*Article* **Weekly Training Load across a Standard Microcycle in a Sub-Elite Youth Football Academy: A Comparison between Starters and Non-Starters**

**José E. Teixeira 1,2,\* , Luís Branquinho 1,3 , Ricardo Ferraz 1,4 , Miguel Leal <sup>3</sup> , António J. Silva 1,5 , Tiago M. Barbosa 1,2 , António M. Monteiro 1,2 and Pedro Forte 1,2,3**


**Citation:** Teixeira, J.E.; Branquinho, L.; Ferraz, R.; Leal, M.; Silva, A.J.; Barbosa, T.M.; Monteiro, A.M.; Forte, P. Weekly Training Load across a Standard Microcycle in a Sub-Elite Youth Football Academy: A Comparison between Starters and Non-Starters. *Int. J. Environ. Res. Public Health* **2022**, *19*, 11611. https://doi.org/10.3390/ ijerph191811611

Academic Editors: Paul B. Tchounwou and Filipe Manuel Clemente

Received: 20 July 2022 Accepted: 13 September 2022 Published: 15 September 2022

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

**Copyright:** © 2022 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/).

**Abstract:** Compensatory training sessions have been highlighted as useful strategies to solve the differential weekly training load between the players' starting status. However, the influence of the players' starting status is still understudied in sub-elite youth football. Thus, the aim of this study was to compare the weekly training load on a standard microcycle in starters and non-starters of a sub-elite youth football academy. The weekly training load of 60 young sub-elite football players was monitored during a 6-week period using an 18 Hz global positioning system (GPS), 1 Hz telemetry heart rate, rating of perceived exertion (RPE), and total quality recovery (TQR). The total distance (TD) covered presented a significant difference between starters and non-starters with a moderate effect (*t* = −2.38, ∆ = −428.03 m, *p* = 0.018, *d* = 0.26). Training volume was higher in non-starters than in starter players (TDStarters = 5105.53 ± 1684.22 vs. TDNon-starters = 5533.56 ± 1549.26 m). Significant interactive effects were found between a player's starting status, playing time, and session duration in overall training load variables for within (F = 140.46; η <sup>2</sup> = 0.85; *p* < 0.001) and between-subjects (F = 11.63 to 160.70; η <sup>2</sup> = 0.05 to 0.76; *p* < 0.001). The player's starting status seems to only influence the training volume in sub-elite youth football, unless one considers the covariance of the playing time and session duration. Consequently, coaches should prioritize complementary training to equalize training volume and emphasize similar practice opportunities for non-starters. Future studies should evaluate the gap between training and match load, measuring the impact of recovery and compensatory sessions.

**Keywords:** workload; recovery; starting status; periodization; youth

### **1. Introduction**

Training load monitoring has been widely reported in youth football research [1,2]. Continuous training monitoring allows the measurement of the players' physical and physiological demands, allowing them to express their changes in performance and wellbeing [3,4]. Currently, analyzing and monitoring the weekly training load has become faster and easier to use due to advancements in tracking system applications [5,6]. Thus, the training representation and the game model can be quickly individually tailored through training load monitoring strategies [7,8]. Although most of the evidence has been produced in elite youth football, recently some studies have applied training load strategies in sub-elite cohorts [9–11]. Load variation over a standard microcycle in sub-elite football players seems to be influenced by week type, player's starting status, playing position, training mode, maturation status, and match-related contextual variables [1,11]. Previous

studies have reported a high intra-week variation with a low inter-week variation across a standard microcycle in a sub-elite youth football team [10]. When comparing elite and sub-elite football contexts, several differences have been reported in training intensity and patterns [10,11]. However, the player's starting status is still poorly studied in sub-elite youth training, with few reports in elite contexts [12–15].

For these reasons, the need inevitably arises to make an adjustment to the training loads of starters and non-starters, as after a game some players may need a complete rest period [16] or regeneration [17], while others must follow their normal training schedule [16,17] or have complete compensatory sessions [18]. In this regard, a recent study indicates that it may be beneficial to use small-sided games (SSG) to control the imposed training load. In fact, even though players can perform the same type of SSG format, there seems to be evidence that the choice of training method (i.e., fractional or continuous) and recovery time between repetitions with the use of the fractional method results in increases and decreases in imposed training loads, respectively [17–19]. Based on the results of these authors, starters should perform continuous SSG formats to decrease training load responses, while non-starters should perform fractional formats with short recovery periods to increase training load responses, thus compensating for the difference in game load between players (compensatory training) during the weekly training microcycle [17]. In this way, SSG can be seen as a powerful tool to ensure that starter and non-starter players achieve the goals set by the coach for the training session (e.g., distances covered, different speed zones, accelerations, decelerations, heart rate among others) [1,17].

However, considering the above differences in competitive levels, it is important to determine the main contributing factors that influence the training load management [19,20]. From a long-term development perspective, managing physical qualities is an important factor in improving a player's future sporting career [21,22]. Load discrepancies based on starting status may require compensatory training sessions or competitive breaks optimization periods [23,24]. In professional football, Anderson et al. [13] described that the total activity volume (i.e., training and match load), as well as the total distance covered, were not different between starters, fringe players, and nonstarters, while Los Arcos et al. [14] stated that the match load was solely responsible for a higher weekly training load in starters compared to non-starters. Dalen and Lorås [12] reported a large amount of match-related high-speed running and sprint distances across the weekly training schedule for elite young football players. Therefore, the present research aims to examine the evidence-based training load and determine any similarities with the training of sub-elite youth football players. Thus, the main purpose of this study was to compare the weekly training load across a standard microcycle in starters and non-starters of a sub-elite youth football academy.

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

### *2.1. Participants and Study Design*

Table 1 presents the baseline characteristics of the subsample of 60 male football players from a sub-elite Portuguese football academy. A total of 60 young football players aged between 13 and 20 years were analyzed in this prospective, observational, and crosssectional study. The daily training load was continuously monitored during a 6-week period of the 2019–2020 competitive season. The training data corresponded to a total of 18 training sessions and 324 observation cases (i.e., starters and non-starters with 164 and 160 observations, respectively).

All participants were informed of the aims and risks of the research. The study only includes players whose legal guardian/next of kin had signed the informed consent to participate. The present research was conducted in accordance with the ethical standards of the Declaration of Helsinki. The experimental approach was approved and followed by the local Ethical Committee from the University of Trás-os-Montes e Alto Douro (3379-5002PA67807).


**Table 1.** Description of the participants' subsamples according to the player's starting status.

### *2.2. Eligibility Criteria for Training Data*

The eligibility for training data was based on previous studies in sub-elite youth football [10,11] considering the following inclusion criteria: (a) young football players aged between 13 and 20 years old [1]; (b) at least five years of competitive experience in football [21]; (c) training files containing at least 35 consecutive minutes of playing time on the pitch [25]; (d) training data considered a competitive one-game per week schedule and complete full training sessions three times a week (~90 min) [10,11]. The exclusion criteria were: (a) total or partial absence from training due to data collection errors, injury events, rehabilitation sessions, individual training sessions, early withdrawal, and/or missing training; (b) football players aged under 13 or over 20 years; (c) the goalkeeper participated in the training session but was excluded from the analysis [1]. The exclusion criteria resulted in the elimination of 36 observation cases.

The players' starting status was divided into starters (i.e., started the game at least 55% of the games) and non-starters (i.e., started in less than 55% of the games) [13,26]. The average playing time was 73.82 ± 12.08 and 24.06 ± 9.67 min for starters and non-starters, respectively. The number of observations was adjusted by age group, specifically under 15 (U15), under 17 (U17), and under 19 (U19) [10,11]. The number of observations in weekly training data for each age was: U15 (*n* = 102), U17 (*n* = 99), and U19 (*n* = 120). The microcycle included three training sessions per week (~90 min) with the following "match day minus format" (MD): MD-3 (Tuesday), MD-2 (Wednesday), and MD-1 (Friday) [7,8]. The number of observations in weekly training data for each age was: MD-3 (*n* = 41), MD-2 (*n* = 38), and MD-1 (*n* = 44). The average training session consisted of 18 players with a training session and all age groups were trained on an outdoor pitch with official dimensions (FIFA standard; 100 × 70 m). The training sessions were performed on synthetic turf pitches, from 10:00 a.m. to 8:00 p.m., and with similar environmental conditions (14–20 ◦C; relative humidity 52–66%) [10,11].

### *2.3. Weekly Training Schedule*

The sampled training sessions were categorized according to a specific focus, following the discussion with the coaching staff. All sampled training sessions started with a standard warm-up with low-intensity running, dynamic stretching for main locomotive lower limb muscles, technical actions, and ball possession. The overview of weekly training was potentially variable across categories, such as different training modes with an emphasis on game-based situations and sport-specific skills for football-specific exercises [27,28]. The typical weekly training schedule was categorized based on a typical training microcycle published on youth football [29,30].

The MD-3 (Tuesday) highlighted the recovery and technical skills with an emphasis on individual and group tactical actions by 1v1 to 6v6 small- and medium-sized games (SSG/MSG) (physiological set: 75–80% HRmax). The MD-2 (Wednesday) focused on the sectorial and collective tactical actions of the game model as training containing the use of large sided games (LSG) (i.e., 7v7 to 10v10) and simulated games (i.e., 11v11) with a physiological set of 75–80% HRmax. The MD-1 (Friday) emphasized goal-scoring situations and tactical schemes (i.e., corners, free-kicks, penalty kicks) (physiological set: 85–90% HRmax).

### *2.4. Procedures*

The young sub-elite football players were monitored using a portable GPS throughout the whole training session duration (STATSports Apex®, Northern Ireland) [10,11]. The GPS device provides raw position velocity and distance at 18 Hz sampling frequencies, including an accelerometer (100 Hz), magnetometer (10 Hz), and gyroscope (100 Hz). Each player wore the micro-tech inner mini pocket of a custom-made vest supplied by the manufacturer, which was placed on the upper back between the two shoulder blades. All devices were activated 30 min prior to training data collection to allow clear and acceptable reception of the satellite signal. Respecting the optimal signal for the measurement of human movement, the match data considered eight available satellite signals as a minimum for the observations [31]. The validity and reliability of the global navigation satellite systems (GNSS) were guaranteed as the GPS has been well established in the literature [31–33]. The current variables and thresholds should consider a small error of around 1–2% reported in the 10 Hz STATSports Apex® units [31].

