*Article* **In-Season Internal and External Workload Variations between Starters and Non-Starters—A Case Study of a Top Elite European Soccer Team**

**Rafael Oliveira 1,2,3, \* , Luiz H. Palucci Vieira 4 , Alexandre Martins 1,2 , João Paulo Brito 1,2,3 , Matilde Nalha 1 , Bruno Mendes <sup>5</sup> and Filipe Manuel Clemente 6,7**


**Abstract:** *Background and Objectives:* Interpretation of the load variations across a period seems important to control the weekly progression or variation of the load, or to identify in-micro- and mesocycle variations. Thus, the aims of this study were twofold: (a) to describe the in-season variations of training monotony, training strain and acute:chronic workload ratio (ACWR) through session ratings of perceived exertion (s-RPE), total distance and high-speed running (HSR); and (b) to compare those variations between starters and non-starters. *Materials and Methods:* Seventeen professional players from a European First League team participated in this study. They were divided in two groups: starters (*n* = 9) and non-starters (*n* = 8). The players were monitored daily over a 41-week period of competition where 52 matches occurred during the 2015–2016 in-season. Through the collection of s-RPE, total distance and HSR, training monotony, training strain and ACWR were calculated for each measure, respectively. Data were analyzed across ten mesocycles (M: 1 to 10). Repeated measures ANOVA was used with the Bonferroni post hoc test to compare M and player status. *Results:* The results revealed no differences between starters vs. non-starters (*p* > 0.05). M6 had a greater number of matches and displayed higher values for monotony (s-RPE, total distance and HSR), strain (only for total distance) and ACWR (s-RPE, TD and HSR). However, the variation patterns for all indexes displayed some differences. *Conclusions:* The values of both starters and non-starters showed small differences, thus suggesting that the adjustments of training workloads that had been applied over the season helped to reduce differences according to the player status. Even so, there were some variations over the season (microcycles and mesocycles) for the whole team. This study could be used as a reference for future coaches, staff and scientists.

**Keywords:** acute/chronic workload ratio; high-speed running; in-season; non-starters; RPE; soccer; starters; training monotony; training strain

#### **1. Introduction**

Monitoring of the training load in soccer has become popular, whereby two main dimensions of load are considered [1]: (i) internal and (ii) external. The external load

**Citation:** Oliveira, R.; Palucci Vieira, L.H.; Martins, A.; Brito, J.P.; Nalha, M.; Mendes, B.; Clemente, F.M. In-Season Internal and External Workload Variations between Starters and Non-Starters—A Case Study of a Top Elite European Soccer Team. *Medicina* **2021**, *57*, 645. https://doi.org/ 10.3390/medicina57070645

Academic Editor: José Antonio de Paz Fernández

Received: 3 May 2021 Accepted: 21 June 2021 Published: 23 June 2021

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

**Copyright:** © 2021 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 (https:// creativecommons.org/licenses/by/ 4.0/).

can be considered as the physical demands that occur in the players in response to the implemented drill/task, while the internal load corresponds to the psychophysiological responses to the external load [2]. Different outcomes can be considered for each of the dimensions, although the rate of perceived exertion (RPE) and heart rate responses are the most used measures associated with internal load [3]. On the other hand, in soccer, the external load is typically characterized by the distance covered at different speed thresholds, or the inertial-derived measures such as accelerations/decelerations or composite variables (e.g., player load) [4].

Monitoring loads allows one to identify the consequence of training plans on the players and to individualize the analysis [5]. Although it is useful to look for accurate measures representing the impact in a training session [6], interpretation of the load variations across a period of time also seems to be important [7]. In fact, calculating workload measures is a part of the strategies to control the weekly progression or variation of the load, or to identify within-week variations [8]. Among the possibilities, acute load (representing the accumulated load during a week), chronic load (typically represented by the mean load in the past weeks), acute:chronic workload ratio (ACWR, representing the relationship between acute and chronic workloads) [9], training monotony (TM) (representing the variability of load within the week) and training strain (TS) (representing the variability of the load multiplied by the acute load) [10] are some examples of how to control load taking into consideration different measures.

Considering that some of these measures are sensitive to load fluctuations, it can be expected that participating or not participating in soccer matches may influence the workload measures reported for the players. For example, it is expectable that players with greater participation in matches present greater values of accumulated load and chronic load. However, as a consequence, players with less participation should be carefully managed to be prepared for participating in matches and coping with a spike in load. Despite the apparently obvious consequence of participating more or less in matches being related with different workload measures, reports on this matter are limited [11]. For example, similar comparisons between starters and non-starters regarding the workload measures of new body load and metabolic power were found [11]. In junior soccer players, it was also found that weekly internal and external load measures were also significantly greater in starters than in substitute players [12].

However, the above-mentioned results still need more research that provides some description about the workload measures' variations in accordance with the level of participation of players in elite soccer. This should be further researched to provide information about how to manage players with match stimulus and to identify possible strategies to level the load with individualized training for those who are not playing. Based on that, the aims of this study were twofold: (a) to describe the in-season variations of TM, TS and ACWR through s-RPE, total distance and high-speed running (HSR); and (b) to compare those variations between starters and non-starters.

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

#### *2.1. Subjects*

Seventeen elite soccer players participated in this study. The players belong to a team that participated in the UEFA Champions League. They were divided into two groups: starters (*n* = 9, age 26.2 ± 3.5 years, 180.1 ± 6.8 cm and 78.7 ± 5.8 kg) and non-starters (*n* = 8, 24.5 ± 4.6 years, 182. ± 6.8 cm and 76.6 ± 4.3 kg). The inclusion criteria were regular participation in most of the training sessions (80% of weekly training sessions), while the exclusion criteria included lack of player information, illness and/or injury for two consecutive weeks. Goalkeepers were excluded from the study. The criteria to define starters and non-starters were assessed week by week against a player´s attendance time at the match and training sessions, and to be considered a starter, a player had to complete at least 60 minutes in three consecutive matches; players who did not achieve this duration were considered non-starters [13]. All participants were familiarized with the

training protocols and signed informed consent prior to the investigation. This study was conducted according to the requirements of the Declaration of Helsinki and was approved by the Ethics Committee of Polytechnic Institute of Santarém (252020Desporto).

#### *2.2. Design*

Training load data were collected over a 41-week competition period, in which 52 matches occurred during the 2015–2016 in-season. The team used for data collection competed in four official competitions across the season, including the UEFA Champions League, the national league and two more national cups from their own country. For the purposes of the present study, all of the sessions carried out as the main team sessions were considered. This refers to training sessions in which both the starting and non-starting players trained together. Only data from training sessions were considered. Data from rehabilitation or additional training sessions of recuperation were excluded. This means that sessions after the match day were included whenever both starters and non-starters trained together, but other kinds of recovery training were excluded. This study did not influence or alter the training sessions in any way. Training data collection for this study was carried out at the soccer club's outdoor training pitches. Total minutes of training sessions included the warm-up, main phase and slow-down phase plus stretching.

The season was organized into 10 mesocycles (M: 1–10). The number of training sessions, number of competitive matches and total training duration for starters and non-starters are presented in Table 1.


**Table 1.** Training sessions and number of competitive matches during the 41-week period.

ST = Starters; NST = Non-starters.

#### *2.3. Internal Training Load Quantification*

During training sessions, the CR10-point scale, adapted by Foster et al. was applied [14]. Specifically, thirty minutes after the end of each training session, players rated their RPE value using an app on a tablet. The scores provided by the players were then multiplied by the training duration to obtain the s-RPE [14,15]. The players were previously familiarized with the scale, and all answers were provided individually to avoid nonvalid scores.

#### *2.4. External Training Load Quantification*

Global positioning system (GPS) units (Viper pod 2, STATSports, Belfast, UK) with 10 Hz frequency were used to monitor the training duration, total distance and HSR (above 19 km/h) for each player. For better satellite reception of the GPS antenna, the GPS unit was placed on the upper back between the left and right scapula through a custom-made vest. Previously, Beato et al. [16] positively tested the validity and reliability of linear, multidirectional and soccer-specific activities through this system. Thirty minutes before the start of a training session, all devices were turned on to acquire satellite signals and to provide synchronization between the GPS clock and the satellite's atomic clock. After the training sessions, the Viper PSA software (STATSports, Belfast, UK) was used to download data and to clip the entire training session (i.e., from the beginning of the warm-up to the end of the last organized drill). In order to avoid inter-unit error, players wore the same GPS device in each training session.

#### *2.5. Calculations of Training Indexes*

Through s-RPE, total distance and HSR, the following variables were calculated: (i) TM (mean of training load during the seven days of the week divided by the standard deviation of the training load of the seven days) [11,17], (ii) TS (sum of the training loads for all training sessions during a week multiplied by training monotony) [11,17] and (iii) ACWR (dividing the acute workload, i.e., the 1-week rolling workload data, by the chronic workload, i.e., the rolling 4-week average workload data) [18–22].

#### *2.6. Statistical Analysis*

Data were analyzed using SPSS version 22.0 (SPSS Inc., Chicago, IL, USA) for Windows. Initially, descriptive statistics were used to describe and characterize the sample. The Shapiro–Wilk and Levene tests were used to test the assumption of normality and homoscedasticity, respectively. A repeated measures ANOVA was used with the Bonferroni post hoc test once variables obtained normal distribution (Shapiro–Wilk > 0.05), and the Friedman and Mann–Whitney tests were used for variables that did not obtain normal distribution in order to compare different M and groups. Hedge's g effect size (95% confidence interval) was also calculated. Hopkins' thresholds for effect size statistics were used, as follows: ≤0.2, trivial; >0.2, small; >0.6, moderate; >1.2, large; >2.0, very large; and >4.0, nearly perfect [21]. Results were considered significant with *p* ≤ 0.05.

#### **3. Results**

Figures 1–3 show an overall view of the weekly average for TM, TS and ACWR calculated through the s-RPE, total distance and HSR across the in-season for starter and non-starter players. Overall, Figure 1 shows that the highest TMs-RPE occurred in week 1 for both starters and non-starters (7.2 and 7.0 AU, respectively), while the lowest value occurred in week 19 for starters (1.5 AU) and week 2 for non-starters (1.5 AU). The highest TSs-RPE occurred in week 41 for both starters (8498.0 AU) and non-starters (15,263.9 AU), while the lowest values occurred in week 30 for starters (110.2 AU) and week 19 for nonstarters (1310.9 AU). The highest ACWRs-RPE occurred in week 21 for starters (1.6 AU) and week 10 for non-starters (1.5 AU), while the lowest ACWRs-RPE occurred in week 36 for starters (0.5 AU) and week 17 for non-starters (0.7).

Figure 2 shows that the highest TMTD occurred in week 21 for both starters and nonstarters (38.2 and 17.1 AU, respectively), while the lowest values occurred in week 2 for both starters and non-starters (2.0 and 1.9 AU, respectively). The highest TSTD occurred in week 21 for starters (558,935.0 AU) and week 15 for non-starters (282,938.6 AU), while the lowest values occurred in week 36 for starters (35,441 AU) and non-starters (42,676.8 AU). The highest ACWRTD occurred in week 10 for both starters (1.6 AU) and non-starters (1.6 AU), and the lowest ACWRTD occurred in week 36 for both starters (0.7 AU) and non-starters (0.8 AU).

