The majority of the recent literature in load monitoring focuses on the association between training load measures and injury risk. However, injury risk is not the only outcome measure we are interested in… performance matters too!
The understanding of the training-performance relationship in team sports is currently not as widely explored as training load and injury risk associations. Furthermore, the attention on injury risk reduction in training load management could lead to interventions that may be to the detriment of performance. This post aims to explore these two outcome measures.
Measuring Performance
Injury data may be more readily available for such analysis, given the (mostly) objective nature of the data as well as the legal and ethical requirements to collect such information. Meanwhile, defining “performance” in the team sport setting can pose more complexity. Game outcome i.e. wins v losses could be used but that doesn’t necessarily reflect performance.
Key Performance Indicators could be established objectively, although this process may require a great deal of data given the wide variety of variables that may influence team performances. Performance scores may be available from a sport-specific statistics provider, such as Champion Data in the Australian Football League (Mooney et al., 2013). Some studies have used match running performance as a representation of performance (Castagna et al., 2009; Castagna et al., 2010) in soccer.
Subjective measures of performance may be collected via coach ratings although this highly confidential data may not be shared internally, let alone be incorporated into published research.
There are, however, a number of examples of this approach in Australian Football. There include Cormack et al. (2013) and Hunkin et al. (2014), who investigated coaches’ rating of performance with neuromuscular fatigue (measured via FT:CT) and pre-match Creatine Kinase respectively. Mooney and colleagues (2013) used coaches’ votes and number of ball disposals to represent both subjective and objective markers of on-field performance in the same sport.
Sports Performance and Training Load
It seems a number of training load regulations have become commonplace in applied practice based on the recent research. These may include attempting to limit week-to-week changes in training load less than 15% (Gabbett, 2016) or the acute:chronic workload ratio (ACWR) below 1.5 (Hulin et al., 2016). Such measures are based on evidence relating them to injury risk but we should also consider their potential impact on performance. After all, physical preparation involves walking a tight rope between performance and injury risk.
Rob Aughey and colleagues (2015) set out to investigate the association between recent training load and strain on match outcome (wins vs losses) in Australian Football. They found likely effects between match wins and both higher weekly-loads and positive training stress balance calculated using strain. Do athletes perform at their best when we push them to tolerate high training loads, despite a potentially greater risk of injury? Does this suggest there may be a trade-off between injury risk and performance?
On the other hand, an innovative study published in Frontiers in November found periods of high acute load and sustained increases in load impaired match performance (Lazarus et al., 2017). The authors created a global weekly-load measure, calculated from a weighted combination of GPS and wellness data, and used quadratic mixed modelling to investigate its associations with performance (using the AFL’s Champion Data). They showed:
Performance was generally highest near or 1SD below the mean for each training measure
Small decrements in performance when 1SD above the mean in a selection of training measures
To my knowledge, this may also be the first time the acute:chronic workload ratio has been examined in relation to performance, rather than injury risk alone. Whilst the effects reported were mostly trivial, performance was generally higher near or below the mean, which across all positions was a ACWR of 1.0. This suggests the so-called “sweet spot” for maximising performance may well be similar to injury risk (Gabbett, 2016). However, I urge caution in extrapolating these results to all sports, teams, situations etc. As Lazarus and colleagues state:
“It is likely that different sports, teams, and load monitoring systems will have different training-performance relationships; therefore, these recommendations should be taken with caution and practitioners are urged to use a framework like that presented here to determine the ideal load in their own cohort of athletes.”
Training Protects from Injury
We have previously discussed some of the history of the proposed fitness-fatigue relationship and how this concept has been developed into Training Stress Balance and, more recently, the ACWR. That guest post on the Plinths and Platforms site is available via this page.
In summary, repeated stimuli applied over time are required to elicit adaptation and ultimately improve fitness and in theory (physical) performance. Whilst this stimulus will put an athlete at risk of injury purely via exposure, it is required in order to improve performance. Naturally then, each and every session is exposing an athlete to some interaction with both injury risk and performance.