### *2.5. Training Load Measures*

### 2.5.1. External Training Load

The external training loads were obtained with time–motion data: total distance (TD) covered (m), average speed (AvS), maximum speed (SPR) (m/s), relative high-speed running distance (rHSR) (m), high metabolic load distance (HMLD) (m), sprinting distance (SPR) (m), dynamic stress load (DSL) (a.u.), number of accelerations (ACC), and number of decelerations (DEC). The number and duration of sprints were also measured (SPR\_D and SPR\_N, respectively (m)). The GPS software provided information only on the locomotor categories above 5.50 m/s: rHSR (5.5–6.97 km·h −1 ) and SPR (>6.97 km·h −1 ). The sprints were measured by the number and average sprint distance (m). The HMLD is a metabolic variable defined as the distance in meters covered by a player when the metabolic power exceeds 25.5 W·kg−<sup>1</sup> . HMLD variables include all high-speed running, accelerations, and decelerations above 3 m/s m·s −2 [31–33]. Both acceleration variables (ACC/DEC) considered the number of accelerations and decelerations performed at maximum intensity (>3 and <3 m/s, respectively). The DSL variable was evaluated by a 100 Hz triaxial accelerometer integrated into the GPS device. The sum of the accelerations is presented in the three orthogonal axes of movement (X, Y, and Z planes) in arbitrary units (a.u.) [34]. The high-intensity activity thresholds were adapted from previous studies [1,2].

### 2.5.2. Internal Training Load

### Heart Rate–Based Measures

Heart rate was recorded by a 1 Hz short-range telemetry system GARMIM TM HR band (International Inc., Olathe, KS, USA). Maximum heart rate (HRmax), average heart rate (AvHR), and percentage of HRmax (%HRmax) values were considered for analysis [35,36]. Training impulse was obtained by Akubat TRIMP [37], reporting a team TRIMP whose equation is based on individual data from the players' TRIMP; however, it was used to calculate the internal load for each player as: Akubat TRIMP = Training duration <sup>×</sup> 0.2053e3.5179x, among which the HRratio is the same in Banisters TRIMP [1], e = Napierian logarithms, 3.5179 is the e exponent, and x = HRratio [37]. HRmax was obtained by the Yo Yo intermittent recovery test level 1 (YYIR1) [38].

### Perceived Exertion and Recovery

The perceived exertion was measured using the 15-point Portuguese Borg Rating of Perceived Exertion 6–20 Scale (Borg RPE 6–20) [39]. The sRPE was obtained by multiplying the total duration of training sessions for each individual RPE score (sRPE = RPE × session duration) following a scale from 6 to 20 [40]. To monitor recovery, each player was asked to report the total quality recovery (TQR) score on a scale from 6 to 20. This scale was proposed by Kenttä and Hassmén [41] to measure the athletes' recovery perceptions. RPE and TQR were individually collected approximately 30 min before and after each training session, respectively. Players

were already familiarized with the procedures and the perceived data were collected using Microsoft Excel® spreadsheet (Microsoft Corporation, Redmond, WA, USA). Previous research has included both scales to examine perceived stress and fatigue in youth football [10,11].

### *2.6. Statistical Analysis*

Robust estimates of a 95% confidence interval (CI) and data heteroscedasticity were calculated using randomly 1000 bootstrap samples [11,42]. Data are presented as the mean ± standard deviation (SD), mean differences (∆) are presented in absolute values, and statistical significance was set at *p* < 0.05. Differences in the players' starting status were tested with an independent sample *t*-test [43]. Effect sizes (ES) were calculated based on Cohen's *d* and classified as: 0.2, trivial; 0.6, small; 1.2, large; and >2.0, very large [42,43]. A repeated-measure ANOVA was applied to compare the differences and interactive effects between playing time, session duration, and player's starting status in the weekly training load [44,45]. Data sphericity was checked by Mauchly's statistic, and where violated, a Greenhouse–Geiser adjustment was applied. For ANOVA, the ES was computed by the eta square (η 2 ) and interpreted as: 0 < η <sup>2</sup> <sup>≤</sup> 0.04, without effect; 0.04 < <sup>η</sup> <sup>2</sup> <sup>≤</sup> 0.25, minimum; 0.25 < η <sup>2</sup> <sup>≤</sup> 0.64, moderate; and <sup>η</sup> <sup>2</sup> > 0.64, strong [46,47]. A comparison of data visualization between starters and non-starters was performed by a violin diagram with a boxplot element (ggplot2). All statistical analyses and data visualization were conducted using JASP software (JASP Team, 2019; version 0.16.3, jasp-stats.org) [43].

### **3. Results**

*Weekly Training Load According to the Player's Starting Status*

The descriptive statistics of weekly training load according to the player's starting status are presented in Table 2.


**Table 2.** Mean weekly training load according to the player's starting status.

Abbreviations: ACC—acceleration; AvS—average speed; DEC—deceleration; HMLD—high metabolic load distance; RPE—ratings of perceived exertion; SPR—sprint distance; SPR\_N—number of sprints; SPR\_D—distance covered at sprinting; sRPE—session ratings of perceived exertion; TD—total distance; TQR—total quality recovery.

Table 3 presents the mean comparison between starters and non-starters for external and internal training loads. Only the TD covered presented a significant difference with a moderate effect when comparing between the player's starting status (*t* = −2.38, ∆ = −428.03 m, *p* = 0.018, *d* = 0.26). Training volume was higher for non-starters than starter players (TDStarters = 5105.53 ± 1684.22 vs. TDNon-starters = 5105.53 ± 1684.22 m). Neither the measures of external training intensity nor the internal training load showed significant differences. However, the high intensity showed a trend towards higher values in non-starters.


**Table 3.** Mean differences between starters and non-starters in the weekly training load.

Abbreviations: ∆—mean differences; ACC—accelerations; ALL—overall independent position group; AvS—average speed; bpm—beat per minute; CD—central defenders; CM—central midfielders; DEC—decelerations; FB—fullbacks; FW—forwards; rHSR—relative high speed running; SPR—sprints; TD—total distance; WM—wide midfielders.

When considering the playing time and session duration as co-variables, to compare the weekly training load in starters and non-starters, there were significant interactive effects between players' starting status, playing time, and session duration in overall training load variables, either for within-subjects (F = 140.46; η <sup>2</sup> = 0.85; *p* < 0.001) or for between-subjects (F = 11.63 to 160.70; η <sup>2</sup> = 0.05 to 0.76; *p* < 0.001). Figure 1 shows the comparison between starters and non-starters for each training load measure.

**Figure 1.** Comparison between starters and non-starters for each training load measure. Note: "Starters" coded 1 (red graph) and "non-starters" coded 2 (green graph).

#### **Figure 1.** Comparison between starters and non-starters for each training load measure. Note: "Starters" coded 1 (red graph) and "non-starters" coded 2 (green graph). **4. Discussion**

**4. Discussion**  The main objective of this study was to compare the weekly training load across a standard microcycle in starters and non-starters of a sub-elite youth football academy. In general, the presented data suggested a trend towards a higher weekly training load in non-starting football players. Additionally, the external and internal training intensity did not seem to differ between the starting status of sub-elite youth football players. However, The main objective of this study was to compare the weekly training load across a standard microcycle in starters and non-starters of a sub-elite youth football academy. In general, the presented data suggested a trend towards a higher weekly training load in non-starting football players. Additionally, the external and internal training intensity did not seem to differ between the starting status of sub-elite youth football players. However, when considering the co-variance of the playing time and session duration, a significant interactive effect between the players' starting status, playing time, and session was reported in the overall training load variables.

when considering the co-variance of the playing time and session duration, a significant interactive effect between the players' starting status, playing time, and session was reported in the overall training load variables. In this study, only the TD covered seems to be influenced by the player's starting status in the young sub-elite, with a higher training volume for non-starters compared to starters (moderate effect). A possible explanation may be that coaches tend to prioritize complementary training to equalize training volume and emphasize similar practice opportunities for non-starters [23,24]. The fact that this sub-elite academy of training football only trains three times a week may represent that one of them might represent recovery training for the starters and compensatory training for the non-starters. The current findings are contrary to the evidence produced on the influence of the player's starting status for elite youth training. In youth elite football, Dalen and Lorås [12] determined a higher In this study, only the TD covered seems to be influenced by the player's starting status in the young sub-elite, with a higher training volume for non-starters compared to starters (moderate effect). A possible explanation may be that coaches tend to prioritize complementary training to equalize training volume and emphasize similar practice opportunities for non-starters [23,24]. The fact that this sub-elite academy of training football only trains three times a week may represent that one of them might represent recovery training for the starters and compensatory training for the non-starters. The current findings are contrary to the evidence produced on the influence of the player's starting status for elite youth training. In youth elite football, Dalen and Lorås [12] determined a higher average weekly physical load for starters than non-starters in total covered distance, Banister's TRIMP, accelerations, and sprints. Furthermore, starters completed more moderate and high-intensity running than non-starters and fringe players in professional football [13].

Both training load analyses were performed during the in-season phase as in the present study [12,13]. On the contrary, this study determined that the non-starters covered more distance across the standard microcycle than starters. Current research also suggests a trend towards high-intensity activity as current training data showed a tendency towards higher values in non-starters, specifically for DEC, HSR, and SPR. The weekly training load disparities between elite and sub-elite football players are due to expertise level, periodization strategy, and training content [48,49], considering that it is possible that shorter training duration in sub-elite contexts may lead coaches to prioritize equity of practice opportunities for non-starters [48]. Otherwise, the intra- and inter-individual variation training load may influence the perceived exertion, pacing strategies, and high-intensity demands [11]. In addition, previous studies have demonstrated that non-starter players tend to have higher training workloads, which may result in overreaching, overtraining syndrome, and poor performance [44,45]. This evidence may also be due, in part, to the influence of maturational and motor development factors on the weekly training load [10,11]. Most importantly, the weekly training load across a standard microcycle should consider the co-variance of the playing time and session duration. This is because a non-starter may have 45 min, as well as a starter, since the players' starting statuses were based on the percentage of started matches and not on the playing time [13,26]. However, this evidence moves in the same direction as the weekly in-season training load verified in professional football players by Los Arcos et al. [14]. According to the study by Los Arcos, although a greater tendency towards a higher perceived exertion-based load for the starters was observed, only the match load was identified as a major factor contributing to a higher weekly training load. In the present study, the perceived exertion tended to be higher for starters than non-starters, for RPE, sRPE, and TQR. Previous studies have demonstrated that the perceived exertion does not seem to show differences either in age group or in maturity status [11]. Given this, the same assumptions seem to occur when considering the player's starting status as an influential factor in the accumulated training load [1]. All HR-based measures showed no statistical differences between starters and non-starters. However, similar to external training intensity, internal training intensity tends to be higher in non-starters. More specifically, non-starters have higher values for HRmax and %HRmax. Teixeira et al. [11] described higher HRmax and Akubat TRIMP in U17, as well as %HRmax, RPE, and sRPE in U15 sub-elite football players. The current weekly training load showed no differences for Akubat TRIMP between starters and non-starters. Although HR-based measures continue to be useful for training load monitoring, the limitations of measuring high-intensity movements are highly dependent on anaerobic components that have been widely described in the literature [1,2]. The standardization of the application of TRIMP methods to youth sub-elite football players should be considered to alleviate these problems [37]. Additionally, there is a need to reduce the dimensionality of the biomechanical and physiological datasets for a better understanding of the training load [11].