Figure 3 shows that the highest TMHSR occurred in week 21 for starters (2.9 AU) and week 36 for non-starters (2.9 AU), while the lowest values occurred in week 20 for starters (0.7 AU) and week 39 for non-starters (0.8 AU). The highest TSHSR occurred in week 4 for starters (3855.6 AU) and week 10 for non-starters (3578.0 AU), while the lowest values occurred in week 18 for starters (218.1 AU) and week 14 for non-starters (365.8 AU). The highest ACWRHSR occurred in week 10 for both starters (1.6 AU) and non-starters (1.6 AU), while the lowest ACWRHSR values occurred in week 9 for starters (0.4 AU) and week 4 for non-starters (0.4 AU).

Table 2 presents the average values and differences between starters and non-starters during the 10 mesocycles for all variables analyzed. There are no differences between the groups.

*Medicina* **2021**, *57*, x FOR PEER REVIEW 5 of 17

**Figure 1. Figure 1.** TM, TS (TM, TS ( **A**) and ACWR (**A**) and ACWR ( **B**) variations calculated through the s-RPE across 41 **B**) variations calculated through the s-RPE across 41 weeks for starters and non-starters. weeks for starters and non-starters.

*Medicina* **2021**, *57*, x FOR PEER REVIEW 6 of 17

**Figure 2.** TM, TS (**A**) and ACWR (**B**) variations calculated through the total distance across 41 weeks for starters and non-starters. **Figure 2.** TM, TS (**A**) and ACWR (**B**) variations calculated through the total distance across 41 weeks for starters and non-starters.

**Figure 3.** TM*,* TS (**A**) and ACWR ( **Figure 3. B**TM, TS () variations calculated through the HSR across 41 we**A**) and ACWR (**B**) variations calculated through the HSR across 41 weeks for starters and non-starters.eks for starters and non-starters.


**Table 2.**Differences between starters and non-starters during the 10 mesocycles, mean±SD.

M = mesocycle; RPE = rating of perceived exertion; s-RPE = session rating of perceived exertion; TM = training monotony; TS = training strain; ACWR = acute:chronic workload ratio; AU = arbitrary units; ST = starters; NST = non-starters.

Figures 4–6 show the differences between mesocycles for TM, TS and ACWR calculated through the s-RPE, TD and HSR across the in-season for the whole team. Figures 4–6 show the differences between mesocycles for TM, TS and ACWR calculated through the s-RPE, TD and HSR across the in-season for the whole team.

**Figure 4.** TM, TS (**A**) and ACWR (**B**) variations calculated through the s-RPE across 10 mesocycles for the whole team. M: mesocycle; a: difference from M5; b: difference from M6; c: difference from M7; d: difference from M8; e: difference from M9; f: difference from M10. **Figure 4.** TM, TS (**A**) and ACWR (**B**) variations calculated through the s-RPE across 10 mesocycles for the whole team. M: mesocycle; a: difference from M5; b: difference from M6; c: difference from M7; d: difference from M8; e: difference from M9; f: difference from M10.

Overall, Figure 4A shows that the highest TMs-RPE occurred in M6 and the lowest value occurred in M5. There only was one significant difference for TMs-RPE in M4 > M5 (ES = 0.17). The highest TSs-RPE occurred in M1 and the lowest value occurred in M5. There was a significant difference in M1 > M5 (ES = 1.50); M3 > M5 (ES = 1.57); M4 > M5 (ES = 1.42); M5 < M8 (ES = −0.62) and <M10 (ES = −0.97).

Figure 4B shows that the highest ACWRs-RPE occurred in M6 while the lowest ACWRs-RPE occurred in M5. There were significant differences in M1 > M5 (ES = 1.63) and <M6 (ES = 7.60); M3 > M5 (ES = 11.75) and <M6 (ES = −10.69); M4 < M6 (ES = −1.42); M5 < M6 (ES = −8.75), <M7 (ES = −9.35), <M8 (ES = −9.25), <M9 (ES = −8.33) and <M10 (ES = −7.17); M6 > M8 (ES = −7.25) and >M10 (ES = 5.85).

Overall, Figure 5A shows that the highest TMTD occurred in M6 and the lowest value in M1. There were significant differences in M1 < M2 (ES = −7.80), <M3 (ES = −5.70), <M4 (ES = −6.18), <M5 (ES = −3.81), <M6 (ES = −1.55) and <M7 (ES = −8.03); M2 < M4 (ES = −6.42); M4 > M7 (ES = −4.89) and M9 (ES = −0.93). The highest TSTD occurred in M6 and the lowest value occurred in M2. There were significant differences in M1 < M2 (ES = −6.52), <M2 (ES = −5.35), <M3 (ES = −5.03) and <M10 (ES = −4.33); M2 < M4 (ES = −4.73), >M5 (ES = −2.92), >M7 (ES = −1.69); M3 > M5 (ES = −2.63), >M7 (ES = −1.63),

>M9 (ES = −3.00) and M4 > M5 (ES = −1.98), >M7 (ES = −1.51), >M9 (ES = −2.00). Additionally, M7 < M10 (ES = −3.52). *Medicina* **2021**, *57*, x FOR PEER REVIEW 11 of 17

**Figure 5.** TM, TS (**A**) and ACWR (**B**) variations calculated through the total distance across 10 mesocycles for the whole team. M: mesocycle; a: difference from M2; b: difference from M3; c: difference from M4; d: difference from M5; e: difference from M6; f: difference from M7; g: difference from M8; h: difference from M9; i: difference from M10. **Figure 5.** TM, TS (**A**) and ACWR (**B**) variations calculated through the total distance across 10 mesocycles for the whole team. M: mesocycle; a: difference from M2; b: difference from M3; c: difference from M4; d: difference from M5; e: difference from M6; f: difference from M7; g: difference from M8; h: difference from M9; i: difference from M10.

Figure 5B shows that the highest ACWRTD value occurred in M6, while the lowest value occurred in M5. There were significant differences in M1 > M2 (ES = −12.21) and >M5 (ES = −17.02). M2 < M3 (ES = −12.18), <M4 (ES = −12.05), <M6 (ES = −10.95), <M7 (ES = −13.75) and <M8 (ES = −13.42). M3 > M5 (ES = −12.99). M4 > M5 (ES = −15.64). M5 < M6 (ES = −14.30), <M7 (ES = −16.41), <M8 (ES = −25.59), <M9 (ES = −23.62) and <M10 (ES = −13.89).

Overall, Figure 6A showed that the highest TMHSR occurred in M6 and the lowest value in M1. There were significant differences in M1 < M3 (ES = −5.42), < M6 (ES = −5.47). M2 < M6 (ES = −4.95). The highest TSHSR occurred in M1 and the lowest value occurred in M5. There were significant differences in M2 > M6 (ES = 1.55), > M5 (ES = 0.16), > M7 (ES = 0.15), > M8 (ES = 0.32) and > M9 (ES = 0.40). M3 > M5 (ES = 0.15) and > M8 (ES = 0.19). M5 < M10 (ES = −0.79). Additionally, M7 < M10 (ES = −1.56).

In Figure 6B, the highest ACWRHSR value occurred in M6 while the lowest value occurred in M5. There were significant differences in M3 > M5 (ES = −5.05). M5 < M6 (ES = −4.93), < M8 (ES = −5.75), < M9 (ES = −5.78) and < M10 (ES = −5.21).

**Figure 6.** TM, TS (**A**) and ACWR (**B**) variations calculated through the HSR across 10 mesocycles for the whole team. M: mesocycle; a: difference from M3; b: difference from M6; c: difference from M5; d: difference from M7; e: difference from M8; f: difference from M9; g: difference from M10. **Figure 6.** TM, TS (**A**) and ACWR (**B**) variations calculated through the HSR across 10 mesocycles for the whole team. M: mesocycle; a: difference from M3; b: difference from M6; c: difference from M5; d: difference from M7; e: difference from M8; f: difference from M9; g: difference from M10.

#### **4. Discussion**

Overall, Figure 4A shows that the highest TMs-RPE occurred in M6 and the lowest value occurred in M5. There only was one significant difference for TMs-RPE in M4 > M5 (ES = 0.17). The highest TSs-RPE occurred in M1 and the lowest value occurred in M5. There was a significant difference in M1 > M5 (ES = 1.50); M3 > M5 (ES = 1.57); M4 > M5 (ES = 1.42); M5 < M8 (ES = −0.62) and <M10 (ES = −0.97). Figure 4B shows that the highest ACWRs-RPE occurred in M6 while the lowest ACWRs-RPE occurred in M5. There were significant differences in M1 > M5 (ES = 1.63) and <M6 (ES = 7.60); M3 > M5 (ES = 11.75) and <M6 (ES = −10.69); M4 < M6 (ES = −1.42); M5 < M6 (ES = −8.75), <M7 (ES = −9.35), <M8 (ES = −9.25), <M9 (ES = −8.33) and <M10 (ES = −7.17); M6 > M8 (ES = −7.25) and >M10 (ES = 5.85). Overall, Figure 5A shows that the highest TMTD occurred in M6 and the lowest value in M1. There were significant differences in M1 < M2 (ES = −7.80), <M3 (ES = −5.70), <M4 (ES = −6.18), <M5 (ES = −3.81), <M6 (ES = −1.55) and <M7 (ES = −8.03); M2 < M4 (ES = −6.42); M4 > M7 (ES = −4.89) and M9 (ES = −0.93). The highest TSTD occurred in M6 and the lowest value occurred in M2. There were significant differences in M1 < M2 (ES = −6.52), <M2 (ES = −5.35), <M3 (ES = −5.03) and <M10 (ES = −4.33); M2 < M4 (ES = −4.73), >M5 (ES = −2.92), >M7 (ES = −1.69); M3 > M5 (ES = −2.63), >M7 (ES = −1.63), >M9 (ES = −3.00) and M4 > M5 (ES = −1.98), >M7 (ES = −1.51), >M9 (ES = −2.00). Additionally, M7 < M10 (ES = −3.52). The main purpose of the current study was to provide a description regarding training monotony (TM), strain (TS) and acute/chronic workload ratio (ACWR) based on perceived exertion, total distance (TD) and high-speed running (HSR) measures collected across in-season soccer. A secondary goal was to compare the time-related behavior of such metrics among starter and non-starter players. Our results in an elite European soccer team squad showed the following: (i) in the mesocycle with a greater number of matches disputed, higher values of various indices occurred, including either monotony (s-RPE, TD and HSR), strain (i.e., only TD in this specific case) or ACWR (s-RPE, TD and HSR); (ii) for all parameters considered, there were no significant differences between starters and non-starters; (iii) despite the similarities observed, players with a distinct status showed peak or lower values in distinct moments of the monitored period for some markers (e.g., lower TMs-RPE in the start and middle of the season, respectively, for non-starters and starters); (iv) higher monotony of perceived exertion was reported in the beginning, while for strain, it happened at the end of the season, independent of player status. In the following paragraphs, we will discuss the role of possible increase in match congestion on the data presented here while also accounting for the absence of differences between starters and non-starters and the common pattern of time-related variations in training monitoring parameters.

