Inter-limb asymmetries: What impact do they actually have on performance?

Chris Bishop has become the go-to guy when it comes to understanding athlete limb asymmetries. Despite the plethera of research that Chris and others have undertaken in this area, there is still a lot of confusion in the field. So we asked Chris six questions which will hopefully help coaches understand this area a little clearer.

You have done a lot of research into asymmetries over a number of years. What do we know about asymmetries and how they actually impact performance?

I’ll answer this in two parts: what do we know, and how do asymmetries impact performance. It seems that this is definitely a topic of interest to the general sport science / strength & conditioning community, given how much research has been done in the last 30 years, in both performance and injury-based settings.

First, inter-limb asymmetry (i.e., between limbs, not within the same limb) is a ratio, meaning it is made up of two component parts (e.g., right vs. left; dominant vs. non-dominant; injured vs. non-injured, etc.), and that means the data is always “noisy.” You can check this for yourself by looking at the mean and standard deviation (SD) reported in almost any study, which often shows the SD to be > 50% of the mean asymmetry value for the group and sometimes as much as 100% of the mean. It doesn’t seem to matter what the population is –  their skill level, in what sport they compete, or what time of year they are tested – the data always shows a lot of within-group variation in asymmetry scores (large SD relative to the mean).

Interestingly, when this is considered in any kind of test-retest design, it’s rare that asymmetry exhibits “significant change” between test sessions or time points. As such, this gives the impression that asymmetry is quite consistent. However, this is false for two reasons: 1) when the SD is so large, it precludes us from finding any “statistically significant” change – but this is hard to believe when we know the variation in scores is so large from person to person; and 2) when we view changes on an individual basis, we see that differences are often very large.

So, it’s important to remember that asymmetry is “noisy” and determining true change at the individual level (discussed more later) is key.

Second, when we consider how asymmetry impacts athletic performance, I have come to believe that it is probably not all that important.

Many correlational studies looked at the association between asymmetry in one task (e.g., jumping) and performance in another (e.g., linear or change of direction speed tests). If “significant” correlations do exist, it’s rare for them to be anything greater than moderate in magnitude, and almost all studies have been conducted at a single time point. This, of course, raises the question: would the results be the same if tested again a month later? From the limited longitidunal data, I’m fairly sure the answer is no.

In addition, a recent meta-analysis from Francesco Bettariga looked at the effects of training interventions on inter-limb asymmetry. Results showed that training interventions only elicited small reductions in asymmetry when compared to a control group, with even fewer studies then investigating how the subsequent changes in asymmetry impacted performance outcomes. Of note, one study by Sannicandro et al. showed that a combination of strength and balance exercises was very effective at reducing jumping asymmetry in youth tennis athletes. However, there was no subsequent improvement in their linear or change of direction speed times. Thus, the evidence trying to link asymmetry and athletic performance is tenuous at the moment, at best.

For coaches out there that decide asymmetry is worth measuring / monitoring, what are the best practices when it comes to calculating them, as it’s not as simple as it first seems?

Let me start by saying that I don’t think this is so much a case of completely “right and wrong” and all our work on the calculation aspects of asymmetry is largely opinion-based. However, I also think we have a pretty logical reasoning behind the calculation choices we have made.

In our opinion, one of the key differentiators is whether you are testing / measuring asymmetry during a bilateral or unilateral test.

During bilateral testing, both limbs are interacting together to produce a sum output of any given metric (e.g., force or impulse), so any evaluation of inter-limb differences might be best done in relation to the sum output value.

However, during unilateral testing, there is no direct ground reaction force from the non-jumping limb. Thus, we may not need to interpret the existing limb difference in relation to the sum total (noting that limbs do not interact together, in this instance). Rather, and remembering that asymmetry is just a percent difference, this is perhaps best calculated in the same way that fractions are computed in mathematics (noting as well, we don’t learn multiple ways to do this in math class).

Moving onto our working examples, if we consider peak force (in Newtons – N) as our metric of interest and we measure this during both bilateral and unilateral countermovement jump (CMJ) tests, here are some hypothetical numbers:

• Bilateral CMJ: Right limb = 675 N; Left limb = 600 N
  • Calculation: (Larger – Weaker)/Total*100
  • Example: (675-600)/(675+600)*100 = 5.9%
• Unilateral CMJ: Right limb = 850 N; Left limb = 700 N
  • Calculation: (Larger – Weaker)/Larger*100
  • Example: (800-700)/800*100 = 12.5%

For a detailed breakdown of the different formulas or equations that have been used to calculate asymmetry and why we make the above suggestions, please read here.

When looking at asymmetry, using 10% as a cut-off to indicate that an intervention is needed seems to be somewhat common practice. Where did that value come from and is it even relevant?