The current study presents some limitations that should be taken into consideration when interpreting and extending the results. First, the training load analysis included only one sub-elite football academy, so the applicability of the results must consider this specificity. Second, quantifying a weekly training load across a standard microcycle should also consider other influencing factors such as periodization structure and match-related contextual factors [10,11,50]. However, the current analysis did not include match data and, consequently, training and match load relationships [1]. The difference between recovery and compensatory sessions from other training days was also not analyzed [10]. Moreover, the training load was extracted from a complete training session, so that in the future the different training exercises should be subdivided to assess the task constraints and modality (i.e., fractional or continuous) such as SSG, high-intensity interval training (HIIT), and simulated game situations [1,51]. Pacing strategies and collective behavior should be considered in future research when analyzing the role of the starting status in match load [20,26,49]. In addition, future research should consider the relationship between compensatory training sessions with match load in youth sub-elite football, as this

is an emerging research topic that has not yet been explored in sub-elite training contexts. Additionally, it is still necessary to compare how the behavior of sub-elite and elite football players differs in specific training drills and constrained tasks [1,10,11]. The lack of access to raw positional data made it challenging to perform the fragmented analysis of the entire training session [49]; therefore, future research should focus on physical, physiological, and technical–tactical analysis with an emphasis on comparing starters and non-starters [49,51]. Hence, more analyses are needed for this purpose with a broader follow-up, given the small sample and size of this prospective, cross-sectional, and observational study design. Research on the weekly training load with an integrative performance perspective should also be considered, as key technical and tactical indicators were not explored in this analysis [49].

### **5. Conclusions**

The current research suggests a trend toward a higher weekly training load in nonstarters, contrary to the published literature to date. The player's starting status only seems to influence the training volume in sub-elite youth football, unless the covariance of the playing time and session duration are considered. Thus, coaches seem to prioritize complementary training to equalize training volume and emphasize similar practice opportunities for non-starters. Future studies should evaluate the gap between training and match load in this comparison between starters and non-starters.

**Author Contributions:** Conceptualization, J.E.T. and P.F.; data curation, J.E.T., L.B. and P.F.; formal analysis, R.F., M.L. and P.F.; funding acquisition, L.B. and P.F.; investigation, J.E.T. and L.B.; methodology, T.M.B., A.M.M. and P.F.; resources, J.E.T. and P.F.; software, J.E.T., L.B. and M.L.; supervision, A.J.S. and P.F.; validation, T.M.B., A.M.M. and P.F.; writing—original draft, J.E.T.; writing—review and editing, L.B., R.F., M.L., T.M.B., A.J.S., A.M.M. and P.F. All authors have read and agreed to the published version of the manuscript.

**Funding:** This project was supported by the National Funds through the FCT—Portuguese Foundation for Science and Technology (project UIDB04045/2021).

**Institutional Review Board Statement:** The study was conducted in accordance with the guidelines of the Declaration of Helsinki and approved by the institutional Ethical Committee from the University of Trás-os-Montes e Alto Douro (Doc2-CE-UTAD-2021).

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

**Data Availability Statement:** Data are available under request to the contact author.

**Acknowledgments:** The authors acknowledge all of the coaches and playing staff for cooperation during all collection procedures.

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

### **References**


### *Article* **The Immunological and Hormonal Responses to Competitive Match-Play in Elite Soccer Players**

**Ryland Morgans 1,\* , Patrick Orme <sup>2</sup> , Eduard Bezuglov <sup>1</sup> , Rocco Di Michele <sup>3</sup> and Alexandre Moreira <sup>4</sup>**


**Abstract:** This study aimed to examine the salivary immunoglobulin A (s-IgA) and salivary cortisol (s-Cort) responses to competitive matches in elite male soccer players. Data were collected for 19 players (mean ± SD, age: 26 ± 4 years; weight: 80.5 ± 8.1 kg; height: 1.83 ± 0.07 m; body-fat 10.8 ± 0.7%) from a Russian Premier League team throughout a 6-week period during the 2021–2022 season. Physical match loads were measured through an optical tracking system. s-IgA and s-Cort were assessed one day before each match (MD − 1), 60-min before kick-off, 30-min post-match, and 48-h post-match (MD + 2). At 60-min before kick-off, s-IgA values were lower than at MD − 1 (90% CI difference 15.7–71.3 µg/mL). Additionally, compared to 60-min before kick-off, s-IgA was higher at 30-min post-match (90% CI difference 1.8–57.8 µg/mL) and at MD + 2 (90% CI difference 5.4–60.5 µg/mL). At 30-min post-match, s-Cort was higher than at 60-min before kick-off (90% CI difference 4.84–7.86 ng/mL), while on MD + 2 s-Cort was higher than at 60-min before kick-off (90% CI difference 0.76–3.72 ng/mL). Mixed model regressions revealed that longer playing time and total distance covered, and higher number of high-intensity accelerations, involved smaller s-IgA differences between 30-min post-match and 60-min before kick-off, and between 60-min before kick-off and MD + 2. Additionally, greater high-intensity and sprint distances, and a higher number of high-intensity and maximal accelerations, involved smaller s-Cort differences between 60-min before kick-off and MD + 2. In conclusion, the present results demonstrate that using salivary monitoring combined with match load may be a useful tool to monitor individual mucosal immunity and hormonal responses to match-play and the subsequent recovery periods in elite soccer players.

**Keywords:** salivary cortisol; salivary immunoglobulin A; physical match performance; recovery; soccer

### **1. Introduction**

The physiological demands of soccer performance have been extensively researched over the past several decades [1]. It is widely accepted that undertaking ~90 min of a soccer match induces significant disruption to bodily homeostatic parameters. The impact that this has on various physiological processes in the hours and days following match-play has also been researched in detail [2,3].

Various methods have been employed within research settings in an effort to quantify the physiological impact following soccer match-play. These methods include assessment of neuromuscular function [4], blood sampling [5], subjective questionnaires [6] and saliva sampling [7]. While these methods have been used effectively to highlight relationships between the physiological status of soccer players and training and match demands, there is a need to fully understand the profile of the response to elite competitive soccer match-play. This further understanding may allow practitioners to individualize the schedule and program of players to ensure full recovery following match-play, reducing the

**Citation:** Morgans, R.; Orme, P.; Bezuglov, E.; Di Michele, R.; Moreira, A. The Immunological and Hormonal Responses to Competitive Match-Play in Elite Soccer Players. *Int. J. Environ. Res. Public Health* **2022**, *19*, 11784. https://doi.org/10.3390/ ijerph191811784

Academic Editor: Paul B. Tchounwou

Received: 31 August 2022 Accepted: 16 September 2022 Published: 18 September 2022

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

**Copyright:** © 2022 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/).

likelihood of injuries, and optimal physical preparation for upcoming matches to maximize subsequent performance.

From the aforementioned methods, saliva sampling methods have been employed to quickly screen players for stress and illness on a regular basis throughout the season [7–10]. Saliva sampling is a relatively simple and non-invasive method that provides practitioners with a variety of markers that can be used to understand players physiological status preand post-match. Previous research has outlined the use of salivary markers such as salivary immunoglobulin A (s-IgA) [7], cortisol (s-Cort) [11], and testosterone [12] in soccer players following match-play.

As outlined above, the stressors of soccer match performance result in disruption to the physiological status of players. Mortatti et al. [8] reported a decrease in s-IgA concentration, a marker of mucosal immunity, in elite U19 soccer players when regularly monitored in a series of seven matches over 20 days, which may leave players more susceptible to illness, specifically through upper respiratory tract infections. Indeed, Springham et al. [10] also identified a cross-season suppression of s-IgA in professional soccer players, which was related to players perceived fatigue, sleep quality and muscle soreness suggesting the need to adopt s-IgA monitoring to aid in the prescription of training load and recovery. Therefore, methods that may be able to provide practitioners with an objective understanding of immune system function, in particular for mucosal immunity, in the period following match-play may be able to minimize the number of training days lost to illness over the course of a season [13].

Cortisol is a steroid hormone, detectable in saliva [14], that reflects catabolic balance [15]. Previous research has reported acute increases in s-Cort post-match in a variety of athletic populations including soccer [15,16], rugby [17], and Australian Rules football (AFL) [18], and differing training methods [19], which may persist for between 24- and 75-h [15,16,20]. Soccer studies that have examined longitudinal s-Cort responses have reported elevated values during periods of increased workload [21] and a reduction in Testosterone: Cortisol ratio toward the end of the competitive season [22]. However, previous longitudinal investigations are limited by infrequent or missing data points [21,22], while studies with short sampling periods have failed to examine the effect of elite competitive match-play or quantified the relationship between physical match performance and objective immunological (s-IgA) and hormonal (s-Cort) markers during the post-match 48-h recovery period. Thus, the ability to accurately analyze acute player responses is diminished.

Morgans et al. [7] presented data that reported fluctuations in s-IgA to be sensitive to changes in the physical demands placed on soccer players as a result of changes in fixture scheduling at different time points across the season. Values for s-IgA were decreased during periods of condensed fixture schedules (2–3 matches per week) but returned to 'normal' baseline measures during regular fixture schedules (one match per week). Similar findings were presented by Mortatti et al. [8] when assessing changes in s-IgA during a period of congested fixtures (seven matches in 20 days). However, these authors found no change in s-Cort concentration during the same period. These authors also suggest that further investigation is required to better understand the potential relationship between s-Cort and the physical demands of elite soccer match-play.

Therefore, this unique investigation aims to examine the s-IgA and s-Cort responses to match-play of elite European soccer players across six competitive fixtures compared with baseline and pre-match values, and to compare if and how these responses differ between starters and non-starters. Furthermore, the study aims to quantify the relationship between physical match performance and objective immunological (s-IgA) and hormonal (s-Cort) markers during the post-match 48-h recovery period. It was hypothesized that elite soccer match-play would induce changes in s-IgA and s-Cort when compared with baseline and that these changes would be greater for starters versus non-starters.

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

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

This study examined 19 elite male soccer players from the same team over a 6-week period during the second phase of the season. The participants had been playing soccer for a minimum of 10 years. Thirteen of the players used in this investigation were members of their respected national teams. The sample was initially recruited based on squad selection across six league matches (home matches (*n* = 4), away fixtures (*n* = 2)) in the 2021–2022 season. The sample was further sub-divided into starting players (*n* = 10) and non-starting players (*n* = 9). Participant data were only included in the analyses as starting player when time spent on the field exceeded 45-min of the match. Players were considered for inclusion as starting player if they completed, based on the inclusion criterion of 45-min playing time, in three (50%) or more of the examined matches. During a regular week, samples were obtained one day before each match (MD − 1), 60-min before kick-off on match-day, 30-min post-match and 48-h post-match (two days (MD + 2)). All samples were collected prior to breakfast in the morning period (09.30–10.30 a.m.) 1-h pre-training except on match-day. In the six examined matches, kick- off time was 2.00 p.m. (*n* = 3), 4.30 p.m. (*n* = 1), and 7.00 p.m. (*n* = 2). Sample collection time on match-day varied due to the official start of the match but was consistently 60-min prior to kick-off. In addition to saliva assessment, all match performance data was collated for analysis. Except on match-day, all participants were in a fasted state and required to abstain from food and caffeine products for a minimum of 2-h prior to the collection of saliva, and all salivary samples were collected at the same time of day for all participants (09.30–10.30 a.m.) to minimize the residual effect of exercise and circadian variations.

