Introduction

The highly individualized, non-linear relationship between age and the development of maturity-related anthropometric characteristics (e.g., stature and body-mass) [1, 2] often leads to the over-selection of early maturing soccer players for academy development programs in comparison to their later maturing counterparts [3–5]. This selection bias is attributed to early maturing players who are often characterized as being temporarily taller and heavier, but not necessarily technically better, in comparison to their later maturing team-mates [3]. Therefore, soccer match and training formats which can control for the confounding effect of biological maturation [4–6] are of importance and relevance to player development practitioners responsible for prescribing suitable training loads and playing environments.

The grouping of athletes according to their maturation status, rather than chronological age is commonly referred to as bio-banding [4, 5, 7]. The exclusive [8] objective of bio-banding is to reduce between player maturity-related differences in anthropometric [9], and physical fitness characteristics, and create a more equitable playing and training environment [10–17]. Bio-banding has been shown to be an effective strategy to reduce the large within group variations associated with the early development of maturity-related anthropometric characteristics, such as stature and body-mass [1, 2] which can be directly associated to the maturation process [9]. Typically, bio-banding methods use one of two popular methods for estimating maturity status [18], including maturity offset methods [19–21] (i.e. estimating the number of years from/past the onset of peak height velocity [PHV]), or estimating the players percentage of final adult stature [22]. Bio-banding has become increasingly popular within academy soccer practices to enhance practitioner understanding of the influence player maturity status has on player characteristics including psycho-social [10, 12], technical-tactical [10, 11, 14], and perceived effort [10, 14]. Which are all considered important by talent practitioners [23, 24] during soccer match-play. Not only is the application of bio-banding important to key stakeholders, such as parents or guardians, and coaches responsible for supporting, and the (de)selection of players for development programs [25], but the consideration of player maturation status is of relevance to practitioners who are responsible for the prescription of training loads and volumes [26]. This concern is evidenced by academy soccer players who have been categorized as being pre-PHV accumulating greater session ratings of perceived exertion training load (sRPE-TL) (being RPE*activity duration in minutes) when competing in small-sided games (SSG) against their more mature (i.e. post-PHV) counterparts, despite internal measures of training load suggesting no meaningful difference between the groups of players [10].

Training loads within soccer are commonly assessed with different metrics that tend to be grouped by internal (e.g. heart rate, blood lactate) or external training loads (e.g. total distance covered, high speed running distances etc.) [27, 28]. Within youth soccer, measurements of external training loads have focused on time-motion analysis data [29, 30], typically reported from micromechanical electrical systems (MEMS) devices or more commonly referred to as global positioning systems (GPS) devices. In contrast, measurement of internal training loads have focused on heart rate and subjective load data [28]. Rather than providing distinct information, it is likely that measures of internal and external training load will demonstrate various associations and relate to other constructs such as intensity and volume of training [31, 32]. In research conducted with academy soccer players, Maughan et al. (2021) used principal component analysis (PCA) to investigate the underlying structure of the relationships between external measures of training load and sRPE-TL. Their results identified that most of the variation within the data could be explained by two components reflecting the training volume (to which sRPE-T) and training intensity [33]. The authors concluded that multivariate techniques should be employed to better understand the complex nature of training loads in youth soccer. Which is of relevance to academy practitioners, given that the highly individualized effect of biological maturation on players anthropometric (i.e. stature and body-mass) development [1, 2] has been associated with subsequent temporary alterations in soccer players functional movement capacity [34]. Such alterations may lead to imbalances between strength and flexibility, and temporary reductions in movement mechanics typically referred to as “adolescent awkwardness” [35]. Ultimately, leading to increased risk of sustaining a non-contact injury [36]. An extension of this conclusion holds that multivariate techniques may enable more effective monitoring and subsequent management of training loads in academy soccer players who are undergoing PHV.