However, with a focus purely on injury risk we are at danger of overlooking the outcome measure of performance itself. Sitting on the sofa all day might help you avoid injury, but it won’t help you win any medals (except perhaps in eSports…!) Wrapping someone in cotton wool will not give them the required positive stimulus they need to increase their threshold and therefore improve performance. If we consider Banister’s original model of human performance, focusing on minimising the fatigue effect but not considering the development of the longer-term fitness effect will, in theory, not result in improved performance.
Recent research suggests that chronic load, achieved in a progressive manner, helps to protect from injury. As Tim Gabbett outlines in his training-injury prevention paradox (2016); for athletes to develop physical capacities that protect them from injury they must train hard.
Conversely, undertraining may increase injury risk. Shane Malone and colleagues (2017) have demonstrated that high chronic training loads and regular exposure to very high speed running (>95% individual max speed) help to protect from injury risk. More recently, the same research group demonstrated that well developed strength and speed qualities help to reduce the risk of injury and allow athletes to tolerate higher workloads (Malone et al., 2018). Furthermore, Windt et al. (2017) found that maximizing preseason participation in elite rugby league was associated with a reduced risk of injury in the following week, as well as less games missed in-season due to injury.
What are the Risks?
There can be pressure on Sports Science staff to use the data to intervene or influence a programme. Rightly so seeing as though we do not want to collect data just for the sake of it! Often the easiest way to do this is to recommend recovery or modify someone’s programme by taking load away. It can be seen as less risky than suggesting adding to an athlete’s programme.
According to the research, what are the levels of risks that are being considered? To discuss this, let’s take a closer look at the “sweet spot” ACWR graph (Gabbett, 2016).
Gabbett (2016) http://bjsm.bmj.com/content/50/5/273
There is a polynomial pattern that suggests the likelihood of injury sharply increases as the ACWR surpasses 1.5. However, there are individual data points in the ‘danger zone’ that have similar injury likelihood (~5%) as other data points found in the ‘sweet spot’. Furthermore, when you consider the highest data points at an ACWR of 2.0, their likelihood of subsequent injury is between 20 and 25%. This means that there is still a 75-80% chance that they will not get injured.
Recent machine learning research helps to highlight this further. Carey and colleagues (2017) used data from three seasons at an AFL club and modelled training load data to try to predict injuries. Even with their best model that predicted hamstring injuries, there was a false positive rate above 10%. This means at least one in ten player-sessions would be unnecessarily modified using this approach.
While a higher risk of injury needs consideration, it is important that this does not lead to an oversight in considering the potential effect of interventions on performance. What impact might a missed session have on an individual’s physical, tactical, technical and mental preparation for example? What are the consequences of removing a stimulus on the individual’s “fitness” within the Banister model or in other terms on their chronic load? If we focus solely on injury risk, do we risk undertraining our athletes and/or not fully developing their physical capacities? Do we potentially protect them from peak performance, as well as injury?
Monitoring the ACWR is also relevant in rehabilitation as we have discussed here. This may be another environment where we regularly observe ACWR in excess of 1.5. For example, Ritchie et al (2017) observed ACWR close to and/or greater than 1.5 in the first and second week after RTP in AFL following lower- and upper- body injuries. However, they observed no recurrence of these injuries in the weeks after RTP.
Final Thoughts
The literature base has recently focused on injury risk as an outcome measure but it is our responsibility to also consider performance. There is clearly a conflict between accumulating training load and minimising fatigue. It is our role to try to quantify the risks involved and weigh up the cost: benefit of subsequent actions. This can be a challenge when trying to ensure players are at peak physical capacity to perform in games that may occur 1, 2, 3, 4, or even 5 days a week, depending on the sport.
I am not suggesting we take no actions as Sports Scientists, nor am I suggesting that we need to suddenly ramp up training loads and train with a positive stress balance every week! We need to ensure that the focus on load monitoring with injury risk does not result in an oversight of training load and performance relationships. After all, the need to avoid spikes in load at all costs is a myth.
Maybe there is a tug of war between injury risk and performance… then again maybe a well-designed and well-executed monitoring programme can assist with optimising both.
Comments