### *2.2. Participants*

A total of 19 male outfield players (mean ± SD, age 26 ± 4 years; weight 80.5 ± 8.1 kg; height 1.83 ± 0.07 m; body-fat 10.8 ± 0.7%) were involved in the study. Players were classified by position and grouped accordingly: Center Defender (CD) *n* = 5, Wide Defender (WD) *n* = 3, Center Midfield (CM) *n* = 7, Wide Forward (WF) *n* = 2, and Center Forward (CF) *n* = 2. All data evolved as a result of employment in which players were routinely monitored over the course of the competitive season. Nevertheless, approval for the study from the club was obtained [23] and the study was performed in accordance with the Helsinki Declaration principles. Ethical approval was granted by the local Ethics Committee of Sechenov University (N 22-21 dated 12 December 2021). To ensure confidentiality, all data were anonymized before analysis. Participants were fully familiarized with the experimental procedures within this study due to the regular testing protocols implemented as part of the clubs' performance monitoring strategy. During the study, players were instructed to maintain normal daily food and water intake, and no additional dietary interventions were undertaken.

### *2.3. Procedures*

The study period included saliva sampling and all match performance across a 6-week phase of the 2021–2022 season. The training sessions performed during the investigation were representative of a typical training micro-cycle implemented within elite European soccer, involving a periodized training week encompassing low, moderate, and high intensity sessions leading to competitive match-play. No player reported a soft tissue injury, illness or infection during the data collection period.

### *2.4. Salivary Sampling*

Given that soccer match-play induces a reduction in s-IgA concentration that return to basal levels within 18-h [24], we reasoned that collection of samples 48-h post-match would allow us to ascertain the effects of the acute suppression in s-IgA concentration from that associated with more chronic levels of stress. The diurnal rhythm of cortisol typically sees the highest concentrations in early morning with decreases as the day progresses [25]. Thus, players provided saliva samples pre-breakfast approximately 60-min before training on MD − 1, 60-min before kick-off on match-day, 30-min post-match and pre-breakfast approximately 60-min on MD + 2.

Saliva samples were collected and analyzed from this cohort of players using the Soma OFC II collection kits in combination with real-time Lateral Flow Device (LFD), respectively. This method has been previously validated for oral fluid collection in the immunoassay of immunoglobulins in sports persons [26,27] and correlates well with other methods (enzyme-linked immunosorbent assay) adopted in the determination of s-IgA [9] and s-Cort [11,20,24]. In accordance with the manufacturer's guidelines, after thoroughly rinsing their mouths with water, un-stimulated saliva samples were obtained. Players were required to place an Oral Fluid Collector (OFC II; Soma Bioscience, Oxfordshire, UK) consisting of a synthetic polymer-based swab attached to a polypropylene volume adequacy indicator stem in their mouth. Participants were instructed to swallow any saliva present within the oral cavity before placing the collection device on top of the tongue. Once the OFC kits collect 0.5 ml (± 20%) of oral fluid (collection time typically in the range of 20–50-s), the volume adequacy indicator turned blue and the player then placed the swab into the buffer bottle. The bottle was then mixed by gentle inversion for a period of 1–2-min, and the collected sample was ready to be analyzed through an IgA/Cortisol Dual LFD and photometric LFD reader (Soma Bioscience, Wallingford, UK). For the LFD, two-to-three drops of saliva/buffer mix were added to the sample window of the LFD cassette. The liquid in turn then ran the length of the test strip through creating a control and test line visible in the test window. Scanning of the LFD took place 15-min after the sample was added, being a competitive assay, the test line intensity was inversely proportional to the s-IgA and s-Cort concentration in the sample analyzed. This method has been previously validated [26–28] against ELISA (r<sup>2</sup> = 0.78) in 208 samples collected from a cohort of English Premier League soccer players [28].

### *2.5. Physical Load*

League physical match performance data were collected using a two-camera optical tracking system (InStat, Moscow, Russia) that was installed to record and examine the technical and physical match performance during competitive league fixtures. The matches were filmed using two full HD, static cameras positioned on the centre line of the field, not less than 3-metres from the field and 7-metres in height. A consistent 25 Hz format was provided. Data were linearly interpolated to 50 Hz, smoothed using a 5-point moving average and then down-sampled to 10 Hz, which allowed analysis of all player actions with and without the ball [29]. The installation process, reliability, and validity of InStat have been previously reported [29]. Physical performance was analyzed using the InStat Analysis Software System and exported to the Microsoft Excel software for further analyses. InStat provided written permission to allow all match data to be used for research purposes. The physical match activity profile included: time on pitch (min); total distance covered (km); high intensity distance (km; total distance covered 5.5–7 m/s); sprint distance (km; total distance covered >7 m/s); number of high-intensity accelerations (peak speed 5.5–7 m/s); number of maximal accelerations (peak speed >7 m/s).

### *2.6. Statistical Analysis*

All data are presented as the mean ± SD. When appropriate, 90% confidence intervals (CI) were also shown. Data were analyzed with the software R, version 4.2.0 (R Foundation for Statistical Computing, Vienna, Austria). Linear mixed models, with random intercepts for individual players' and match IDs, were used to assess the differences between the mean s-IgA and s-Cort values at the examined time points (MD − 1, 60-min before kick-off, 30-min post-match, and MD + 2) in starters compared to non-starters. The sample 60-min before kick-off was taken as the reference category to which values of MD − 1, 30-min postmatch, and MD + 2 values were compared. Additionally, linear mixed-effect regressions with random intercept for players' and match IDs were performed to examine the effect of

playing time (min) and variables related to the match physical effort (distances covered and number of accelerations), on s-IgA and s-Cort, respectively. The s-IgA and s-Cort differences between post- (30-min after) and pre-match (60-min before), and the s-IgA and s-Cort differences between 48-h post- and pre-match, were taken as outcome variables. Effect sizes were calculated from the coefficients of linear mixed models as Cohen's d through the lme.dscore function from the EMAtools package [30]. The absolute d value was interpreted as very small (<0.2), small (0.2–0.5), medium (0.5–0.8), large (>0.8). For all analyses, statistical significance was set at *p* < 0.10 due to the relatively small number of examined matches. examine the effect of playing time (mins) and variables related to the match physical effort (distances covered and number of accelerations), on s-IgA and s-Cort, respectively. The s-IgA and s-Cort differences between post- (30-min after) and pre-match (60-min before), and the s-IgA and s-Cort differences between 48-h post- and pre-match, were taken as outcome variables. Effect sizes were calculated from the coefficients of linear mixed models as Cohen's d through the lme.dscore function from the EMAtools package [30]. The absolute d value was interpreted as very small (<0.2), small (0.2–0.5), medium (0.5–0.8), large (>0.8). For all analyses, statistical significance was set at *p* < 0.10 due to the relatively small number of examined matches.

Foundation for Statistical Computing, Vienna, Austria). Linear mixed models, with random intercepts for individual players' and match IDs, were used to assess the differences between the mean s-IgA and s-Cort values at the examined time points (MD − 1, 60-min before kick-off, 30-min post-match, and MD + 2) in starters compared to non-starters. The

− 1, 30-min post-match, and MD + 2 values were compared. Additionally, linear mixedeffect regressions with random intercept for players' and match IDs were performed to

*Int. J. Environ. Res. Public Health* **2022**, *19*, x 5 of 13

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

The mean and SD s-IgA at the examined time points are shown in Figure 1. Sixty minutes before kick-off, the mean s-IgA value was significantly (*p* = 0.0108) lower than MD − 1, with an estimated difference of 43.5 µg/mL (90% CI: 15.7 to 71.3; d = 0.26, small). Additionally, compared to 60-min pre-match, there was a significantly higher value of s-IgA 30-min post-match (*p* = 0.083; estimated difference 29.8 µg/mL (90% CI: 1.8 to 57.8; d = 0.17, very small) and 48-h post-match (*p* = 0.051; estimated difference 33.0 µg/mL (90% CI: 5.4 to 60.5; d = 0.19, very small). No significant differences were observed between starters and non-starters at any time point, and there was no significant group x time interaction (*p* > 0.10). The mean and SD s-IgA at the examined time points are shown in Figure 1. Sixty minutes before kick-off, the mean s-IgA value was significantly (*p* = 0.0108) lower than MD − 1, with an estimated difference of 43.5 μg/mL (90% CI: 15.7 to 71.3; d = 0.26, small). Additionally, compared to 60-min pre-match, there was a significantly higher value of s-IgA 30-min post-match (*p* = 0.083; estimated difference 29.8 μg/mL (90% CI: 1.8 to 57.8; d = 0.17, very small) and 48-h post-match (*p* = 0.051; estimated difference 33.0 μg/mL (90% CI: 5.4 to 60.5; d = 0.19, very small). No significant differences were observed between starters and non-starters at any time point, and there was no significant group x time interaction (*p* > 0.10).

**Figure 1.** Mean and SD values of s-IgA the day before the match (MD − 1), 60-min before kick-off (M − 60 min), 30-min post-match (M + 30 min), and 48-h post-match (MD + 2). **Figure 1.** Mean and SD values of s-IgA the day before the match (MD − 1), 60-min before kick-off (M − 60 min), 30-min post-match (M + 30 min), and 48-h post-match (MD + 2).

Figure 2 shows the mean and SD values of s-Cort at the four examined time points. There was no significant difference between MD − 1 and 60-min before kick-off (*p* = 0.118). At 30-min post-match, s-Cort was significantly (*p* < 0.001) higher than 60-min pre-match, with an estimated difference of 6.35 ng/mL (90% CI: 4.84 to 7.86; d = 0.68, medium), while at 48-h post-match, s-Cort showed a decrease though it was still slightly higher (*p* = 0.014) than 60-min before kick-off, with an estimated difference of 2.47 ng/mL (90% CI: 0.76 to 3.72; d = 0.25 small). No differences were observed between starters and non-starters, and no significant time x group interaction was observed (*p* > 0.10). Figure 2 shows the mean and SD values of s-Cort at the four examined time points. There was no significant difference between MD − 1 and 60-min before kick-off (*p* = 0.118). At 30-min post-match, s-Cort was significantly (*p* < 0.001) higher than 60-min pre-match, with an estimated difference of 6.35 ng/mL (90% CI: 4.84 to 7.86; d = 0.68, medium), while at 48-h post-match, s-Cort showed a decrease though it was still slightly higher (*p* = 0.014) than 60-min before kick-off, with an estimated difference of 2.47 ng/mL (90% CI: 0.76 to 3.72; d = 0.25 small). No differences were observed between starters and non-starters, and no significant time x group interaction was observed (*p* > 0.10).