In addition to delineating between training volume and intensity, it has been suggested that perceptual measures to quantify the contributions of specific cardiovascular and neuromuscular/musculoskeletal systems can provide relevant information to monitor and prescribe training in team sports [37, 38]. Differential ratings of perceived exertion (dRPE) that separate scores for breathlessness (RPE-B), leg muscle exertion (RPE-L), and technical/cognitive exertion (RPE-T) [38] may provide a viable alternative to quantify internal loads [39, 40]. To date, there has been no research investigating the full range of dRPE measures within an academy soccer context. However, studies conducted using Australian Football League have shown there to be likely small differences between RPE-L and RPE-T (5.5%), a likely small difference between RPE-L and RPE-B (3.5%) and a possibly small difference between RPE-B and RPE-T (1.9%) [38]. With combined dRPE training loads explaining 66–91% of the variance in sRPE training loads within professional rugby union players, and the strongest associations being exhibited with sRPE-TL being sRPE-L for high-intensity intervals (r = 0.67), sRPE-B for repeated high-intensity efforts (r = 0.89) and sRPE-T for Speed (r = 0.63) and Skills (r = 0.51) [37]. Additionally, most research investigating dRPE has used only correlation analyses that are limited in their ability to explore underlying relationship structures, with data comprising more than two variables. Whilst PCA provides a more effective method to explore structures [31], the method has limitations when applied to repeated measures data, where many series of measurements are made across the same variables for a relatively small number of individuals. In contrast, more modern multivariate techniques, including MultiLevel Simultaneous Component Analysis (MLSCA) can better investigate both the sources of inter- and intra-individual variability in multivariate data and identify individuals or groups of individuals that differ with regards to measures of location, variance and covariance [41]. In the context of team sports, MLSCA can be used to identify whether the variability in multivariate measurements are primarily caused by differences across individuals, or differences within individuals due to variation in relevant interventions or independent variables. Additionally, MLSCA can be used to identify individuals who tend to score highest/lowest or are the most/least variable on multivariate components described by combinations of the outcome variables [42]. Given the popularity of bio-banding methods and the increased focus on multivariate statistical techniques, the aims of the current study were to investigate the use of dRPE with academy soccer players to: 1) examine the effect of bio-banded and non-bio-banded maturity groups within SSG on players dRPE; 2) describe the multivariate relationships between dRPE measures investigating the sources of intra and inter-individual variation, and the effects of maturation and bio-banding.

Methods

Having institutional ethics committee approval (FHS189) and parental/guardian consent, the current study used previous methods [16] which have investigated the perceived exertion of academy soccer players during SSG`s, when teams were selected based upon their chronological age or their bio-banding grouping. Additionally, this study investigated if dRPE measures provide distinct information, and to establish the associated sources of variation including between and within players.

Results

Descriptive statistics of dRPE values for pre-PHV and post-PHV players are presented in Table 1. Results from mixed effects models (Table 2) identified that in general, pre-PHV players reported higher values for perceived exertion, with point estimates indicating relative increases ranging from ~6 to 16 units (p <0.001 to p = 0.121). Main effects were also identified for game type, with players reporting higher values (~8 to 9 units) across all RPE variables (p<0.001). However, interaction effects between maturity and game type across all RPE variables (p≤0.012) identified that increases in RPE values for mixed games were highest for post-PHV players (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Means and standard deviations of dRPE values across groups.

https://doi.org/10.1371/journal.pone.0270259.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Results of mixed effects regression models of maturation and game type effects on dRPE values.

https://doi.org/10.1371/journal.pone.0270259.t002

The MLSCA analysis identified that variation in dRPE measures were explained relatively equally from between (59.8%) and within (40.2%) sources. CHull tests identified that two components should be selected for the between model, and a single component for the within model (Table 3). For the between model, the first component loaded sRPE-B and sRPE-L, with the second component loading sRPE and sRPE-T (correlation between components equalled 0.82). For the within model, the single component represented equal weighting across all four RPE measures. Adding another component to the within model revealed the same structure that was obtained for the between model, with loadings of sRPE-B and sRPE-L, and loadings of sRPE and sRPE-T (correlation between components equalled 0.87). Use of the MLSCA loadings to generate multivariate perceived exertion values aligned with the univariate regression models, with the between model identifying greater values from post-PHV players (component 1: d = 0.51, p = 0.094; component 2: d = 0.95; p<0.0015). In contrast, analysis of the within component variation identified higher mean values for post-PHV players with a medium effect size (d = 0.53; p = 0.085).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. MLSCA between- and within-model component loadings and percentage of explained variance.

https://doi.org/10.1371/journal.pone.0270259.t003

Discussion

The current study examined the effect of bio-banding in academy soccer players within 4v4 SSG’s and the influence on players dRPE. The main findings of the study were; 1) In general, pre-PHV players exhibited higher dRPE compared with their post-PHV counterparts; 2) Mixed maturity games reported higher dRPE outputs overall; 3) Variation in dRPE measures across a series of bio-banded games are caused by both between and within sources of variation in relatively equal amounts; 4) Across a series of bio-banded games, the four dRPE measures do not provide unique information, and between variation is best expressed by one or two highly correlated components, with within variation best explained by a single equally loaded component.