According to our results, the most intense period of games notably induced increases in monotony and ACWR (all variables) and concerning TSTD. Eight soccer fixtures were played across one month, rendering an average of at least two per week during the mesocycle. Indeed, this can characterize a full, congested schedule as per previous definitions [22,23]. Despite the monotony of s-RPE being more than twofold above the suggested threshold of 2 AU [10] as in the case of starters, the total duration of training sessions decreased for such players, while the same was not valid for non-starters. This is possible given the requirements of the latter to be more involved in active training/matches during that moment of the season and given the likely need to rotate players. In fact, congested fixture periods are linked to the possibility of inducing greater TS [8], whilst they can impair physical match performance [23–25] and raise injury risk [22,26]. However, the values for ACWR were all below 1.3, independent of player status (see Table 2), which, in theory, may not represent exacerbated injury likelihood [27], despite the fact that such a question lacks consensus to date (see, for example, Impellizzeri et al. [28]). Based on these assumptions, it seems that adjustments promoted during training may help avoid worst scenarios relating to management of players' workloads across the most congested period of matches in a season. However, particular attention should be paid to non-starters since they presented a high monotony and no reduction in total training time as compared to previous ones and aligned with a partly higher strain (i.e., TSTD) during the intense period of games.

One key finding of the present study was that when players were grouped according to their playing status as starters or non-starters, no significant differences were detected. This can suggest that contemporary soccer training methods require players to respond to stimuli delivered in a homogenous way, i.e., irrespective of whether generally starting the games or not. Importantly, one previous work verified opposing results, considering training TM and TS from accelerometer-derived variables, where starters showed greater values compared to their non-starter peers [14]. In contrast, non-starters may experience greater overall in-game physical exertion as compared to starters or players who participated in a whole match [29], and a similar condition was verified considering the most demanding passages of play [30]. Reports both have [31] and have not [32] confirmed the match–training-load associations in soccer. Of note, although there was no statistical significance here in the comparisons depending on player status, starters and non-starters reached maximal and minimal values for various markers at distinct moments. This finding can be related to distinct demands placed over each player across the season owing to situational-induced variations [33] and their prominent non-linear usage. Taken together, these assumptions could indicate that monitoring players on an individualized basis seems necessary, accounting for whether players generally start games on the pitch or the bench. Notwithstanding, traditional measures of workload such as TD and HSR may not be sensitive enough to detect possible status-related differences in monitoring strain, monotony and ACWR in training routines.

The findings from our investigation may assist in understanding the role of player status in various parameters used to control training in soccer as well potentially serving as a benchmark for future prescriptions and monitoring. Regardless of playing status and considering just the s-RPE, training monotony peaked at the beginning of the period (week 1), while the strain reached the largest values at the end (week 41). Such observations are different when compared to a six-week congestive period, which found lower values of monotony and strain in the first week, but the highest values in the last week for both variables [34]. The present results also disagree with the idea that a high degree of strain is often achieved when there is no competition (e.g., pre-season) [35]. High monotony early in the period may be indicative of either a poor ability of athletes to recognize the initial training loads or a true heavy stimulus applied, making it difficult to cope with as per the common fitness status of players at that moment. For example, the training session durations or strain levels were not the highest in the first week, whilst this does not hold true for TMs-RPE. Indeed, prior off-season training is recognized to impair physical capacity aspects [36] and it may have contributed to the prime TMs-RPE outputs.

Aside from the aforementioned potential derived implications (e.g., informing conditioning professionals on the effects of playing status and provision of reference values), a number of limitations of the present investigation should be highlighted. With the ever increasing energy requirements in soccer, the data gathered here may be outdated to some extent. The mostly descriptive nature of the work may limit its practical application. The generalizability of the results to other teams/countries, competitive standards and ages is also not warranted and requires replication studies. Complete description of training drills in further research can facilitate field implementation. Finally, co-variables such as match location, results and opponent quality should be considered in future studies as previously recommended [33].

#### **5. Conclusions**

To summarize, here, we observed across in-season soccer that spikes in training monotony, ACWR and strain for both internal and external load parameters (except regarding strain) may occur during match congestion intensification in elite soccer. Most importantly, apart from the extreme values being slightly discrepant (i.e., highest/lowest outcomes of the monitored markers varied according to playing status), starters and nonstarters behaved equally across the period, thereby suggesting a lack of differences between them in the adjustments of training workloads during the period. Finally, the progression of the training cycle phases elicited distinct responses of monitoring indices, such as the monotony of perceived exertion, which reached peak values at the early season, and major strain was reported at the end-season stage. The results suggest that the training load and management of load were properly addressed, despite some play-time differences across the season. Moreover, the present study shows that it is possible to have a congested mesocycle with eight matches with higher workloads (M6). In addition, this is the first study to report data for the 10 mesocycles of the in-season period and could be considered a reference for future studies.

**Author Contributions:** Conceptualization, R.O. and F.M.C.; methodology, R.O. and F.M.C.; software, R.O., A.M. and M.N.; validation, R.O., A.M. and J.P.B.; formal analysis, R.O. and M.N.; investigation, R.O., L.H.P.V., A.M., J.P.B., M.N., B.M. and F.M.C.; resources, R.O., L.H.P.V., A.M., J.P.B., B.M. and F.M.C.; data curation, B.M.; writing—original draft preparation, R.O., L.H.P.V., A.M. and F.M.C.; writing—review and editing, R.O., L.H.P.V., A.M., J.P.B. and F.M.C.; visualization, R.O., L.H.P.V., A.M., J.P.B. and F.M.C.; supervision, R.O., A.M. and F.M.C.; project administration, R.O. and J.P.B.; funding acquisition, R.O. and J.P.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Portuguese Foundation for Science and Technology, I.P., Grant/Award Number UIDP/04748/2020.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Ethics Committee of Polytechnic Institute of Santarém (252020Desporto).

**Informed Consent Statement:** Written informed consent was obtained from the participants to publish this paper.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** Luiz H. Palucci Vieira: ongoing PhD fellowship provided by São Paulo Research Foundation—FAPESP, under process number (2018/02965-7). "The opinions, hypotheses and conclusions or recommendations expressed in this material are the responsibility of the authors and do not necessarily reflect the views of FAPESP". Additionally, the authors would like to thank the team's coaches and players for their cooperation during all data collection procedures.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


## *Article* **Relationships between Fitness Status and Match Running Performance in Adult Women Soccer Players: A Cohort Study**

**Lillian Gonçalves 1, \* , Filipe Manuel Clemente 2,3 , Joel Ignacio Barrera 4 , Hugo Sarmento 4 , Francisco Tomás González-Fernández 5 , Luiz H. Palucci Vieira 6 , António José Figueiredo 4 , Cain C. T. Clark <sup>7</sup> and J. M. Cancela Carral 1**


**Abstract:** *Background and Objectives:* The aim of this study was twofold: (i) to analyze the relationships between fitness status (repeated-sprint ability (RSA), aerobic performance, vertical height jump, and hip adductor and abductor strength) and match running performance in adult women soccer players and (ii) to explain variations in standardized total distance, HSR, and sprinting distances based on players' fitness status. *Materials and Methods*: The study followed a cohort design. Twenty-two Portuguese women soccer players competing at the first-league level were monitored for 22 weeks. These players were tested three times during the cohort period. The measured parameters included isometric strength (hip adductor and abductor), vertical jump (squat and countermovement jump), linear sprint (10 and 30 m), change-of-direction (COD), repeated sprints (6 × 35 m), and intermittent endurance (Yo-Yo intermittent recovery test level 1). Data were also collected for several match running performance indicators (total distance covered and distance at different speed zones, accelerations/decelerations, maximum sprinting speed, and number of sprints) in 10 matches during the cohort. *Results*: Maximal linear sprint bouts presented large to very large correlations with explosive match-play actions (accelerations, decelerations, and sprint occurrences; *r* = −0.80 to −0.61). In addition, jump modalities and COD ability significantly predicted, respectively, in-game high-intensity accelerations (*r* = 0.69 to 0.75; R <sup>2</sup> = 25%) and decelerations (*r* = −0.78 to −0.50; R <sup>2</sup> = 23–24%). Furthermore, COD had significant explanatory power related to match running performance variance regardless of whether the testing and match performance outcomes were computed a few or several days apart. *Conclusion*: The present investigation can help conditioning professionals working with senior women soccer players to prescribe effective fitness tests to improve their forecasts of locomotor performance.

**Keywords:** football; athletic performance; match analysis; sports training; GPS; high-intensity running

**Citation:** Gonçalves, L.; Clemente, F.M.; Barrera, J.I.; Sarmento, H.; González-Fernández, F.T.; Palucci Vieira, L.H.; Figueiredo, A.J.; Clark, C.C.T.; Carral, J.M.C. Relationships between Fitness Status and Match Running Performance in Adult Women Soccer Players: A Cohort Study. *Medicina* **2021**, *57*, 617. https://doi.org/10.3390/ medicina57060617

Academic Editors: Jan Bilski and Tatiana Moro

Received: 14 April 2021 Accepted: 11 June 2021 Published: 13 June 2021

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

**Copyright:** © 2021 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 (https:// creativecommons.org/licenses/by/ 4.0/).

#### **1. Introduction**

Soccer matches represent a well-known intermittent mode of exercise in which short periods of intense efforts are interspaced by periods of low-to-moderate intensity [1,2]. Thus, players must maintain a desired level of running intensity and recover rapidly to perform to the best of their ability [3,4]. The literature has demonstrated that women soccer players may cover 9 to 11 km per match while spending 99 ± 8.3 m·min−<sup>1</sup> performing low-speed running and 9.7 ± 3.7 m·min−<sup>1</sup> performing high-speed running [2,5–8].

In female soccer, sprinting is considered a high-intensity effort [9], and high-speed activity is considered an essential component of matches. Usually, such efforts occur during decisive moments in a match [7], though they represent only 8% to 12% of the total distance covered in a typical match [10]. Additionally, female players were found to perform between 70 and 190 high-intensity runs (>19.8 km·h −1 ) during a match [5,10,11], covering between 210 and 520 m [6,7,12,13].

To sustain such efforts, female soccer players should present well-developed fitness statuses that allow them to meet the various demands of a match. Regarding sprinting performance, typical fitness status values observed in women soccer players suggest that they can cover 10 m in 2.31 ± 0.21 s and 25 m in 4.52 ± 0.20 s [14–17]. For another determinant variable (i.e., lower limb power), typical values exhibited by women soccer players are 30.1 ± 3.7 cm in the squat jump and 31.6 ± 4.0 cm in the countermovement jump [18]. Both sprinting and lower-limb power are neuromuscular determinants of soccer performance. However, the sport overwhelmingly involves running at low-to-moderate intensities—thus, good cardiorespiratory performance is crucial.