Tables 1 and 2 shows the coefficients of fixed effects obtained with linear mixed model analysis with playing time and physical match performance variables as fixed factors, and individual values of s-IgA differences, 30-min post-match vs. 60-min before kick-off (Table 1), and 48-h post-match vs. 60-min before kick-off (Table 2), as outcome variables. These coefficients indicate the change in s-IgA differences post-match involved by a oneunit increase of the independent variable in that given match.

**Figure 2.** Mean and SD values of s-Cort the day before the match (MD − 1), 60-min before kick-off (M − 60 min), 30-min post-match (M + 30 min), and 48-h post-match (MD + 2). **Figure 2.** Mean and SD values of s-Cort the day before the match (MD − 1), 60-min before kick-off (M − 60 min), 30-min post-match (M + 30 min), and 48-h post-match (MD + 2).

Tables 1 and 2 shows the coefficients of fixed effects obtained with linear mixed model analysis with playing time and physical match performance variables as fixed fac-**Table 1.** Effects of playing time and physical performance on s-IgA differences calculated between 30-min post-match and 60-min before kick-off time points.


\* *p* < 0.10. TD = Total distance; CI: Confidence Interval.

48-h post-match, with d values ranging from medium to large (Tables 1 and 2). Additionally, greater high-intensity distance covered involved a smaller s-IgA difference between **Table 2.** Effects of playing time and physical performance on s-IgA differences calculated between 48-h post-match and 60-min before kick-off time points.

ations involved smaller s-IgA differences between 60-min before kick-off and 30-min or


TD (km) −6.82 (−12.36 to −1.31) 0.051 \* 0.61 \* *p* < 0.10. TD = Total distance; CI: Confidence Interval.

High-intensity distance (km) −64.91 (−128.34 to −1.16) 0.102 0.43 Sprint distance (km) 61.99 (−200.82 to 320.77) 0.697 0.09 Number of high-intensity accelerations <sup>−</sup>1.18 (−2.16 to 0.19) 0.057 \* 0.48 Number of maximal accelerations −0.42 (−4.89 to 3.98) 0.876 0.03 \* *p* < 0.10. TD = Total distance; CI: Confidence Interval. **Table 2.** Effects of playing time and physical performance on s-IgA differences calculated between 48-h post-match and 60-min before kick-off time points. A 1-min longer time on pitch involved a 0.74 µg/mL smaller 30-min post-match/60-min before kick-off difference, with a medium effect (Table 1), and a 1.32 µg/mL smaller 48-h post-match/60-min before kick-off s-IgA difference, with a medium effect (Table 2). Similarly, a greater total distance covered and a higher number of high-intensity accelerations involved smaller s-IgA differences between 60-min before kick-off and 30-min or 48-h postmatch, with d values ranging from medium to large (Tables 1 and 2). Additionally, greater high-intensity distance covered involved a smaller s-IgA difference between measurements taken 48-h post-match and 60-min before kick-off, with a medium effect (Table 2).

**Coefficient (90% CI)** *p***-Value Cohen's d** Playing time (mins) −1.32 (−2.18 to −0.45) 0.013 \* 0.80 TD (km) −12.61 (−20.20 to −4.95) 0.007 \* 0.81 The fixed effects obtained from linear mixed models, with time on pitch and physical match performance variables as fixed factors, and individual values of s-Cort differences as outcome variables are presented in Table 3 (30-min post-match vs. 60-min before kick-off difference) and Table 4 (48-h post-match vs. 60-min before kick-off difference).

High-intensity distance (km) −125.65 (−211.21 to −37.23) 0.018 \* 0.61 Sprint distance (km) −323.09 (−679.52 to 49.44) 0.145 0.33


**Table 3.** Effects of playing time and physical performance on s-Cort differences calculated between 30-min post-match and 60-min before kick-off time points.

TD = Total distance; CI: Confidence Interval.

**Table 4.** Effects of playing time and physical performance on s-Cort differences calculated between 48-h post-match and 60-min before kick-off time points.


\* *p* < 0.10. TD = Total distance; CI: Confidence Interval.

There was no significant effect of playing time, distances covered or the number of high-intensity or maximal accelerations on s-Cort differences between 30-min postmatch and 60-min before kick-off (all *p* > 0.10) (Table 3). Conversely, greater high-intensity and sprint distances, and a higher number of high-intensity and maximal accelerations, involved smaller s-Cort differences between 48-h post-match and 60-min before kick-off, with small effects (Table 4).

### **4. Discussion**

This investigation aimed to examine the s-IgA and s-Cort responses to match-play of elite soccer players across six competitive fixtures in the 2021–2022 season compared with baseline and pre-match values. Furthermore, the study aimed to quantify the relationship between physical match performance and objective immunological (s-IgA) and hormonal (s-Cort) markers during the post-match 48-h recovery period. One of the main findings of the present study was the significant though slight decrease in s-IgA concentration from MD − 1 to 60-min before kick-off. It is reasonable to suggest that this result is somewhat unexpected as the release of s-IgA is under strong neuroendocrine control [31], and the activation of the sympathetic nervous system associated with player's match preparation would, on the contrary, increase s-IgA concentration. Previously, it has been suggested that these mechanisms are responsible for the increases in s-IgA concentration induced by acute stress [32]. This result however, is unique in elite professional male soccer players and may suggest that psychological factors related to official match-play preparation may affect s-IgA concentration, and consequently, mucosal immune function. Moreira et al. [33], demonstrated in elite male volleyball players a significantly lower prematch s-IgA concentration for a final championship match compared with pre-match s-IgA values for a regular season match. This result suggests that players' perceived importance of the match affect s-IgA concentration, highlighting therefore, the role of psychological factors in modulating the mucosal immunity in team-sport athletes. Indeed, this result further indicates that monitoring resting s-IgA in team-sports athletes would provide valuable information regarding how athletes cope with competition induced stress.

Regarding coping with stress related to competitive match preparation, the present results reported lower s-IgA concentration 60-min before kick-off compared to MD − 1, which may be partly explained by the well-known differences in responses to acute stress

between active and passive coping strategies [34]. Bosch et al. [34] examined the acute immunological effects of two different laboratory stressors ("active coping" via a timepaced memory test and "passive coping" via a stressful video showing surgical operations). The results of the study showed that active coping led to increases in s-IgA concentration, while, passive coping induced a decrease in s-IgA concentration. Considering that the preparation for an official match may impose a significant psychological stress on teamsports athletes [33,35], it may therefore be inferred that the adoption of passive coping strategies before official match-play may negatively impact the mucosal immune function which in turn may increase the likelihood of upper respiratory tract infection occurrences. The present results in conjunction with the aforementioned data may possibly provide an opportunity for sport scientists and professionals working with soccer players to adopt active coping strategies during the preparation period for official soccer matches, and highlight the potential for the introduction of affective or positive emotional engagement. Further studies should focus on examining whether structured active coping tasks minimize the negative effect (i.e., decreasing s-IgA concentration) of the inherent stress associated with preparation for official match-play.

The current results also demonstrated an increasing trend in s-IgA concentration at 30-min and 48-h post-match, compared to 60-min before kick-off. These results suggest a short-term (acute) stress response induced enhancement of mucosal immune function [36]. Psychological and physiological stressors have been shown to stimulate biological stress. These responses are signals to cells and tissues, which express themselves as receptors for the released biological factors, leading therefore to the activation of all bodily systems, including the immune system. In contrast to chronic stress, that may lead to suppression or dysregulation of immune function, while impacting negatively the mucosal immunity [37], the present results suggest that the short-term stressors related to official soccer matchplay may induce enhancement of immune function in professional soccer players. This is a positive response which prepares athletes for the imposed challenges associated with competition. It is important to highlight that previous studies have shown that factors such as corticosterone and epinephrine, released due to the presence of a stressor, are mediators of a short-term stress induced immuno-enhancement, while a variety of studies have shown increases or no changes in s-IgA concentration from pre- to post-match in team-sport athletes [9,33,38], professional female soccer players [39], and professional male soccer players [40]. Previous studies in soccer players demonstrated that elevated levels of psycho-physiological stress may negatively affect the mucosal immune function, with decreases in s-IgA concentration across periods of congested fixtures or intensive training loads [7,8,12,41]. Considering our results in combination with the existing literature, it could be reasonable to suggest that the probability of observing no changes or even increases in s-IgA concentration is high for acute stress (i.e., from pre- to post-match), while on the other hand, the chronic effect of accumulated stress, notably, when performing successive matches in a short period of time, may negatively affect the mucosal immunity of players.

The design of the present study allowed the observation of s-IgA responses to actual physical match load that have not yet been demonstrated in official soccer matches with elite professional male players. Despite the observed trend to increase s-IgA concentration from pre-match to 30-min and 48-h post-match, it is notable that, when performing a higher workload, players seemed to present a slower return to their initial s-IgA concentration. The 1-min longer playing time on pitch produced a 0.74 µg/mL smaller 30-min post-match/60-min before kick-off difference and a 1.32 µg/mL smaller 48-h postmatch/60-min before kick-off s-IgA difference. Smaller s-IgA concentration differences between 60-min before kick-off and 30-min or 48-h post-match were also observed in association with greater total distance covered, and with a higher number of high-intensity accelerations. Additionally, greater high-intensity distance covered involved a smaller s-IgA difference between 48-h post-match and 60-min before kick-off. This unique and important finding of the present study suggests that an inverted-U/bell-shaped relationship may be observed between match-workload and the effects on mucosal immune function.

Thus, when performing higher workload, above a given threshold, players may be more prone to trivial increases or even reductions in s-IgA concentrations. In addition, this result may aid in explaining the increased likelihood of a suppressed effect from accumulated and successive match-play in s-IgA concentration, as this workload accumulation would affect plasma cells functions (immunoglobulin-secreting plasma cells) and the rate of IgA transcytosis across the epithelial cell. This result suggests a novel role for physical match workload monitoring and its impact on mucosal immunity in professional soccer players.

In relation to s-Cort, there was no significant difference between MD − 1 and 60-min before kick-off. This result suggests that the expected anticipatory stress response to match participation [42] did not occur. This finding might be associated with the high-level of the examined players and with their habitual lead-in process to cope with the pressure and anxiety involved in the period preceding the start of official matches. In this sense, van Paridon et al. [42] reported in their systematic review that the anticipatory stress response and cortisol reactivity, in both male and female athletes competing at international level, do not present a significant anticipatory cortisol response. Moreover, in earlier research, Alix-Sy et al. [43] despite showing a significant increase in s-Cort concentration at prematch compared to a non-training day in professional French soccer players, reported a significant positive association between unpleasant somatic emotions and cortisol. Indeed, Alix-Sy et al. [43] also demonstrated no differences in s-Cort between starters and nonstarters, as observed in the current study. Furthermore, it should be highlighted that in their study, the authors compared a non-training day with official matches, while in the present study, saliva collection occurred during habitual training sessions performed one day before matches. This difference may influence, at least in part, the present result of no change in s-Cort.