Bio-banding in soccer is a method for grouping adolescent players to reduce the within group variations of maturity related anthropometric characteristics [9], which is purported to support the talent identification and player development pathway [10]. Within the current study, those players defined as pre-PHV reported 20–50% higher dRPE values across all constructs for both bio-banded and mixed maturity SSG’s. Despite evidence to suggest maturity-related differences in perceptual effort during SSG match-play exist [9], to date, studies exploring all constructs of dRPE within academy soccer SSG games based upon maturity status are absent. The findings observed when playing against post-PHV are like that observed within senior soccer players, when playing against opponents of a higher rank, with sRPE-T showing large differences [53]. Towlson and colleagues (2021) suggested that players bio-banded as ‘pre-PHV’ had increased challenges to overcome playing against post-PHV players, which would suggest an increase in perceived load consistent with the findings of the current study. Interestingly, during mixed maturity SSG`s, both pre-PHV and post-PHV developers dRPE scores increased. This may be because of accumulated player fatigue, with the players having competed in 60 minutes of SSG match-play. However, given that the players received at least 5 minutes of passive recovery between games, accompanied by an additional 20 minute of passive recovery between the bio-banded and mixed maturity categorisation formats, it is unclear if this elevation in RPE was a direct of result of accumulated fatigue or change in game format. During analysis of passing networks during SSG’s, there is an increased reliance on the post-PHV players during mixed games in comparison to bio-banded games [16]. This subjective reliance was characterised by post-PHV players becoming more integral to passing networks and team dynamics. This was assessed via betweenness centrality, a measure of how often a player lies on the shortest path in the passing network from one player to another, and page rank, a measure of player importance within team dynamics [16]. These findings, alongside those in the current study provide some context and consideration for practitioners considering the practical benefits of using bio-banding during talent identification and development processes. Whilst post-PHV players have been shown to report lower perceived exertion within SSG`s, their involvement appears to be higher than pre-PHV players. Pre—PHV players report higher values of perceived exertion, however, are possibly more reliant on post-PHV players [16] during mixed maturity SSG`s. This could further contribute to the over-selection of early maturing players, who often possess advanced maturity-related anthropometric (primarily enhanced stature) and physical fitness characteristics for key defensive roles (e.g., central defence and goalkeeper) [43, 54]. Subsequently homogenising the type of player within the academy system which a senior squad can select from.

Ratings of perceived exertion have been used to assist academy practitioners to assess the internal load ensued by academy players during chronologically [10] and bio-banded [14] match-play. However, it has been suggested that the one-dimensional approach of overall session ratings of perceived exertion (sRPE) oversimplifies the self-perception of effort [37], suggesting the perception of effort should be differentiated in to specific categories [37]. However, previous research has shown that different dRPE measures may provide similar information when analysed using multivariate methods [33]. When using PCA to investigate relationships between objective and subjective load measures, Maughan et al. [33] found that measures of sRPE, sRPE-L and sRPE-B were heavily loaded within the first principal component that they suggested represented a measure of total training load. Further multivariate analysis through exploratory factor analysis, identified four latent factors, one of which represented subjective load [33]. The authors suggested that univariate correlations, alongside the relationships found in the multivariate assessments suggested that differentiating subjective measures of load had limited benefits for the population analysed. Similar findings were obtained in the present study when analysing the dRPE measures with MLSCA. The analyses identified that both between and within sources were relatively equal (59.8 vs. 40.2%, respectively) in explaining variation in dRPE values. That is, dRPE values were systematically different across players (e.g. due to factors such as maturation), but also varied within players across games (e.g. due to factors such as game type). Statistical tests suggested that a single component model was best to explain with variation, and a two-component model was best to explain between variation. However, the components for the two-component were highly correlated (r = 0.82). When utilising a two-component model, both the between and within-models produced a split between sRPE-B and sRPE-L which loaded heavily within the first component and sRPE and sRPE-T which loaded heavily in the second component. These findings suggest that when taking subjective measures during SSG in academy soccer players, the individuals gestalt sRPE provided similar information to a separate subjective measure assessing technical demands (sRPE-T). Equally these measures may be somewhat distinct to measures of physical exertion regarding breathlessness (sRPE-B) and leg muscle exertion (sRPE-L). The findings that the four dRPE measures did not measure distinct constructs and that they were more likely to represent a single weighted component, or two closely correlated components may be influenced by the data collection environment and its relative homogeneity (e.g., collection of data across very similar SSG’s). The potential for dRPE values and relationships to be influenced by the type of training has previously been argued (26). Given the data collected here was from SSG, placing demands on the technical abilities of players, it is reasonable to expect there to be some relationship between sRPE and sRPE-T. Previous research in rugby union players found relationships between sRPE and sRPE-T for training categorised as ‘skills’, however weaker relationships for ‘skill-based conditioning’, the category which most resembled SSG [37]. Collectively, these findings suggest there is a modality effect on the relationship between dRPE measures, and this is likely to be affected by the participants and the training modality investigated.