Female players usually present maximal oxygen uptake values between 49.4 and 56.7 mL/kg/min [2,17]. Based on one of the most common field-based tests used in soccer (namely, the Yo-Yo intermittent recovery test level 1), elite women soccer players can cover 1224 ± 255 m during the test, while players from lower divisions cover 826 ± 160 m [2,17].

Since high-intensity runs and sprinting tend to decrease at the end of the match, they could be associated with fatigue [19–22]. Therefore, sustaining good aerobic levels can help players avoid the effects of fatigue when performing power-related actions. Naturally, a player's performance will be affected by multiple factors, such as their position [23,24]. For instance, research indicates that central defenders perform fewer high-intensity runs than other players [23,24].

Fitness status can support match running performance—however, the strength of this relationship differs depending on the type of demand imposed on the player and the physical quality. For example, repeated sprint ability seems to be significantly correlated with total and high-intensity distances covered in matches [23–38]. Total distance also presented large correlations with high-intensity running activities and aerobic performance in field-based tests performed by male and female youth soccer players [29–42].

However, very few studies have tested the relationships between fitness status and match running performance among female soccer players. Nevertheless, it is pertinent to consider which kind of fitness status best relates to specific efforts in matches since match running performance is a determinant of a player's ability to sustain a high performance level. Understanding this matter will help to emphasize and specify the training process. However, fitness status changes over time. As such, analyzing the relationships between match running performance and fitness status in different moments throughout a season can help to explore whether these relationships are influenced by time.

Following the above discussion, the present study aims to (i) analyze the relationships between fitness status (repeated sprint ability (RSA), aerobic performance, vertical jump height, and anthropometry) and match running performance and (ii) run a regression analysis to explain variations in total distance, high-speed running (HSR), and sprinting distance. The hypothesis of the study is that match running variables are explained by the fitness status of the players.

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

#### *2.1. Experimental Approach to the Study*

This 22-week study followed an observational analytic cohort design. Players were assessed three times during the cohort (Figure 1). The first and second assessments were separated by four weeks, whereas the second and third assessments were separated by 18 weeks. The intervals were varied to determine the relationship between the physical capacities assessed with the match running and variations observed in total distance, HSR, and sprinting distance during matches.

#### **Figure 1.** Timeline of the study.

Three participants were excluded from the analysis, and 22 participants remained. Assessment 2 was correlated with matches 1 to 4 (weeks 6 to 15), while assessment 3 was correlated with matches 7 to 10 (weeks 22 to 27). A correlation analysis was conducted between fitness variables and match running performance for each period of assessment. Additionally, multi-linear regression analysis was carried out considering the match running performance variables to determine how each of the three variables of interest influenced running performance.

#### *2.2. Participants*

Twenty-two women soccer players from a team participating in the first Portuguese league were observed (Table 1). The participants presented a mean age of 24.77 ± 6.49 years old and a height of 162.51 ± 7.08 cm. In the first assessment mean weight was 59.06 ± 9.50 kg. In the second assessment mean weight was 59.01 ± 9.30 kg and body mass 61.62 ± 9.50 kg. The sample included three goalkeepers, four external defenders, four central defenders, six midfielders, and five attackers. During the season, players participated in four training sessions per week and official matches on weekends.

The eligibility criteria that players had to meet to be included in the final sample were as follows: (i) completion of all three assessments; (ii) participation in at least 85% of training sessions, (iii) not being out of action for treatment for more than four weeks, and (iv) at least five years of experience.

Before the study began, all players were informed of the study's design and procedures. Afterward, each player signed an informed consent form. The study was approved by the local university (code: CTC-ESDL-CE001-2021; date: 18 March 2021) and followed the ethical standards as per the Declaration of Helsinki for studies involving humans.

∙ −

∙ −


**Table 1.** Physical fitness assessment (mean ± SD).

Note: VO2max was estimated by the next equation: Yo-Yo IR1 test: VO2max (mL/min/kg) = IR1 distance (m) × 0.0084 + 36.4 (Bangsbo. 2008); ADD: adductor strength; ABD: abductor strength; SJ: squat jump; CMJD: countermovement jump; COD: change-of-direction; Pmax: maximum power at repeated-sprint test; Pmin: minimum power at repeated-sprint test; Paverage: average power at repeated-sprint test; FI: fatigue index at repeated-sprint test; YYIR1: intermittent recovery test level 1; HRmax: maximal heart rate; VO2max: maximal oxygen uptake.

#### *2.3. Measures*

#### 2.3.1. Physical Fitness Assessment

Between August and January, three fitness assessments with similar demands occurred in three microcycles. For each assessment period (week), three days were dedicated to run the tests, interspaced by 24 h between them. Players had 48 h of rest before the first day of assessments of each week analyzed.

In the first training session of the week, players were tested for anthropometry and hip adductor and abductor strength. In the second training session, vertical jump height, changes of direction, and linear speed were assessed. In the third session, the repeated sprint ability test and the Yo-Yo intermittent recovery test level 1 were applied.

These tests always occurred at the same time (7:30 p.m.) and location. The linear speed, repeated sprint ability, and Yo-Yo intermittent recovery tests were performed on synthetic turf without rain at a mean temperature of 19.5 ± 3.4 ◦C and a relative humidity of 63 ± 4%. A warm-up was performed before all evaluations. Warm-ups consisted of low and self-paced running, followed by calisthenic exercises in which players performed two sets of 10 repetitions of walking lunges, single-leg deadlifts, and fontal and lateral high knee movements.

#### Anthropometry

Body weight (kg) was measured without shoes with a bioelectrical impedance analysis (BIA) device (Tanita BC-730) to the nearest 0.1 kg. Height (cm) was measured using a stadiometer (Type SECA 225, Hamburg, Germany) to the nearest 0.1 cm.

#### Repeated Sprint Ability

The running anaerobic sprint test (RAST) test was applied to test players' repeated sprint abilities. This test consisted of six runs of 35 linear meters (each interspaced by 10 s of rest), with no COD required [43]. The time (sec) of each effort was recorded using a photocell timing gate (Photocells, Brower Timing System, UT, USA), with one device positioned at the starting line and the other positioned at the finish line. The device had a resolution of one-thousandth of a second. The minimum and maximum peak power and the fatigue index were determined using the following equation [43]: Power = Weight × Distance<sup>2</sup> Time<sup>3</sup> and Fatigue = maxpower− minpower Sum of 6 sprints (s) .

#### Linear Sprinting

Players' 10- and 30-m linear sprint abilities were tested using photocell timing gates (Photocells, Brower Timing System, UT, USA) positioned at the start and finish lines. Participants began the test positioned 0.5 m behind the starting line in a two-point split stance. As with the repeated sprint test, the device used to measure the players' performance had a resolution of one-thousandth of a second. Each player's best result obtained from three separate trials was recorded as their sprint time.

#### Change-of-Direction

The zig-zag 20 m [40] test was used to assess COD. This test consists of four 5 m each set out at 100◦ . Times were once again recorded using photocells timing gates (Photocells, Brower Timing System, UT, USA) with a resolution of one-thousandth of a second. The typical error of the Photocells was between 0.04 and 0.06 s, while the smallest worthwhile change was between 0.11 and 0.17 s [41]. Subjects performed three trials, resting for at least three minutes between trials. The best time (lowest time in seconds) of the three trials was used for the analysis.

#### Squat and Countermovement Jump

Squat and countermovement jump heights were assessed, with the highest jumps (cm) recorded and used in the analysis. Both jumps were tested with an optical measurement system consisting of a transmitting and receiving bar (Optojump, Microgate, Bolzano, Italia).

Each participant started the squat jump test in a squat position (although self-selected, the recommendation was to stay approximately at 90◦ relative knee joint angle) with their hands on their waist. After spending three seconds in the squat position, the participant jumped by extending their legs and then landed in the same place. Each participant performed three trials, with 30 s of rest provided between jumps.

Each participant started the countermovement jump test from a standing position, with their hands on their waist. After spending three seconds in the standing position, the participant flexed their legs and then immediately extended them while jumping. Each participant performed three trials, with 30 s rest provided between jumps.

#### Yo-Yo Intermittent Recovery Test—Level 1

For the Yo-Yo Intermittent Recovery test, participants were to run 20 m from one mark to another and then return to the starting mark. After every 40 m covered, a 10-s recovery period is provided, during which time participants jog between two marks that are five meters apart (an audio beep is utilized to control participants' speed). The speed starts at 10 km/h, increasing progressively thereafter. The test ends when the athlete achieves voluntary exhaustion or does not reach one of the 20-m marks before or at the same time as the beep. At the end of the test, the number of completed levels and shuttles, as well as the total distance covered, were recorded. The total distance (meters) was recorded.

#### Hip Adductor and Abductor Strength

A dynamometer (Smart Groin Trainer, Neuro excellence, Portugal) was positioned on the thigh area of participants, who were asked to squeeze the tool for 20 s. Three trials were performed, with 10 s of rest between trials. The strength of the hip adductor and abductor was measured in kilograms. The highest value was used in the analysis.

#### 2.3.2. Match Running Performance

During the match, participants used a Global Position System (GPS) (SPI HPU, GP-Sports, Canberra, Australia). This device has a frequency of 15 Hz and accelerometer of 100 Hz, 16 G Tri-axis, and a magnetometer of 50 Hz. Participants were asked to use a tight-fitting vest during the match and the device was placed between the left and right scapula. The GPS device collected the speed (km·h −1 ), the maximal speed (km·h −1 ), the number of sprints, the time of each sprint (sec), and accelerations and decelerations executed during each match observed. Speed achieved during a match was divided into the following 6 zones: zone 1 (0–5.9 km·h −1 ), zone 2 (6–11.9 km·h −1 ), zone 3 (12–13.9 km·h −1 ), zone 4 (14–17.9 km·h −1 ), zone 5 (18–23.9 km/h), and zone 6 (>24 km·h −1 ). The acceleration and deceleration were also recorded and split into 3 zones: ace1 (1.0–1.9 m·s 2 ), ace2 (2.0–2.9 m·s 2 ), ace3 (3.0–4.0 m·s 2 ) and des1 (1.0–1.9 m·s 2 ), des2 (2.0–2.9 m·s 2 ), des3 (3.0–4.0 m·s 2 ). The external load collected for analysis were: total distance covered (m), the distance covered (m) in the different speed zones, accelerations (m·s 2 ), decelerations (m·s 2 ), the maximum speed achieved (km/h), and the number of sprints (n).