Considering these findings, we might suggest that the players evaluated in the present study did not show a s-Cort rise from MD − 1 to 60-min before kick-off possibly due to their positive evaluation of the potential match challenges, which in turn may be related to their perception of relative situational control and the non-decisive nature of regular season matches. Considering the results of the present study it may be suggested that due to the nature of the evaluated matches and the level of the assessed players, the s-Cort anticipatory responses (MD − 1 vs. 60-min before kick-off) indicated an optimal cognitive and behavioural player state to participate in the matches.

As expected, a significant increase in s-Cort from 60-min before kick-off to 30-min post-match was observed, while at 48-h post-match, s-Cort showed a decrease though still slightly higher than 60-min before kick-off, and no differences were observed between starters and non-starters. The increases in s-Cort reinforces that official soccer match-play induce significant psychophysiological stress likely related to physical demands associated with the volume and intensity of match-play, leading to increased secretion of s-Cort, as also reported in A-League [16] and intercollegiate soccer players [44]. It is important to highlight that the psychological factors involved in official match-play may play a role in this result. These results in conjunction allow us to infer that besides the well-known effect of increased s-Cort related to exercise stress, which represents per se a potent physiological stressor [45], the pressure of official match-play may be considered as an additional stress factor, possibly due to its social-evaluative task characteristics combined with other contextual factors inherent to sports competition as proposed by Arruda et al. [46].

Indeed, as demonstrated more recently by Rowell et al. [16], a substantial individual variability in s-Cort response to soccer match-play may be expected, including the responses within 48-h post-match. Furthermore, the psycho-physiological relationships and the impact of situational factors have been reported to influence cortisol responses to matchplay in soccer players [47]. Thus, the present results add to the literature and suggest that contextual factors other than being a starter or non-starter may influence the variability in players s-Cort responses. The uniqueness of the present study allowed us to examine the effect of match-load measures on s-Cort time-course responses. A novel and interesting finding of the present study was that the greater high-intensity distance, sprint distance, number of high-intensity and maximal accelerations performed, the smaller the s-Cort differences between 48-h post-match and 60-min before kick-off. This result suggests that performing a greater number of high-intensity actions during the match would increase the associated stress, that in turn may hinder s-Cort recovery to resting values. Moreover, the present results suggest that high-intensity distance, sprint distance, number of highintensity and maximal accelerations may be employed as reliable markers of individual external match-load inducing stress, and possibly predict catabolic state induced by matchplay, rather than dividing players into starters and non-starters groups. In addition, the findings highlight the need to monitor in conjunction with the individual external matchload and the s-Cort response of players to account for individual variability in recovery from match-play. The results also suggest that examining s-Cort responses from pre-match to 30-min post-match might not aid in observing true changes in s-Cort during the recovery time-course.

Despite the interesting findings of the current study, some limitations should be acknowledged. Firstly, our study only focused on one elite European professional soccer team across a 6-week period, and as a result, the findings and practical implications must be considered with caution when applying to another set of players from a league with different characteristics such as match demands, travel [48], climate [49], and over an extended period of time during a different phase of the season (early-, mid-, late-season, congested Christmas schedule). Furthermore, the sample size was also a limitation due to this study being conducted in the real-world, conducted with players from an elite soccer club. Our sample was selected as a convenience sample by recruiting all available outfield players from the first team of the club involved. Nevertheless, similar sample sizes have been used in previous studies conducted in elite soccer players in this research field. Secondly, match outcome was not considered, which has the potential to affect immunological and hormonal recovery profiles. Future investigations are warranted to evaluate these factors as they may be particularly relevant in different leagues across varying athletic populations during the season. Other limitations include the absence of training load, fatigue, and fitness profiling data [50].

### *Practical Implications*

The present findings may provide practitioners with detailed knowledge about acute and chronic variations in physical match performance and the subsequent recovery responses, that can be practically useful to assess and interpret change in individual and team performance. Previously, a number of practical recommendations to monitor immune function in athletes have been documented [7,9,10,16,51]. Match-play with higher physical outputs did not necessarily produce disturbances to mucosal immunity and hormonal balance. Therefore, accordingly, designing a structured, planned and individualized tailored recovery strategy and potential for squad rotation should be considered during demanding stages of the season to ensure immunological and hormonal recovery. Previous results highlighted that this might be particularly important during congested fixture schedules (Christmas fixture period) [7] and toward the end of the season [10]. Our findings support the use of s-IgA and s-Cort monitoring in professional soccer players and devising individual thresholds to determine values associated with inadequate recovery.

### **5. Conclusions**

As a result of this specific investigation, the data demonstrate for the first time that the use of salivary monitoring in combination with physical match load may be a useful tool to monitor individual mucosal immunity and hormonal responses to elite soccer match-play and the subsequent recovery periods. However, surprisingly no significant differences were observed between starters and non-starters at any time point, thus additional research is required. Finally, analysis of specific time points during recovery also warrants further investigation.

**Author Contributions:** Conceptualization, R.M. and R.D.M.; methodology, R.M.; software, R.D.M.; validation, R.M. and R.D.M.; formal analysis, R.D.M.; investigation, R.M. and E.B.; resources, R.M. and E.B.; data curation, R.M.; writing—original draft preparation, R.M., P.O., R.D.M. and A.M.; writing review and editing, R.M., E.B., R.D.M. and A.M.; visualization, R.D.M.; supervision, R.M.; project administration, R.M. 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 local Ethics Committee of Sechenov University (N 22-21 dated 12 December 2021).

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

**Data Availability Statement:** The data are not publicly available due to privacy reasons.

**Acknowledgments:** The authors want to thank all the players, and medical staff involved in the study for the professionalism shown throughout. The authors also want to state that the results of the present study do not constitute endorsement of any products.

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

### **References**


### *Article* **Effect of Small-Sided Games with and without the Offside Rule on Young Soccer Players: Reliability of Physiological Demands**

**Igor Junio Oliveira Custódio † , Renan Dos Santos † , Rafael de Oliveira Ildefonso, André Andrade, Rodrigo Diniz , Gustavo Peixoto, Sarah Bredt, Gibson Moreira Praça and Mauro Heleno Chagas \***

School of Physical Education, Physiotherapy and Occupational Therapy, Federal University of Minas Gerais,

Av. Antônio Carlos, 6627, CEP 31270-901, Belo Horizonte 31270-901, Brazil **\*** Correspondence: mauroufmg@hotmail.com; Tel./Fax: +55-31-3409-7443

† These authors contributed equally to this work.

**Abstract:** This study aimed to compare the physiological demand between three vs. three small-sided games (SSGs) with (3vs.3WITH) and without (3vs.3WITHOUT) the offside rule, as well as the withinand between-session reliability of this demand. Twenty-four U-17 soccer athletes performed various three vs. three (plus goalkeepers) SSGs with and without the offside rule. The data collection was performed within an eight-week period. Athletes' heart rate was monitored during the SSG. The variables analyzed were the percentage mean heart rate (HRMEAN%) and the percentage peak heart rate (HRPEAK%). For the analysis of within-session reliability, the mean value of the first two and last two SSG bouts performed within one day were used. The between-session reliability was calculated using the mean value of the four SSG bouts of each SSG type performed on two different days. In both SSGs, the values for reliability were significant and were classified as moderate to excellent. There were no significant differences in the physiological demand among SSG types. We concluded that the offside rule does not influence the physiological demand in a three vs. three SSG and the HRMEAN% and HRPEAK% present moderate to excellent reliability in a three vs. three SSG with and without the offside rule.

**Keywords:** small-sided games; task constraints; physiological demand; young soccer player; peak heart rate; offside rule; reliability

### **1. Introduction**

In recent years, the physical conditioning of soccer players has developed according to an integrated approach involving tactical and technical aspects of the game [1,2]. In this context, small-sided games (SSGs) provide high-intensity activity, including both tactical and technical demands, and optimize the available training time [3]. Knowledge of the effect of changing SSG characteristics (e.g., the number of players per team, the pitch size, and the rules) helps strength and conditioning coaches to adequately prescribe an SSG during the training process [4]. Although there are many studies on the effect of changing the pitch size and the number of players in a team [5–7], there has been less research on how rule changes in SSGs affect the players' physical and physiological responses [8–11]. One task constraint that can induce changes in players' available space is the offside rule, as it might reduce the effective playing area when the defending team moves towards the opponent's goal. To the best of our knowledge, the influence of the offside rule on the physiological demands of SSGs has not been investigated. Considering the importance of this rule on the game dynamics and the possibility of implementing it during game-based tasks such as an SSG, it is essential to understand its impact on athletes' physiological responses.

The players' movements and displacements during official matches are determined by the effective playing area, which is influenced by the offside rule. This constraint causes

**Citation:** Custódio, I.J.O.; Dos Santos, R.; de Oliveira Ildefonso, R.; Andrade, A.; Diniz, R.; Peixoto, G.; Bredt, S.; Praça, G.M.; Chagas, M.H. Effect of Small-Sided Games with and without the Offside Rule on Young Soccer Players: Reliability of Physiological Demands. *Int. J. Environ. Res. Public Health* **2022**, *19*, 10544. https://doi.org/10.3390/ ijerph191710544

Academic Editor: Paul B. Tchounwou

Received: 18 July 2022 Accepted: 20 August 2022 Published: 24 August 2022

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

**Copyright:** © 2022 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/).

the playing area to be dynamic, restricting or allowing players to move across the field length according to the position of teammates and opponents [12]. Hence, the relative area (i.e., area per player) also constantly changes during the game [13]. Previous studies have suggested reducing the relative area by decreasing the absolute pitch size (area in m<sup>2</sup> ) while maintaining the number of players [14], or keeping the absolute pitch size and increasing the number of players [15]. A smaller relative area generally decreases players' physical [9,16] and physiological [4] responses, because it constrains players' displacements. Therefore, another possibility to modulate the relative area in SSGs is the inclusion of the offside rule, because it can reduce the effective space in which players can move. However, some studies on soccer SSGs have included the offside rule [17,18], while others have not [19,20]. Castillo et al. [9] compared the physical demands of a soccer SSG with and without the offside rule and found a greater total distance and a larger distance covered between 13 and 16 km/h on the pitch without the offside rule. Therefore, it might be expected that non-offside SSGs lead to greater physiological responses from the players. Nonetheless, the influence of this rule on the relative area and consequently on the physical and physiological demands of soccer SSGs requires deeper investigation. Moreover, this knowledge may add a new interpretation to previous studies on SSGs that have or have not implemented the offside rule. Understanding the impact of this rule on athletes' responses can help the coaches to better use game-based activities during training.

