power of the results, due to the limited sample. As stated above, our study sample is represented by a highly homogenous population of the same gender, age, social status, lifestyle, education and physical activity. While this consideration allowed us to achieve an internal validity of the study design and results, at the same time, it may limit the generalization of the results. Regarding the choice of intervention, recent data suggested that a combination of model observation and self-observation had better short-term effects on motor performance than each VO method applied separately [11]. Nevertheless, only ideal model observation was used in the present study to explore, for the first time, its effect on the simple reaction time in association with MI. Further studies may be required to evaluate if the combination of both VO variants is more beneficial. Additional uncontrolled factors influencing the results of our study are placebo and expectation effects. Our choice of sham intervention in the control group, in which the VO/MI procedure was followed, but it was not related to the end task for which the reaction time was measured, allowed us to create an active control group. Recent studies indicate, however, that it may

not be sufficient for eliminating a placebo effect, since the experimental and active control group can develop different expectations of improvement, even if interventions are incomparable [31].

In activities that take place in relatively unpredictable and constantly changing environment, movements have to be continuously adapted. Thus, developing physical and motor capabilities is as important as improving sensory and cognitive skills. This is particularly relevant for, but not limited to, open-skill sport activities [32]. In closed-skill sports (swimming, running), reaction time to auditory stimulus is often determinant for success. Additionally, in some professional activities, such as aviation, reaction must follow prompts of different types, including visual and auditory stimuli. The current study found that specific VO/MI training related to the end task, whereby the reaction time is measured, allows a parallel reduction in the reaction time for both types of stimuli. Further research in the cognitive and neurophysiological field, possibly incorporating the concept of spatiotemporal window for multisensory integration [33], is needed to explore how this correlation influences the information process; this could lead not only to better simple reaction time, but also to better recognition or choice reaction times.

#### **5. Conclusions**

In conclusion, we observed that the reaction times for upper and lower limbs to auditory stimuli were always lower, at baseline, than those registered for visual stimuli. The neuromotor training comprising specific VO and MI procedures in the experimental group did not determine a significantly higher reduction in the simple reaction time to auditory and visual stimuli than the sham procedure in the control group. Indeed, a significant reduction of reaction time to auditory stimulus and a trend towards reduction of reaction time to visual stimulus was observed post-intervention for the upper and lower limbs in both groups. Only the specific VO/MI training, however, produced a linear correlation between the improvement in the reaction time to auditory and visual stimuli. Interestingly, the reductions in reaction times in the experimental and control groups were always more significant for the auditory than visual stimuli. These findings could be crucial in training programs for aviation cadets and other professionals who need to improve their reaction times to a multitude of stimuli.

**Author Contributions:** Conceptualization, F.S., V.R. and G.S.; Data curation, I.B. and V.D.; Formal analysis, A.M.S. and C.C.; Investigation, A.M.S., I.B. and D.N.; Methodology, F.S. and E.D.V.; Project administration, S.M. and F.D.M.; Resources, V.D. and G.S.; Supervision, F.S.; Validation, E.D.V. and F.D.M.; Visualization, C.C. and S.P.; Writing—Original draft, A.M.S., S.P. and F.D.M.; Writing—Review & editing, V.R., D.N. and S.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Brief Report*