#### *2.4. Statistical Analysis*

Descriptive statistics were represented as mean ± SD. Normal distribution and homogeneity was tested with the Kolmogorov-Smirnov test on all data before analysis. A Pearson correlation coefficient *r* was used to examine the relationship between values of fitness assessment (hip strength (ADDs and ABDs); squat and countermovement jump (SJ and CMJ); change-of-direction test (COD in seconds); linear Sprinting (10 m and 30 m in seconds); repeated sprint ability test (Pmax, Pmin and FI); Yo-Yo intermittent recovery test 1 (YYIR1 distance)) and match running performance (total distance covered (D); speed achieved in zone 1 (Z1), zone 2 (Z2), zone 3 (Z3), zone 4 (Z4), zone 5 (Z5), and zone 6 (Z6); acceleration (ace1, ace2, ace3) and deceleration (des1, des2, des3); maximum speed achieved (MSA); and number of sprint (NS)). To interpret the magnitude of these correlations we adopted the following criteria: *r* ≤ 0.1, trivial; 0.1 < *r* ≤ 0.3, small; 0.3 < *r* ≤ 0.5, moderate; 0.5 < *r* ≤ 0.7, large; 0.7 < *r* ≤ 0.9, very large; and *r* > 0.9, almost perfect [44]. The changes over the assessment were determined using repeated measures ANOVA. Significant main effects were subsequently analyzed using a Bonferroni post hoc test. Effect size is indicated with partial eta squared for Fs. To interpret the magnitude of the eta squared we adopted the following criteria: η <sup>2</sup> = 0.02, small; η <sup>2</sup> = 0.06, medium; and η <sup>2</sup> = 0.14 large. Regression analysis was used to identify which fitness outcomes can better explain match running performance. All variables were examined separately in this regression analysis. The magnitude of R2 was interpreted as follows: >0.02, small; >0.13, medium; >0.23, large. Data were analyzed using Statistica software (version 10.0; Statsoft, Inc., Tulsa, OK, USA).

#### **3. Results**

Descriptive statistics were calculated for each variable (see Tables 1 and 2 for more information).

A repeated measures ANOVA with participants' mean hip strength (ADDs and ABD) did not reveal any effect of assessment *F* > 1, in both cases. Another repeated measures ANOVA with participants' mean squat and countermovement jump (SJ and CMJ) did not reveal an effect of assessment in SJ, *F* (1.12) = 2.42, *p* = 0.11, η <sup>2</sup> = 0.16. However, data showed a significant effect of assessment in CMJ, *F* (1.12) = 6.13, *p* = 0.01, η <sup>2</sup> = 0.33. Continuing with the same type of repeated measures ANOVA analysis with participant 's mean change-ofdirection (COD (s), COD (km·h −1 ), and COD (m·s −1 )) did not reveal any effect of assessment *F* > 1. In the same line, another ANOVA analysis with participants mean linear sprinting (10 m (s), 10 m (km·h −1 ), 10 (m·s −1 ), 30 m (s), 30 m (km·h −1 ), 30 (m·s −1 )) did not reveal any effect of assessment *F* > 1. A repeated measures ANOVA with participants' mean repeated sprint ability test (Pmax (s), Pmin (s), Paverage (s) and FI (%)) revealed an effect of assessment for Pmax (s), Pmin (s), and Paverage (s), *F* (1.12) = 4.86, *p* = 0.01, η <sup>2</sup> = 0.28, *F* (1.12) = 8.84, *p* = 0.001, η <sup>2</sup> = 0.42, and *F* (1.12) = 6.23, *p* = 0.01, η <sup>2</sup> = 0.34, respectively. Nevertheless, there was no effect of assessment for FI (%), *F* > 1. Particularly remarkable, a repeated measures ANOVA with participants' mean Yo-Yo intermittent recovery test level 1 (stage (n), YYIR1, distance (m), HRmax (bpm), and V02max (mL·kg−<sup>1</sup> ·min−<sup>1</sup> )) revealed an effect of assessment for stage (n), YYIR1, distance (m), and V02max (mL·kg−<sup>1</sup> ·min−<sup>1</sup> ), *F* (1.8) = 7.40, *p* = 0.001, η <sup>2</sup> = 0.48, *F* (1.8) = 7.40, *p* = 0.001, η <sup>2</sup> = 0.48, *F* (1.8) = 7.40, *p* = 0.01, η <sup>2</sup> = 0.42, respectively. However, HRmax (bpm) data did not show any effect of assessment, *F* > 1.

**Table 2.** Descriptive table of match running performance (mean ± SD).


Note: *n* per minute was calculated considering the time in match; NS: number of sprints; MSA: maximum speed achieved; zone 1 (Z1), zone 2 (Z2), zone 3 (Z3), zone 4 (Z4), zone 5 (Z5), and zone 6 (Z6); acceleration (ace1, ace2, ace3) and deceleration (des1, des2, des3); D: distance covered.

The effect of match running performance tested repeatedly (D, Z1, Z2, Z3, Z4, Z5, Z6, ace1, ace2, ace3, des1, des2, des3, MSA and NS = between match 1 to match 4 (n), match 7 to match 10 (n), match 1 to match 4 (n per min), match 7 to match 10 (n per min)) did not reveal any effect of assessment of any studied variable, *F* > 1, in all cases.

On the basis of data obtained, correlations analysis was performed in order to find the possible association between fitness assessment and match running. First, we performed analysis of assessment 2 and matches 1–4, and second, assessment 3 and matches 7–10. Consequently, the correlation between fitness assessment and match running (assessment 2 and matches 1–4) are summarized in Table 3. No significant correlations were found between fitness assessment and the next variables of match running (D, Z1, Z2, Z3, Z4, Z5, ace1, ace2, des1, des2, MSA, and NS). However, negative correlation was found between 30 m linear sprinting and ace3, *r* = −0.52, *p* = 0.24. Crucially, other negative correlations were found between COD and Z6 and ace3 and des3 (*r* = −0.57, *p* = 0.024; *r* = −0.59, *p* = 0.011; *r* = −0.50, *p* = 0.034, respectively).

Correlation analysis was performed in order to find possible association between fitness assessment and match running (assessment 3 and matches 7–10). All data are summarized in Table 4. No significant correlations were found between fitness assessment and the next variables of match running (D, Z2, Z3, ace1, and des1). Nevertheless, positive correlation was found between SJ and ace3, des2, des3, and NS (*r* = 0.75, *p* = 0.007; *r* = 0.64, *p* = 0.035; *r* = 0.63, *p* = 0.035, and *r* = 0.70, *p* = 0.016, respectively). Other positives correlations were found between CMJ and Z1, Z4, ace2, ace3, des2, des3, and NS (*r* = 0.61, *p* = 0.048; *r* = 0.63, *p* = 0.040; *r* = 0.64, *p* = 0.036, *r* = 0.69, *p* = 0.019, *r* = 0.67, *p* = 0.022, *r* = 0.62, *p* = 0.039, and *r* = 0.70, *p* = 0.016, respectively). Furthermore, negative correlations were encountered between 10 m and ace2, *r* = −0.61, *p* = 0.047; des2, *r* = −0.61, *p* = 0.050; and NS *r* = −0.75, *p* = 0.008. Negative correlations were encountered between 30 m and Z5, Z6, ace2, ace3, des2, des3, MSA, and NS (*r* = −0.63, *p* = 0.039; *r* = −0.70, *p* = 0.016; *r* = −0.68, *p* = 0.021, *r* = −0.77, *p* = 0.006, *r* = −0.68, *p* = 0.022, *r* = −0.68, *p* = 0.022, *r* = −0.68, *p* = 0.023, and *r* = −0.80, *p* = 0.003, respectively). In the same line, more negative correlations were found between COD and Z4, Z5, Z6, ace2, ace3, des2, des3, and NS (*r* = −0.68, *p* = 0.022; *r* = −0.80, *p* = 0.003; *r* = −0.77, *p* = 0.006, *r* = −0.76, *p* = 0.007, *r* = −0.84, *p* = 0.001, *r* = −0.78, *p* = 0.005, *r* = −0.75, *p* = 0.007, and *r* = −0.74, *p* = 0.010, respectively). In addition, another positive correlation was encountered between FI and MSA, *r* = 0.61, *p* = 0.043.

Lastly, a multilinear regression analysis was performed to verify which variable of fitness assessment (agreement with the correlation analysis) could be used to better explain match running performance (See Table 5. for more information).


**Table 3.**Correlations between fitness assessment and match running (assessment 2 and matches 1–4).

ADD: adductor strength; ABD: abductor strength; SJ: squat jump; CMJD: countermovement jump; COD: change-of-direction; Pmax: maximum power at repeated-sprint test; Pmin: minimum power at repeated-sprint test; <sup>P</sup>average: average power at repeated-sprint test; FI: fatigue index at repeated-sprint test; YYIR1: intermittent recovery test level 1; HRmax: maximal heart rate; VO2max: maximal oxygen uptake; NS: number of sprints; MSA: maximum speed achieved; zone 1 (Z1), zone 2 (Z2), zone 3 (Z3), zone 4 (Z4), zone 5 (Z5), and zone 6 (Z6); acceleration (ace1, ace2, ace3) and deceleration (des1, des2, des3); D: distance covered; \*: significant at*p*< 0.05. 71


**Table 4.**Correlations between fitness status and match running (assessment 3 and matches 7 to 10).

ADD: adductor strength; ABD: abductor strength; SJ: squat jump; CMJD: countermovement jump; COD: change-of-direction; Pmax: maximum power at repeated-sprint test; Pmin: minimum power at repeated-sprint test; <sup>P</sup>average: average power at repeated-sprint test; FI: fatigue index at repeated-sprint test; YYIR1: intermittent recovery test level 1; HRmax: maximal heart rate; VO2max: maximal oxygen uptake; NS: number of sprints; MSA: maximum speed achieved; zone 1 (Z1), zone 2 (Z2), zone 3 (Z3), zone 4 (Z4), zone 5 (Z5), and zone 6 (Z6); acceleration (ace1, ace2, ace3) and deceleration (des1, des2, des3); D: distance covered; \*: significant at*p*< 0.05.


**Table 5.** Values of regression analysis explaining fitness assessment and match running performance.

SJ: squat jump; CMJD: countermovement jump; COD: change-of-direction; Pmax: maximum power at repeatedsprint test; Pmin: minimum power at repeated-sprint test; Paverage: average power at repeated-sprint test; FI: fatigue index at repeated-sprint test; YYIR1: intermittent recovery test level 1; HRmax: maximal heart rate; VO2max: maximal oxygen uptake; NS: number of sprints; MSA: maximum speed achieved; zone 1 (Z1), zone 2 (Z2), zone 3 (Z3), zone 4 (Z4), zone 5 (Z5) and zone 6 (Z6); acceleration (ace1, ace2, ace3) and deceleration (des1, des2, des3); D: distance covered; \*: significant at *p* < 0.05.

#### **4. Discussion**

The main aim of the current study was to determine the magnitude of relationships between various fitness status measures (strength, power, single/repeated sprinting, and intermittent endurance) and match running performance in adult women soccer players competing at a high level. We also aimed to explain the match running variations based on fitness status. The main findings in the present Portuguese players indicate the following: (1) correlations between fitness and match running performance were dependent on the time frame separating the testing battery and the collection of running performance during actual match-play. (2) With only rare exceptions, isolated strength, intermittent endurance, and repeated sprint ability performance were not associated with, nor did they predict, match running performance. (3) Even considering the fact that tests for separate maximal sprint bouts (10 and 30 m) were largely to very largely associated, they failed to significantly explain the variance of match-play (e.g., explosive) locomotor variables. (4) Jump and

COD ability clearly allowed a good (medium-to-large) prediction to be obtained of in-game high-intensity accelerations and decelerations, respectively. Finally, (5) the latter evaluation method was the only fitness indicator that had significant power to predict match running performance independent of the interval between testing and match performance.