Another critical issue regarding the use of SSGs is their reliability as a means of training. This analysis is crucial to test whether specific demands can be achieved when an SSG format is repeatedly applied during the training process. Weir [21] suggested using the intraclass correlation coefficient (ICC) and the standard error of the measurement (SEM) to analyze the reliability. The ICC provides information on the variability between individuals and the consistency of this variability in repeated test measures [21], while the SEM reflects the degree of fluctuation of the individual's scores in a test or condition, representing the expected natural variability (the random error) for that score [21]. Some studies have investigated the reliability of the physiological responses during different soccer SSGs and presented within- [20] and between-session designs [22]. Many of these studies showed high reliability for physiological demands [22–28]. A recent systematic review indicated that internal loads—average heart rate (%HRavg), peak heart rate (%HRpeak), and maximum heart rate (%HRmax)—showed small within-session variations (~0.5–6% of change between the lowest and the highest sets/repetitions), irrespective to the SSG format. Therefore, it is possible to expect high reliability of internal load measures in both with and without offside SSGs in the current study [29].

Considering these issues, this study aimed to (i) compare the physiological demands of a three vs. three SSG with and without the offside rule and (ii) to verify the within- and between-session reliability in these two SSGs.

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

### *2.1. Participants*

Twenty-four U-17 male soccer athletes (age: 16.7 ± 0.6 years; body mass: 64.8 ± 6.7 kg; height: 176.5 <sup>±</sup> 6.5 cm; body fat: 9.7 <sup>±</sup> 1.6%; and estimated VO2MAX: 52.1 <sup>±</sup> 2.5 mL·kg−<sup>1</sup> ·min−<sup>1</sup> ) from an elite club participated in this study. This club was considered elite as players compete at the national level regularly. The club achieved first position in the national U-18 competition in the same year the data collection was performed. The athletes competed at a national level and had seven training sessions per week. Data from two athletes were excluded from the analyses due to technical problems, which reduced the final sample to twenty-two players. Players were included if they volunteered to participate in the study and were not injured or returning from injury. On the other hand, the exclusion criteria comprised being injured, not participating in the whole data collection, or refusing to provide written consent to participate in the study. Goalkeepers participated in the data collection but were not evaluated. The participants and their legal guardians were informed about all the research procedures and provided written consent for participating

in the study. The local Ethics Committee from the Universidade Federal de Minas Gerais (70103017.0.0000.5149) approved the study, and all the guidelines from the Declaration of Helsinki were followed. Helsinki were followed. *2.2. Teams' Composition for the SSG*

twenty-two players. Players were included if they volunteered to participate in the study and were not injured or returning from injury. On the other hand, the exclusion criteria comprised being injured, not participating in the whole data collection, or refusing to provide written consent to participate in the study. Goalkeepers participated in the data collection but were not evaluated. The participants and their legal guardians were informed

study. The local Ethics Committee from the Universidade Federal de Minas Gerais (70103017.0.0000.5149) approved the study, and all the guidelines from the Declaration of

#### *2.2. Teams' Composition for the SSG* The 24 athletes were randomly allocated into eight teams of three players (A to H).

*Int. J. Environ. Res. Public Health* **2022**, *19*, x FOR PEER REVIEW 3 of 11

The 24 athletes were randomly allocated into eight teams of three players (A to H). Each team had a defender, a midfielder, and a forward to allow teams to explore the physical, technical, and tactical specificities of each playing position during the different SSGs [30,31]. The eight teams were divided into two groups. Group 1 was composed of teams A to D, and Group 2 was composed of teams E to H. Each team within the group played against the same opponent during the entire study (e.g., Team A always played against Team B) to reduce the possible variability related to differences in the opposing teams during the SSGs [32]. The procedures for the composition of the teams and groups are described in Figure 1. Each team had a defender, a midfielder, and a forward to allow teams to explore the physical, technical, and tactical specificities of each playing position during the different SSGs [30,31]. The eight teams were divided into two groups. Group 1 was composed of teams A to D, and Group 2 was composed of teams E to H. Each team within the group played against the same opponent during the entire study (e.g., Team A always played against Team B) to reduce the possible variability related to differences in the opposing teams during the SSGs [32]. The procedures for the composition of the teams and groups are described in Figure 1.


**Figure 1.** Team and group composition procedures. Legend: d = defender; m = midfielder; f = forward.

#### *2.3. Data Collection 2.3. Data Collection*

Athletes performed several 3 vs. 3 SSGs (where goalkeepers were included but not evaluated) with (3vs.3WITH) and without (3vs.3WITHOUT) the offside rule. Both of the SSGs were played in the 3 vs. 3 format, on a 36 × 27 m pitch of natural grass, with goals measuring 6 × 2 m (see Figure 2). In the 3vs.3WITH game, two referees were positioned on the sides of the pitch to observe the game and apply the offside rule when necessary. The defending team received a free kick when an offside situation was detected. In the 3vs.3WITHOUT game, the offside rule was not applied, so players could play freely. Each session comprised four SSG bouts, which lasted for four minutes, with five minutes of passive rest. Additional balls were placed around the pitch to ensure a fast game restart when the ball went off the pitch. Coaches and researchers did not give the players verbal encouragement or technical instructions. Athletes performed several 3 vs. 3 SSGs (where goalkeepers were included but not evaluated) with (3vs.3WITH) and without (3vs.3WITHOUT) the offside rule. Both of the SSGs were played in the 3 vs. 3 format, on a 36 × 27 m pitch of natural grass, with goals measuring 6 × 2 m (see Figure 2). In the 3vs.3WITH game, two referees were positioned on the sides of the pitch to observe the game and apply the offside rule when necessary. The defending team received a free kick when an offside situation was detected. In the 3vs.3WITHOUT game, the offside rule was not applied, so players could play freely. Each session comprised four SSG bouts, which lasted for four minutes, with five minutes of passive rest. Additional balls were placed around the pitch to ensure a fast game restart when the ball went off the pitch. Coaches and researchers did not give the players verbal encouragement or technical instructions.

*Int. J. Environ. Res. Public Health* **2022**, *19*, x FOR PEER REVIEW 4 of 11

**Figure 2.** Representation of the 3vs.3WITHOUT game. Legend: G = goalkeeper; D = defender; M = midfielder; F = forward. **Figure 2.** Representation of the 3vs.3WITHOUT game. Legend: G = goalkeeper; D = defender; M = midfielder; F = forward.

The SSGs were performed on Tuesdays and Wednesdays for eight consecutive weeks at the end of the competitive season. We chose the same weekdays to minimize the influence of the distribution of training loads on athletes' physical responses. Group 1 performed the SSGs during the first four weeks (one SSG format each day), while Group 2 performed the SSGs in the last four weeks. This was to avoid a long break between SSG sessions for each team, which could lead to changes in physical fitness, and to minimize the disruption to the athletes' training routines. Therefore, each SSG format was performed twice, with an interval of six to eight days between trials for each SSG format, according to the club availability. The SSGs were performed on Tuesdays and Wednesdays for eight consecutive weeks at the end of the competitive season. We chose the same weekdays to minimize the influence of the distribution of training loads on athletes' physical responses. Group 1 performed the SSGs during the first four weeks (one SSG format each day), while Group 2 performed the SSGs in the last four weeks. This was to avoid a long break between SSG sessions for each team, which could lead to changes in physical fitness, and to minimize the disruption to the athletes' training routines. Therefore, each SSG format was performed twice, with an interval of six to eight days between trials for each SSG format, according to the club availability.

To standardize the influence of circadian rhythm on the observed responses, all sessions were performed at the same time (between 8 a.m. and 10:30 a.m.). The mean (the standard deviation) temperature and relative humidity of all sessions were 31.1 °C (± 2.6 °C) and 28.1% (± 4.8%), respectively, recorded by a portable digital thermometer (Big Digit Hygro-Thermometer, Extech Instruments, Massachusetts, EUA). To standardize the influence of circadian rhythm on the observed responses, all sessions were performed at the same time (between 8 a.m. and 10:30 a.m.). The mean (the standard deviation) temperature and relative humidity of all sessions were 31.1 ◦C (± 2.6 ◦C) and 28.1% (± 4.8%), respectively, recorded by a portable digital thermometer (Big Digit Hygro-Thermometer, Extech Instruments, Massachusetts, EUA).

To control for the possible effect of changes in physical conditioning on the reliability analysis, athletes performed the Yoyo Intermittent Recovery Test Level 1 (Yo-YoIR1) [31] and a 20 m sprint test one week before and two weeks after the data collection. To control for the possible effect of changes in physical conditioning on the reliability analysis, athletes performed the Yoyo Intermittent Recovery Test Level 1 (Yo-YoIR1) [31] and a 20 m sprint test one week before and two weeks after the data collection.

In detail, the protocol used for the 20 m sprint test consisted of taking four attempts at the 20 m test, and time recording the distance covered. An interval of three minutes of passive recovery between attempts was established. It is noteworthy that the distance of In detail, the protocol used for the 20 m sprint test consisted of taking four attempts at the 20 m test, and time recording the distance covered. An interval of three minutes of passive recovery between attempts was established. It is noteworthy that the distance of 20 m was chosen for the measurement of running speed due to evidence that, in official games, sprint running distances longer than 20 m are infrequent [33].

The Yo-YoIR1, on the other hand, is an intermittent, progressive aerobic capacity test, in which athletes perform a series of round-trip runs on a 20 m course [31]. So after each

round trip, there is an interval of 10 seconds of active rest in which the athlete trots or walks a course of 10 m, covering 5 m going and 5 m returning. The running speed is determined by sound signals, starting at 10 km/h and increasing progressively throughout the test. In the present study, when the athlete was unable to maintain the rhythm (the speed) determined by the sound signals for two consecutive series, the test was closed, and the total distance covered was recorded. The peak heart rate achieved during Yo-YoIR1 was considered as the athletes' maximum heart rate and was used to relativize heart rate values as a percentage of the maximum.

### *2.4. Physiological Demand*

The heart rate (HR) of the players during the SSGs was recorded using a 1 Hz heart rate monitor (Polar T31 Electro Oy®, Kempele, Finland). The reliability of this device has been previously tested in the literature. Physiological demands were characterized by the percentage of mean heart rate (HRMEAN%) and the percentage of peak heart rate (HRPEAK%). The HRMEAN% was calculated as the mean of all the values recorded by HR monitors during the SSG bouts (HR values of the rest intervals were excluded). The HRPEAK% was considered to be the highest value recorded during the SSG bouts. All HR values were relativized by the peak HR presented by each athlete in the Yo-YoIR1.

### *2.5. Statistical Analyses*

The data did not present significant deviations from normality (using Shapiro–Wilk's test) or homoscedasticity (using Levene's test). An independent t-test was used to compare means between the 3vs.3WITH and 3vs.3WITHOUT games. Cohen's d effect size was calculated to characterize the magnitude of the significant differences in paired comparisons and was classified as insignificant (<0.19), small (0.20–0.49), medium (0.50–0.79), or large (≥0.80) [32].