Despite this study showing early evidence to suggest that maturity related differences in perceived exertion of players during SSG`s and that bio-banding may be a suitable method to control for this, there are limitations to this study which should be considered. Firstly, data were collected from two English professional male soccer academies and, as per previous research [14], the findings may not be generalisable to other clubs, levels, or sports. Given that the adolescent growth spurt in male academy soccer players typically manifests between 9.7–10.7 years to 13.8–15.2 years [1, 2], methods used for assessing subjective perceptions of exertions were considered appropriate. Previous research using the Borg CR10 scale highlighted a lack of sensitivity which may influence relationships between variables. The CR100 scale, which was used in the present study, has been suggested due to its increased sensitivity in comparison to the CR10 scale [37]. This increase in sensitivity was not evidenced within this study and may be due to factors such as the cognitive maturity of the players which was not measured within the present study. Previous research in youth soccer players has endorsed the use of scales which contain both pictorial and verbal anchors, such as the OMNI scale, to aid players in differentiating between RPE categories [55]. Further research investigating the use of different scales, alongside multivariate methods of analysis, may further the understanding of practitioners with regards to relationships between dRPE measures.

Conclusion

Using a bio-banding SSG design study, we have shown that pre-PHV players report higher subjective measures of exertion than post-PHV players during SSG`s. Additionally, when evenly mixing players based on measures of maturation, higher measures of perceived exertion were generally reported. Collectively these findings support previous findings by Towlson and colleagues (2021) and further suggest that the use of maturity-matched bio-banded formats create a more equitable (physical) playing environment to supplement academy soccer player development and talent identification pathways. Whilst maturity-matched bio-banded formats may control the exertion experienced by young players, practitioners should also consider that exposure to adversity that comes from being exposed to bigger, stronger and faster players may limit players opportunity demonstrate key psychological behaviours [12] which talent practitioners perceive as most important during the talent identification process [25]. Mixed formats may also promote an over-reliance on early-developed players, leading to higher perceived exertion generally across both post-PHV-, and pre-PHV players [16]. Finally, measures of dRPE were shown to have unique relationships, with gestalt sRPE providing similar information to sRPE-T and measures of sRPE and SRPE-L also showing distinct relationships. Further research may wish to consider whether measures of dRPE within similar populations are training mode dependent, providing further information to support practitioners aims of enhancing talent ID pathways. Therefore, the practical applications of this study are twofold: 1) practitioners should be aware that during SSG match-play pre-PHV players perceive higher levels of exertion than post-PHV maturation players, additionally mixed bio-banded games result in higher reported dRPE values overall, therefore player maturation status must be considered by practitioners during SSG team selection and be influenced by the desired outcome of the intervention. These considerations should create more optimum training environments, thus supporting player development. 2) Whilst collection of dRPE measures may provide useful information with regards to the subjective experience of players, practitioners should be aware that during SSG gestalt sRPE provides similar information to sRPE-T, and similarly sRPE-B and sRPE-L provide similar information. Depending on the aims of the training intervention, practitioners may wish to only consider one, or at most two, measures of subjective load.

Acknowledgments

The authors would like to thank all players and coaching staff at the soccer club for their cooperation and assistance during this study.