Some innovative aspects in the current study should be highlighted. First, while the majority of previous works adopted only correlations as a statistical treatment to evaluate the possible link between fitness status and match running performance in soccer [41,45–48], a decision was made to move further when providing recommendations mainly based on regression analysis, which reveals the weighted influence of players' physical capacity on their locomotor outputs during match-play. In addition, various investigations on the subject have tested players at a single time point [36–38,40,41,49,50]. Meanwhile, here, two distinct approaches were considered using a cohort design, one testing associations between fitness status determined near the match occurrences and another with a longer interval between them. Most importantly, less than 3% of evidence on the complexity of fitness-match running performance relationships in a soccer context [51] were derived from scientific studies including female players according to knowledge collated in reviews [27,52]. Nonetheless, only intermittent endurance (Yo-Yo IR1/IE2), aerobic fitness (laboratory treadmill tests) [2,53], and Wingate measurements [11] were previously related to match running performance in women's soccer. Again, this reinforces the originality of data presented in the current work and supports the critical appraisal of the findings' strengths and weaknesses, which is developed in the following paragraphs.

An important finding of the present investigation is that a fitness testing battery seems to have a relatively short expiration date to help preview match physical performance in female soccer players. Such is indicated by the frequency and strength of correlations between fitness status measures and match running performance, as well as the number of variables involved, which varied in the distinct moments. According to a recent critique piece [53–57], manipulating the interval between players' evaluations and matches was never previously addressed when the objective was to understand their associations. Here, when looking at fitness assessment 2, we noticed only four moderate-to-large (ranging from −0.59 to −0.50) correlations with matches 1 to 4 (e.g., COD and 30 m sprint tests with in-game very-high intensity accelerations) (Table 3). All these were performed across a 10-week period, where fitness tests and matches were separated by three to nine weeks.

In contrast, more than 30 large to very large correlation coefficients (ranging from −0.84 to 0.75) were found between fitness assessment 3 parameters and running outputs during matches 7–10 (e.g., SJ, CMJ, 10/30 m sprint and COD with in-game sprint occurrences) (Table 4). Such second analysis comprised a 6-week period in total, with an interval between test and match equal to no more than three weeks. Naturally, changes occurring between fitness assessments were not explained in this work, but possibly may be affected by the training process [26,58,59]. Reports have not yet confirmed the same for match running performance, even though these are related to each other crosswise [3,15,57]. However, based on the current observations, the usefulness of some fitness data may become outdated (or at least its relevance might be reduced) after approximately a month in women's soccer.

The local strength of hip adductor and abductor muscles and intermittent highintensity or endurance running bouts were not associated, nor were they predictors. However, 10- to 30-m sprint performances were largely to very largely related to match running outputs in the female players of the present study, though their shared variance had no statistical significance.

When comparing such results to those presented in the available literature, discrepancies are identified. For example, this was the case in intermittent endurance capacity, which was previously linked to match running performance in female senior players [3,57] as well as in senior and youth male populations [27]. However, recent studies have demonstrated that running outputs during small-sided games were associated with the outputs obtained in competitions [54] independent of the player's intermittent endurance profile [55]. Force, maximal velocity, and aerobic and anaerobic resistance are important fitness components arguably contributing to sustaining physical efforts experienced in soccer match-play, thereby representing frequent determinants to winning [56,57].

Notwithstanding, many issues likely compromise the utility of some fitness assessment protocols in the current format. For example, testing maximal sprint ability (single or successive running bouts) using only linear paths is criticized nowadays given the curved trajectory of most explosive in-game actions [58]. The very short intervals often offered between repeated all-out efforts do not match those encountered in actual matches [59]. Occurrences of near-to-maximal displacements can also be very uncommon in elite standards [60]. Hip adduction/abduction strength allows one to discriminate between distinct performance levels [61], yet the effective contribution of hip muscle strength to running kinematics is low [62,63]. To summarize, such points of view are in alignment with our results. The external validity of some popular fitness status markers in women's soccer is not always supported, and its indiscriminate use needs to be re-thought.

Conditioning professionals need to be aware of assessment tools that can remain consistent over time when classifying players based on their fitness status, as well as the potential implications in terms of match performance. In this sense, although the construct validity of the various methods tested here has been challenged, jump and COD ability provided reasonable predictions (R<sup>2</sup> ranging from 23 to 25%, respectively) (Table 5) for in-game high-intensity accelerations and decelerations. Furthermore, COD predicted match running performance regardless time between testing and match observations. In other words, regression models with inputs being COD data remained significant from assessment 2/matches 1–4 until assessment 3/matches 7–10.

Studies that have aimed to extract the most relevant game indicators in soccer suggest that accelerations and decelerations are among the main components of athletes' external loads [64]. Interestingly, decelerations are more frequent than accelerations in soccer matchplay [65]. In addition to being paramount to COD performance, skilled decelerations are also fundamental to a range of match events (e.g., rapid changes in speed, cutting maneuvers, and regaining ball possession) [66]. Therefore, change-of-direction seems to be a sensitive indicator of fitness status in female soccer players, as it may provide meaningful information about the next match profile, in particular the players' deceleration performance.

Aside from the novelty of the current investigation, a number of limitations are recognized and need to be accounted for in future work, as well as when making interpretations and generalizations based on the present evidence. For one, female players were grouped regardless of the general exertion of their positional role during matches. Studies in male soccer players have shown that fitness status and match running performance relationships can be position dependent [25,26,53]. Another limitation is that the recommended sample size (*N* = 80 players) as per Gregson et al. [67] was not met in the present investigation, though this has frequently been the case in similar research. It is possible given the practical difficulty of involving multiple clubs that the pertinence of large/potentially heterogeneous datasets in solving this problem also lacks consensus. In addition, none of the conducted tests required players to rely on technical-tactical performance. Instead, they evaluated physical capacity markers. Adopting protocols that more closely mimic game demands could enhance the ecological validity and, in turn, the predictive ability of a testing battery in informing to some extent and advancing physical performance during matches [54]. Finally, the games involved different opposition over the course of the study, which may have impacted the running demands completed in matches.

#### **5. Conclusions**

Including a change-of-direction ability test seems pertinent when assessing women soccer players, as it may partly predict match-play running performance regardless of whether the time separating the assessment and competition is shorter (testing immediately prior to/after competing) or longer (test-match moments interspaced by at least three weeks). It is something that provides preliminary evidence about the construct validity of COD testing and its likely robustness regarding common time-related changes in match running performance.

We also demonstrated that the further apart a fitness testing battery is carried out in relation to actual matches, the lower its value in predicting in-game running outputs. Finally, caution is required concerning conditioning professionals' extensive use of common testing procedures, such as isolated maximal sprints, intermittent high-intensity actions, or endurance bouts, as evidence in the present study revealed that these do not always provide useful information for forecasting inter-individual variations in match-play locomotor performance.

**Author Contributions:** L.G., F.M.C., and J.M.C.C. led the project, established the protocol and wrote and revised the original manuscript. J.I.B. and H.S. collected the data and wrote and revised the original manuscript. F.T.G.-F., C.C.T.C., A.J.F., and L.H.P.V. wrote and revised the original manuscript. All authors have read and agreed to the published version of the manuscript.

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

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board (or Ethics Committee) of the Polytechnic Institute of Viana do Castelo. School of Sport and Leisure (code: CTC-ESDL-CE001-2021).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Acknowledgments:** Filipe Manuel Clemente: This work is funded by Fundação para a Ciência e Tecnologia/Ministério da Ciência. Tecnologia e Ensino Superior through national funds and, when applicable, co-funded via EU funds under the project UIDB/50008/2020. Hugo Sarmento gratefully acknowledges the support of a Spanish government subproject "Integration ways between qualitative and quantitative data, multiple case development, and synthesis review as the main axis for an innovative future in physical activity and sports research" [PGC2018-098742-B-C31] (Ministerio de Economía y Competitividad. Programa Estatal de Generación de Conocimiento y Fortalecimiento Científico y Tecnológico del Sistema I+D+i), which is part of the coordinated project "New approach of research in physical activity and sport from mixed methods perspective" (NARPAS\_MM) [SPGC201800X098742CV0]. Luiz H Palucci Vieira: ongoing PhD fellowship from São Paulo Research Foundation–FAPESP, under process number [#2018/02965-7].

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

#### **References**


**Isaac López-Laval 1, \* , Rafel Cirer-Sastre 2 , Francisco Corbi <sup>2</sup> and Sebastian Sitko 1**


**\*** Correspondence: isaac@unizar.es

**Abstract:** *Background and Objectives:* The aim of the present study was to compare the impact of an incremental exercise test on muscle stiffness in the rectus femoris (RF), vastus lateralis (VL), biceps femoris (BF), and gastrocnemius (GL) among road cyclists of three performance levels. *Materials and Methods:* The study group consisted of 35 cyclists grouped according to their performance level; elite (*n* = 10; professional license), sub-elite (*n* = 12; amateur license), and recreational (*n* = 13; cyclosportive license). Passive muscle stiffness was assessed using myometry before and after an incremental exercise test. *Results*: There was a significant correlation between time and category in the vastus lateralis with stiffness increases in the sub-elite (*p* = 0.001, Cohen's *d* = 0.88) and elite groups (*p* = 0.003, Cohen's *d* = 0.72), but not in the recreational group (*p* = 0.085). Stiffness increased over time in the knee extensors (RF, *p* < 0.001; VL, *p* < 0.001), but no changes were observed in the knee flexors (GL, *p* = 0.63, BF, *p* = 0.052). There were no baseline differences among the categories in any muscle. *Conclusions*: Although the performance level affected VL stiffness after an incremental exercise test, no differences in passive stiffness were observed among the main muscles implicated in pedaling in a resting state. Future research should assess whether this marker could be used to differentiate cyclists of varying fitness levels and its potential applicability for the monitoring of training load.

**Keywords:** cyclist; myometry; stiffness; incremental cycling test

#### **1. Introduction**

Road cycling is a popular endurance sport characterized by its cyclic nature, variable intensity, and large training volumes [1]. Among the key determinants of road cycling performance, maximal oxygen uptake (VO2max) stands out as one of the values that best represents cardiorespiratory fitness [2]. Furthermore, this parameter has been proposed to classify endurance athletes based on fitness level [3]. Other performance determinants in road cycling include cycling economy or efficiency, tactical and technical skills, psychological resilience, body composition in accordance with the cycling discipline, and muscles' mechanical properties adapted to the riders' specialty [2,4]. The muscle mechanical properties might differ among cycling specialties: the most powerful riders are characterized by greater and shorter muscles and a greater predominance of fast twitch fibers when compared to climbers or time-trialists [5]. The relationship between cardiorespiratory fitness and muscle performance factors has been detailed by Hoper et al. (2013), who determined that the percentage of muscle fibers is influenced by the VO2max of the cyclist [6].