For the within-session reliability of the HRMEAN% and HRPEAK% for the 3vs.3WITH and 3vs.3WITHOUT games, athletes' mean values of the first two and the last two SSG bouts in each session (day 1 and day 2) were used. To determine the between-session reliability, athletes' mean values of the four SSG bouts performed in each session were used. For both within- and between-session reliability, the intraclass correlation coefficient 2,k (ICC2,k) and the standard error of the measurement (SEM) were used [21]. The ICC2,k values were classified as weak (<0.4), moderate (0.40–0.59), good (0.60–0.74), or excellent (0.75–1.00) [34].

A two-way analysis of variance (groups × moments) was used to compare the data on aerobic power (from the Yo-YoIR1) and sprint performance (from the 20 m sprint) among the two groups and moments (from the pre- and post-data collection).

The level of statistical significance was set at 5% (α = 0.05). All analyses were performed using SPSS version 23.0 (Chicago, IL, USA).

### **3. Results**

Table 1 shows the descriptive data (the means and standard deviations) of HRMEAN% and HRPEAK% in the investigated SSGs. There were no significant differences between the SSGs with and without the offside rule (giving a small effect size).

**Table 1.** Means (standard deviations) of the variables related to the physiological demand of SSGs with and without the offside rule.


Legend: **3vs.3WITH** = small-sided games with the offside rule; **3vs.3WITHOUT** = small-sided games without the offside rule; **FCPEAK%** = percentage peak heart rate; **FCMEAN%** = percentage mean heart rate.

Table 2 shows the within-session (bouts within days 1 and 2) intraclass correlation coefficient values (95% CI), the ICC classification, and the SEM values for the variables related to the physiological demand of SSGs with and without the offside rule. The ICC values were classified as "good" or "excellent" (values above 0.60), except for the HRMEAN%, which was classified as "moderate" on day 2.

**Table 2.** Within-session intraclass correlation coefficients (95% CI), ICC classification, and SEM for the variables related to the physiological demand of SSGs with and without the offside rule.


**3vs.3WITH** = small-sided games with the offside rule; **3vs.3WITHOUT** = small-sided games without the offside rule; **FCPEAK%** = percentage peak heart rate; **FCMEAN%** = percentage mean heart rate; **CI** = confidence interval; **SEM** = standard error of the measurement. **\*** indicates statistical significance (*p* < 0.05).

Table 3 shows the between-session (between days 1 and 2) intraclass correlation coefficient values (95% CI), the ICC classification, and the SEM values for the variables related to the physiological demand of SSGs with and without the offside rule. The ICC values were classified as "good" or "excellent" (values above 0.60), except for the HRMEAN% in the 3vs.3WITH game, which was classified as "moderate".

**Table 3.** Between-session intraclass correlation coefficients (95% CI), ICC classification, and SEM for the variables related to the physiological demand of SSGs with and without the offside rule.


**3vs.3WITH** = small-sided games with the offside rule; **3vs.3WITHOUT** = small-sided games without the offside rule; **FCPEAK%** = percentage peak heart rate; **FCMEAN%** = percentage mean heart rate; **CI** = confidence interval; **SEM** = standard error of the measurement. **\*** indicates statistical significance (*p* < 0.05).

The two way analysis of variance of the control variables (aerobic power—pre-test: 1850.9 ± 288.7 m; post-test: 1950.0 ± 277.6 m and 20 m sprint performance—pre-test: 22.7 ± 0.6 km/h; post-test: 23.3 ± 0.6 km/h) showed no significant interaction (aerobic power—F = 0.68; *p* = 0.41; 20-m sprint performance—F = 0.985; *p* = 0.325) or main effects (aerobic power—F = 3.47; *p* = 0.07; 20 m sprint performance—F = 0.352; *p* = 0.556). These data show the lack of differences in physical conditioning during the period of the data collection, mitigating the possible effect of variability on the between-session reliability analysis.

### **4. Discussion**

This study aimed to investigate the effect of the offside rule on the physiological demands of three vs. three soccer SSGs in U-17s and the reliability of the physiological demands in three vs. three SSGs with and without the offside rule. The results show that the physiological demands, characterized by the HRPEAK% and HRMEAN%, did not differ among the SSGs with and without the offside rule, and thus, our hypothesis was rejected. Furthermore, the within and between-session reliability of physiological demands confirmed our hypothesis, with moderate to excellent ICC values for all variables, regardless of the rules of the SSGs.

We expected that the offside rule would decrease the physiological demands of the SSGs because of the reduction in the effective playing area. This hypothesis was based on previous results that showed a decrease in the physiological demands when the absolute playing area was decreased for the same number of players or when the number of players was increased within the same playing area [16]. These changes result in smaller relative areas (i.e., area per player), restricting the available space for players to move around and, consequently, reducing the intensity of the game (i.e., lower physiological demands) [14,35]. A previous systematic review included studies with similar playing areas and showed that reducing the relative area per player tends to reduce physical and physiological responses [16]. However, the results of the present study do not corroborate this hypothesis. A possible explanation for these divergent results may be related to the magnitude of the change in the effective playing area in the three vs. three SSG with the offside rule. Previous studies have shown that small changes in the relative area may not be sufficient to influence players' physical [9] and physiological [4] responses. Specifically, the reduction in the effective playing area depends on the defending team moving up the pitch to constrain the available space for the offensive team. Therefore, the number of times the players adopted this behavior might have been smaller than what was required to induce different responses when considering the whole bout. Moreover, although the heart rate has been widely used in studies on SSGs [36] and is considered a valid variable to measure SSG intensity [37], it may not be sensitive enough to detect differences in the frequency of specific actions (i.e., jumps, duels, accelerations, decelerations, sprints, and changes of direction) during the game, which could, in turn, also reflect game intensity [38]. Considering this issue, future studies should collect information through other variables, such as accelerations, decelerations, mean speed, and distances covered in different speed zones, to increase the understanding of exercise intensity during game-based activities, such as SSGs [39].

The HRPEAK% and HRMEAN% values found in both SSGs investigated in this study are similar to those reported in previous studies on the three vs. three SSG format performed by soccer players of a similar age (Sub-17) [40–42]. Furthermore, studies on SSG training (training periods above four weeks) indicate the necessity for HRmean values to be above 80% of HRmax to improve aerobic performance [43–46]. Therefore, the results of the present study reinforce the potential use of different SSGs for the improvement of aerobic performance in soccer athletes, including the offside rule.

The investigation of SSG reliability is essential to support using SSGs during training. In addition, with the knowledge of the demands imposed on athletes by different SSGs, strength and conditioning coaches can examine if those demands are reproducible when the same SSG is performed at different moments. In the present study, high ICC (>0.60) and low SEM (<1.7) values were found in the within-session reliability analysis of HRMEAN. These data corroborate the results of previous studies on the reliability of heart rate variables collected during SSGs, despite the differences in the SSG formats. Hill-Haas et al. [24] compared different SSG formats (two vs. two, four vs. four, and six vs. six) and found percentage values of SEM (SEM%) of 1.9 and 4.4% for the HRPEAK% and 1.1 and 3.6% for the HRMEAN%. Another study also reported small SEM percentage values for the HRMEAN% (5.4%) and HRPEAK% (3.0%) in a three vs. three SSG with similar characteristics [22]. Finally, Stevens et al. [28] found good reliability values for the HRMEAN% during a six vs. six SSG (ICC = 0.61 and SEM% = 2.2%). On the other hand, the results of the present

study on between-session reliability suggest good reproducibility of the HRPEAK% and HRMEAN%, despite an interval of one week between the sessions (ICC = 0.56), with a low variability among these measures (SEM < 2.6%). These results are similar to previous research that indicated good reproducibility for the physiological demands represented by heart rate variables in different SSGs. Da Silva et al. [23] and Rampinini et al. [27] investigated the reliability of the HRMEAN% in SSGs with different numbers of players and pitch sizes and found that values of SEM percentage ranged from 2.2 and 3.4%, and the percentage of typical error (TE%) values (similar to SEM) ranged from 2.0% and 5.4%, respectively. Additionally, Hill-Haas et al. [25] found low variability for the HRMEAN%, with TE% values ranging from 2 and 4% in a four vs. four SSG. This result is similar to that found by Hulka et al. [47], which showed high ICC (0.88) and low SEM% (2.35%) values in a four vs. four SSG. Additionally, both with and without the offside rule, SSGs showed similar classifications regarding the reliability measures. However, when looking at both within- and between-session reliability, the SSG without the offside rule showed lower ICC values than the SSG with it. It has been proposed in the literature that a higher movement variability can be detected in lesser-known game formats [48–50]. It can be argued that U-17 soccer players usually engage in more specific tasks than those that are general game-based tasks—therefore, the game with the offside rule seems to be more representative of the requirements of the official match. Consequently, the reduction in the reliability might indicate a more variable displacement behavior in the SSGwithout condition due to the players' need to readapt to the new constraints.

This study investigated U-17 athletes, which hinders the generalization of the results to other age categories. Future studies should be carried out with athletes of different ages to provide more precise information on the physiological demands of the three vs. three SSGs investigated in this study. Moreover, this study did not monitor athletes' recovery levels during the data collection, which could have added a deeper understanding of athletes' conditions while recording the variables. In this case, further research should investigate athletes' recovery behavior over SSG bouts and between training sessions to provide information that better supports the use of SSGs for the physical conditioning of soccer players.

### **5. Conclusions**

Using the offside rule in a three vs. three SSG did not influence the physiological responses of young soccer athletes. The within- and between-session reliability values of the physiological variables in both SSGs with and without the offside rule were high, supporting the reproducibility of the physiological demands of SSGs despite their natural unpredictability and variability. The absence of difference between the protocols indicates that coaches might choose between the two SSG formats based on other goals—for example, tactical missions related to enlarging the surface area—instead of considering the impact the offside rule will have on players' physiological responses.

**Author Contributions:** Conceptualization, R.D.S., G.M.P. and M.H.C.; formal analysis, A.A. and R.D.; investigation, R.D.S., I.J.O.C., R.d.O.I. and S.B.; methodology, R.D., G.P. and S.B.; writing—original draft preparation, I.J.O.C., R.d.O.I. and G.P.; writing—review and editing, R.D.S., A.A., S.B. and M.H.C.; project administration, R.D.S. and M.H.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the Pró-Reitoria de Pesquisa da Universidade Federal de Minas Gerais (PRPq-UFMG), Coordenação de Aperfeiçoamento de Pessoal de Nível Superior—Brasil (CAPES), Conselho Nacional de Desenvolvimento Científico e Técnológico (CNPq), and Fundação de Amparo à Pesquisa do Estado de Minas Gerais (FAPEMIG).

**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 Federal University of Minas Gerais/Brazil (Protocol number 70103017.0.0000.5149 and date of approval was 8 September 2017).

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

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

### **References**