Among the mechanical characteristics of the muscle, stiffness provides information regarding its intrinsic property and post-effort response; furthermore, it is one of the main parameters that characterizes the viscoelastic properties of the myofascial complex [7]. A proper conceptualization of muscle stiffness requires an analysis of both muscular architecture and its functional aspect. Concretely, it should be defined as the biomechanical

**Citation:** López-Laval, I.; Cirer-Sastre, R.; Corbi, F.; Sitko, S. Characteristics of Pedaling Muscle Stiffness among Cyclists of Different Performance Levels. *Medicina* **2021**, *57*, 606. https://doi.org/10.3390/ medicina57060606

Academic Editor: Simon M. Fryer

Received: 20 May 2021 Accepted: 9 June 2021 Published: 11 June 2021

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

**Copyright:** © 2021 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 (https:// creativecommons.org/licenses/by/ 4.0/).

capacity of the tissue (characterized by the type, number, and composition of its muscular fibers) that impedes stretching and distensions [8].

Current literature establishes different conclusions regarding stiffness levels [9–12]. On the one hand, optimal levels of musculotendinous stiffness are highly correlated to significant increases in muscle performance, especially in situations where the stretch-shorter cycle component is key to optimal performance [9–11]. On the other hand, higher stiffness can also be considered a potential threat, since a greater risk of injury has been reported in those athletes who presented greater levels of muscular stiffness as a consequence of high training loads [12]. The literature regarding endurance activities is scarce but has concluded that the reduction in musculotendinous stiffness observed in these types of sports is a consequence of the fatigue generated by submaximal muscle contractions that are sustained in time [13,14]. Related to this, García-Manso et al. (2011) determined that the loss in contractile capacity induced by a long-distance race reflects changes in the neuromuscular response and fluctuations in the contractile capacity of the muscle [13]. Furthermore, Andonian et al. (2016) reported a decrease in quadriceps stiffness caused by an extreme mountain ultra-marathon [14].

Since road cycling is an endurance sport with no impact and in which power is applied to the pedals, the lower limb muscle stiffness of cyclists might differ from other endurance disciplines. To the authors' knowledge, only four previous studies have analyzed road cycling muscle stiffness with non-invasive tools [15–18]. Several aspects could be highlighted from these studies: muscular stiffness seems to be an important contributory factor to sprint performance [15] and is proportional to the cyclists' power output during sprints [16]. Klich et al. (2020) observed higher stiffness in sprinters compared to endurance track cyclists [17]. In this same sense, it is important to highlight the results obtained by Ditroilo et al. (2011), who established that cyclists with higher baseline muscular stiffness suffered greater stiffness losses under fatigue than those with lower baseline levels [18].

The active and passive measurement of muscle stiffness is normally a complex procedure as it requires either muscular biopsies or repeated maximal isometric contractions [8]. Both methods may generate pain and require recovery after the procedure [10]. The utilization of non-invasive techniques for passive measurement of muscle stiffness could be highlighted as a viable alternative, especially when used in a field setting. Tensiomyography, elastography, electromyography, and ultrasounds have been the preferred non-invasive methods in recent years [10], although they still require educated staff and extended time periods for data obtention [8]. The MyotonPRO® (Myoton Ltd., Tallinn, Estonia) is a non-invasive tool that allows for the measurement of passive muscle stiffness through short oscillatory impulses [19] that are generated on the skin and over the area of the analyzed muscle [20]. Previous studies have demonstrated that the device is valid and reliable (ICC = 0.75–0.96; *R* <sup>2</sup> = 0.95) [8,21] and has been used to measure the main muscles that participate in the pedaling action: rectus femoris (RF), vastus lateralis (VL) [22], hamstring [23], and gastrocnemius (GL) [20]. Klich et al. (2019) proposed myotonometry as an easy and suitable tool to assess the viscoelastic characteristics of muscles in cyclists [24].

It is accepted that both fatigue and performance level of the athlete can influence the changes in muscle stiffness [15,18,20]. Given that previous studies have determined a relationship between stiffness and performance level in other sports, it could be hypothesized that this relationship could also exist in an endurance discipline such as road cycling. However, to date, the differences in muscle stiffness of road cyclists of different performance levels have not been examined. Further, the relationship between initial stiffness level and the response to fatigue after an incremental test is also an area of enquiry that has yet to be investigated. Accordingly, the aim of the present study was to compare the impact of an incremental exercise test on muscle stiffness among road cyclists of three performance levels.

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

#### *2.1. Research Design*

Thirty-five participants completed this cross-sectional study. The same order was followed during each individual assessment: anthropometry, evaluation of passive muscle stiffness (stiffness pre), incremental exercise test, and assessment of passive muscle stiffness after the incremental exercise test (stiffness post) (see Figure 1). In addition, during the assessment of passive muscle stiffness, the order was always the same between before and after the incremental exercise test and was standardized to avoid the influence of recovery time in muscle stiffness. Measurements were always taken in the same place (University Lab, Río Isuela Sport Center, Huesca, Spain), with a mean temperature of 21 ± 2 ◦C and mean relative humidity of 52% ± 9%). All participants were assessed during the preparatory phase of the annual training cycle, between the months of September and October and coinciding with the first preparatory meso-cycle. The evaluations were always carried out on Saturdays and Sundays at the same time (between 10 and 12 in the morning) to standardize the measurements and thus organize the schedule of travel to the laboratory. A 48-h rest period was established prior to the initial measurement to guarantee an adequate baseline assessment without fatigue. The assessment of the stiffness was carried out immediately after the completion of the incremental exercise test.

**Figure 1.** Experimental approach timeline and muscle assessment points. Blue marks: proximal and distal measurement area. Yellow mark: Myoton assessment point.

Assessment points placed at VL: 2/3 on the line from the anterior spina iliaca to the lateral side of the patella, RF: 50% on the line from the anterior spina iliaca superior to the superior part of the patella, BF: 50% on the line between the ischial tuberosity and the lateral epicondyle of the tibia, and GL:1/3 of the line between the head of the heel [25].

#### *2.2. Participants*

G∗Power version 3.1.9.2 [26] was used to estimate the required sample size in a 2 × 3 mixed design for a minimum expected effect size (Cohen's *F*) of 0.4, an α level of 0.05, and a power (1−β) of 0.95. This procedure returned a minimum number of 30 participants [17]. Thirty-five male cyclists volunteered to participate in the present study. The main characteristics of the participants are reported in Table 1. Participants were allocated to groups according to their performance level: elite, cyclists with a professional license (*n* = 10); sub-elite, cyclists with an U23 or amateur license (*n* = 12); recreational, cyclists that participate in cyclosportive events (*n* = 13). The inclusion criteria were to own a professional, U23, cyclosportive, or masters license. The exclusion criteria were (a) surgical procedures and injuries in the six months prior to the study and (b) use of performance-enhancing drugs in the six months prior to the study. After being informed of the benefits and potential risks of the investigation, all participants signed an informed consent form. The study followed the ethical guidelines of the 2013 Declaration of Helsinki and received approval from the Research Ethics Committee of the autonomous region of Aragon, Spain (the approval code is PI19/447, approved on 4th December 2019).

**Table 1.** Summary of the characteristics of the participants.


Values are expressed as mean SD; superscripts indicate statistically significant differences and their direction. E = elite, S = sub-elite, R = recreational.

#### *2.3. Data Collection*

Participants were weighed and measured by an internationally certified anthropometrist (ISAK level 2). Height (cm) was measured using the SECA-360 measuring rod (SECA©, Spain) with a precision of 1 mm, and bodyweight (kg) was measured using scales of the same brand with a precision of 0.1 kg [27]. The tests were carried out between 10 and 12 in the morning. The participants were asked to follow the feeding protocol used for races 3 h before the laboratory appointment. The tests were supervised by a sports doctor and two Bachelor of Science in Physical Activity specialists in performance assessment. The participants performed the entire measurement protocol with their own shorts, slippers, and clipless pedals.

A MyotonPRO® (Myoton Ltd., Estonia) was used to assess the passive stiffness of the main muscles involved in the pedaling action, RF, VL, biceps femoris (BF), and GL, before and after the incremental exercise test. To ensure correct measurements, the assessment points were drawn on the skin following the indications of Hermens et al. [25] (see Figure 1). After removing the tights, and with the cyclist in a lying position on a stretcher, the device was held perpendicular to the skin surface. It was then pushed (0.58 N for 15 ms) against the skin above the muscle area to reach the required depth (*d* = 3 mm). After the red light turned green, five short impulses (tap interval was 0.8 s) were produced automatically by the device in order to induce mechanical oscillations in the soft tissues. In order to guarantee the validity of the data obtained, only those evaluations in which the coefficient of variation was lower than 3% were considered. Otherwise, the assessment was repeated. All measurements were made by the same experienced researcher, and the intra- and inter-rater reliability for this device have been estimated in previous studies. [8,21]. The MyotonPRO® device provides data on the recorded passive muscle stiffness (S, N/m) [28]. The mean values for stiffness were calculated from the responses to the five impulses delivered.

Participants performed an incremental exercise test with gas exchange analysis (CPX/D Med Graphics, St. Paul, MN, USA, EE. UU. Measurement accuracy = 1%) [29] in the laboratory. Cyclists completed the graded exercise tests on their own bikes set up on the Wahoo KICKR Power Trainer (Wahoo Fitness, LLC, Atlanta, Georgia), which allows for power and cadence measurements and has been previously validated [30]. The incremental test was based on the following protocol: 10 min of warm up (5 min 100 W + 5 min 150 W) and increases of 25 W every 3 min [31,32]. The test stopped when a plateau of VO<sup>2</sup> was reached or, when not seen, at voluntary fatigue when at 100% of estimated HRmax, a respiratory exchange ratio of ≥1.15 and a rate of perceived exertion (RPE) of ≥18 [32]. The 6–20-point Borg scale was used [33]. All the participants were familiarized with the RPE scale as it was commonly used by their coaches. The scale was shown to the participant in the last 30 s of each one of the steps of the incremental test.

#### *2.4. Statistical Analyses*

Statistical analyses were performed in R version 4.0.1 (R Core Team 2020) using RStudio (RStudio Team 2020). Variables were visually inspected and described as mean (standard deviation) using the package *rstatix*. Stiffness differences were assessed by fitting an independent linear mixed-effects model for each muscle (rectus femoris, biceps femoris, gastrocnemius lateralis, and vastus lateralis) using the packages *lme4* and *lmerTest*. The models included fixed-effects terms for time (pre and post), category (elite, sub-elite, and recreational), and their interaction. Time at pre and the recreational category were the reference category in each factor, respectively. Random slopes were allowed to vary between moments (time) and random intercepts were allowed to vary among participants (id). Main effects were obtained performing an analysis of variance with each model, and post-hoc pairwise comparisons were performed comparing estimated marginal means using the emmeans package. The effect size of main effects was reported using partial eta squared (η 2 <sup>P</sup>) and interpreted as follows: η 2 <sup>P</sup> < 0.01 "small", η 2 <sup>P</sup> < 0.06 "medium", η 2 <sup>P</sup> < 0.14 "large". Differences in estimated marginal means and their 95% confidence intervals were reported as absolute effect size, and Cohen's d with Hedges correction and their 95% confidence intervals were reported as standardized effect size. Cohen's d was interpreted as follows: |*d*| < 0.2 "negligible", |*d*| < 0.5 "small", |*d*| < 0.8 "medium", otherwise "large". Normality of residuals was assessed using the Shapiro–Wilk test and Q–Q plots, heteroscedasticity was assessed using the Breusch–Pagan test, and model performance was evaluated using Akaike information criterion and R<sup>2</sup> . All assumptions and performance functions were assessed using the package performance. Statistical significance was assumed when *p* < 0.05.

#### **3. Results**

There was a significant correlation between time and category in the VL with stiffness increases in the sub-elite (*p* = 0.001, Cohen's *d* = 0.88) and elite groups (*p* = 0.003, Cohen's *d* = 0.72), but not in the recreational group (*p* = 0.085) (Figure 2). There were no differences among categories in the RF (F(2, 32) = 0.7, *p* = 0.53), GL (F(2, 32) = 0.9, *p* = 0.41), and BF (F(2, 32) = 1, *p* = 0.39). Additionally, baseline stiffness was comparable between categories in all muscles. Stiffness increased over time in both knee extensors, RF (F(1, 32) = 31.9, *p* < 0.001, η 2 <sup>P</sup> = 0.5) and VL (F(1, 32) = 24.4, *p* < 0.001, η 2 <sup>P</sup> = 0.2), but no changes were observed in the knee flexors, GL (F(1, 32) = 0.2, *p* = 0.63, η 2 <sup>P</sup> = 0) and BF (F(1, 32) = 4.1, *p* = 0.052, η 2 <sup>P</sup> = 0.1).

**Figure 2.** Stiffness by group, muscle, and time. Dots indicate group means whereas vertical error bars indicate standard deviations for each mean.

#### **4. Discussion**

The purpose of this study was to analyze the passive muscle stiffness of the main muscles involved in the pedaling action in a group of 35 cyclists classified by performance level. Furthermore, the effect of an incremental exercise test until exhaustion on the variations of passive muscle stiffness was also studied to determine whether performance level has an effect on the post-effort muscular response. The main findings of this study could be highlighted as; (i) there were no differences in the resting passive muscle stiffness of the muscles involved in the pedaling action between cyclists categorized by performance level, (ii) an exposition to an incremental exercise test until exhaustion caused an increase in passive muscle stiffness of the knee extensor muscles regardless of the performance group without resulting in modifications of the knee flexor and ankle extensor muscles, and (iii) only the VL differed in its behavior when differentiating elite and sub-elite categories from recreational cyclists. Therefore, it could be determined that there were no significant differences in the passive muscular stiffness analyzed in a resting situation regardless of the level of the cyclist. Furthermore, the subjection to an incremental test until exhaustion only caused an increase in the stiffness of the knee extensors (RF and VL) with significant differences between performance levels (elite and sub-elite vs. recreational) only found

for the VL. To the best of the authors' knowledge, this has been the first study to compare cyclists' muscle stiffness both in a resting and fatigued situation. In addition, the conclusions reported in this manuscript offer information for cyclists, coaches, and medical staff that help to understand the internal behavior of the muscles and could be considered as a training response variable.

The muscular properties and their behavior during the pedaling action have been studied in the scientific literature [34–36]. However, muscle stiffness and especially its variations among cyclists of variable performance levels have been scarcely studied [17,18]. The results obtained in this study differ from those reported by other authors. No significant differences between the analyzed groups (recreational, sub-elite, and elite) or in relation to other parameters such as age, height, or fat percentage in any of the studied muscles (RF, VL, BF, and GL) were obtained. Contrarily, recent studies reported differences in stiffness between performance levels in other sport disciplines. Pruyn et al. analyzed muscle stiffness in netball players classified as elite, sub-elite, and recreational, and reported significant differences between groups (*p* = 0.018). In addition, they concluded by stating that muscular stiffness could be a characteristic that could contribute to a player's ability to physically perform at an elite level. They also provided an explanation for the high injury rates at elite levels of performance by associating greater levels of stiffness with a higher injury rate [37]. Additionally, Kalkhoven et al. established a relationship between greater muscular stiffness and higher performance in a group of soccer players, highlighting the importance of high stiffness and its contribution to better athletic performance [38].

Regarding cycling, the studies performed by Wastford et al. and Uchiyama et al. should be highlighted. Both determined the importance of high stiffness levels for sprint specialties [15,16]. Uchiyama et al. established a mean value of 186–626 N/m in the VL and determined that this value is proportional to the workload and the power developed by the cyclist [16]. Finally, in the only previous study that analyzed cycling stiffness through myometry, higher values were reported for the knee extensor muscles (VL and RF) in sprinters than in less powerful riders. Our results showed that, in a resting situation, the stiffness characteristics are not associated with a typical endurance performance parameter such as VO2max. This finding could be explained because previous studies have analyzed sport disciplines in which the speed component is key to performance [7,37]. This finding does not occur in road cycling, which is a non-impactful discipline characterized by a continuous cyclical movement through coordinated submaximal contractions of the muscles involved in the pedaling action, characteristics that may explain these results. In this study there were no differences in muscle stiffness regardless of the VO2max (348 ± 55 N/m–433 ± 115 N/m). It should be taken into consideration that, despite the lack of statistically significant differences, greater stiffness was observed in those groups with greater aerobic capacity (greater VO2max) in the analysis of the VL (recreational; 348 ± 55, sub-elite; 404 ± 85, and elite; 433 ± 115). This increase was not observed for the rest of the analyzed muscles: RF, BF, and GL.

The literature regarding post-effort stiffness in endurance sports is scarce. Both studies by García-Manso et al. [13] and Andonian et al. [14] on long-distance events (Ironman and ultra-marathon, respectively) determined a clear decrease in contractile capacity and a decrease in the stiffness of the quadricep muscles. The results reported in our work showed an increase in the muscle tone of the knee extensors (RF and VL), with significant differences between both muscle groups (*p* < 0.001). Contrarily to what was reported by previous authors, this increase could be explained because the time until the assessment was clearly different in these studies: 4h in the case of the Ironman and days in the case of the ultra-marathon. In our study, muscle stiffness was tested immediately after the incremental test, which may be considered as a relevant factor that may influence the results.

Our results match those obtained by Silva et al., who determined that the RF and VL were the muscles with the highest activation rates during the pedaling action. Regarding the antagonist muscle (BF), a lower but longer total activation was observed [39]. Two years later, the same authors performed a similar analysis, this time with more muscle groups that were analyzed after an incremental test until exhaustion [40]. Again, activation of both RF and VL increased together with some parts of the hamstring muscles (long head of the BF, semitendinosus, and semimembranosus), while there was no activation in the short portion of the BF. The results of our work determined that the BF did not suffer significant differences in stiffness after undergoing a situation of induced fatigue (*p* = 0.052). This could be due to the fact that most of the power during the pedaling action is produced by the RF and VL and to a lesser degree by the BF. This aspect is closely linked to the elevation of post-effort muscle stiffness [15,41]. Additionally, it should be considered that the technique used for the analysis of activation in these studies was EMG and not myometry, an aspect that could influence the results.

In relation to the GL, only three studies have studied the muscular stiffness of this muscle through myometry [20,35,42]. The participation and activation of this muscle in the pedaling action is indisputable, but our results determined that there were no changes in stiffness in the post-effort situation (*p* = 0.63), an aspect that contradicts the results of some studies that highlight the importance of this muscle group involved in flexion and extension of the ankle. Pruyn et al. studied the relationship between muscle stiffness using myometry and variables related to performance in different modalities of team sports and highlighted the importance of enhancing the muscle group composed by the gastrocnemius, soleus, and Achilles tendon in order to achieve success in these sports modalities [42]. Again, the disciplines analyzed to reach this conclusion were based on short, high-intensity motor actions and not cyclical actions composed of submaximal muscle contractions as occurs in a sport such as cycling.

Finally, the correlation between the moment of measurement (pre and post effort) and the study category should be highlighted. In this case, cyclists in the elite and sub-elite groups presented significant differences in the stiffness values of the VL muscle compared to the recreational group. This finding is related to what was reported by Ditroilo et al., who found that cyclists with higher baseline stiffness levels presented greater reductions in muscle stiffness after fatiguing [18]. This would explain why the lower-level group (lower stiffness and lower VO2max) presented a behavior that did not match what was seen in the higher-level group. This discrepancy between riders characterized by different fitness levels may suggest that the stiffness control of the knee extensor muscles could be useful as a possible reference for functional tests in the periodic evaluation of cyclists.

Despite the fact that the study sample used in this work covered very high levels of performance, the recreational group presented high basal aerobic levels (VO2max = 49.9 ± 3.1 mL/kg/min), and this could be a limitation that had an impact on the results. In the same way, an incremental test generates maximum aerobic metabolic stimulation but does not induce the same levels of structural fatigue. Future lines of research should not only use groups characterized by lower performance levels but also control groups that would allow a global vision of the muscular stiffness. In addition, the behavior of the muscle should be analyzed in more fatiguing situations such as stage races while, at the same time, considering the specialty of the cyclist. Given that the stiffness values of the VL muscle differed between recreational, elite, and sub-elite cyclists, future research should assess whether this marker could be used to differentiate cyclists of varying fitness levels and its potential applicability for the monitoring of training load.

#### **5. Conclusions**

The results of this study suggest that there are no differences in the passive muscle stiffness of the muscles involved in the pedaling action between cyclists categorized by performance level. Exposition to an incremental exercise test until exhaustion caused an increase in passive muscle stiffness of the knee extensor muscles, regardless of the performance group, without resulting in modifications of the knee flexor and ankle extensor muscles. Only the VL differed in its behavior when differentiating elite and sub-elite categories from recreational cyclists.

**Author Contributions:** I.L.-L., R.C.-S., F.C. and S.S. were involved in conceptualizing and design this study. I.L.-L. and S.S. were involved in the data collection. R.C.-S. carried out the statistics. All authors were involved in manuscript writing (review and editing) and supervised this research study. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research did not receive any external funding.

**Institutional Review Board Statement:** The study followed the ethical guidelines of the 2013 Declaration of Helsinki and received approval from the Research Ethics Committee of the autonomous region of Aragon, Spain (the approval code is PI19/447, approved on 4 December 2019).

**Informed Consent Statement:** All participants signed an informed consent document.

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

#### **References**


MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Medicina* Editorial Office E-mail: medicina@mdpi.com www.mdpi.com/journal/medicina

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18

www.mdpi.com ISBN 978-3-0365-3868-6