This is an interesting one and it’s probably quite a complicated and somewhat context-specific answer. That is to say, I probably can’t tarnish every study with this answer, but there are some common themes I have picked up in my reading over the last 6-7 years, so here goes…

When looking at some of the asymmetry literature in sport from the early 1990’s, it seems that many studies suggested that 15% was a marker of increased injury risk. The issue here is that once this makes it out into the academic community, it only takes a handful of other studies to cite this information for some to start believing it to be true.

More recently, that value has transitioned to 10%, and a study by Kyritsis et al. is a classic example of this.

However, considering that studies cannot agree upon one figure and asymmetry is so noisy, who’s to say that we’d obtain the same asymmetry value if we tested again one or two weeks later. In addition, we’ve already outlined the weak link between asymmetry and athletic performance, so the only question to remain is whether a threshold exists for heightened injury risk in athletes?

I recently wrote an article on thresholds for asymmetry in NSCA’s Strength and Conditioning Journal and outlined some of the pitfalls with using thresholds for asymmetry.

First, as previously mentioned, the literature doesn’t seem to be able to agree on one specific threshold. Second, much of the asymmetry data in injury-based studies to date revolves around outcome measures (e.g., jump distance or height), as opposed to strategy metrics during jump tests. Previous research on jumping has shown that jump strategy metrics may be more sensitive to change than outcome measures, and, from an injury perspective, a recent study showed that asymmetry may mask persisting limb deficits during the rehabilitation process. Third, asymmetry is highly task-specific, with an abundance of evidence corroborating this point. Simply put, we should not expect an asymmetry score of 10% in a different test.

To take this further, asymmetry is also metric-specific within the same test. If I monitored asymmetry in jump height, contact time and reactive strength during a unilateral drop jump test, I am 99.99% sure you would not get the same values for each metric. Thus, and if we consider our original question, how can we apply or act upon one threshold or cut-off of any magnitude if asymmetry values differ values for each metric within the same test?

Naturally, then, this raises another question: how do we then interpret asymmetry data when we measure it? I will discuss this in Question 5.

Should asymmetry and asymmetry alone be a red flag to signal an intervention?

The short answer is, no. I’m not sure any single factor or variable in isolation is a risk factor to signal an intervention. The study by Kyritsis et al. is another good example. To summarise, six clinical discharge tests had to be met in order for professional soccer athletes not to be at a 4x risk of suffering a second ACL rupture: i) < 10% quadriceps strength asymmetry on the isokinetic dynamometer at 60°/s, < 10% asymmetry for the: ii) single leg, iii) triple and iv) crossover hop for distance tests, v) running an agility t-test in < 11-s, and vi) complete an on-field sports-specific rehabilitation.

The first assumption is that this is the correct test battery for soccer athletes. While nothing stands out as being wrong per se, these tests largely reported outcome measures only (remembering that this misses a big piece of the picture). In addition, none of these tests had any predictive capability on their own, again highlighting the somewhat limited information an outcome measure of distance or time might provide us as a single metric in a single test.

Finally, while I am at the risk of repeating myself here, we must remember the associated noise that asymmetry brings. If asymmetry data changes so much from session to session, especially at an individual level, even without any kind of intervention, it is challenging to know whether doing nothing would have been just as effective. Instead, and as experienced practitioners like Matt Jordan often allude to, monitoring asymmetry is just another piece of the performance puzzle. We should not hang our hat on this metric as the golden ticket to improve performance or judge whether athletes are ready to return to training and competition. I’d probably argue that the physical characteristic of “strength” is more important to monitor and make decisions off when it comes to assessing limb differences.

As Sean Maloney points out well in this article, a persistently weaker limb should be seen as a “window of opportunity” for increasing output or capacity. In turn, and assuming the stronger limb doesn’t get concurrently weaker, the inter-limb % difference may start to reduce over time.

How should we frame asymmetries in a RTP setting vs. a performance setting? How do asymmetries impact injured athletes?

I’ll start by saying that a more detailed read about our thoughts on this is in a special issue article we wrote for the Aspetar Sports Medicine journal. I also think this may not be as simple as delineating so clearly between injured and healthy athletes. Rather, a continuum of what we monitor and why may be an alternative way to look at this.

For athletes with a serious injury, an asymmetry is present for a very specific reason: one limb is currently incapacitated. Naturally, this provides the necessary context for why large inter-limb differences may be evident in a given test. In line with the suggestions by Sean Maloney, and of course what is no doubt deemed standard practice in rehabilitation, providing improved capacity for that injured limb is necessary so they can start training again and hopefully get back to competing.

With that in mind, monitoring how the magnitude of asymmetry changes during the rehabilitation journey seems like an obvious thing to do, noting that it provides context of one limb relative to the other. However, as with all ratio numbers, this should not be done solely by using the percent value, but in conjunction with the raw limb scores, as well.

Something our research has shown over the years is that the direction of asymmetry (i.e., which limb performs better) in healthy athletes can be subject to change, in some cases just as much as the magnitude of asymmetry. Thus, for those athletes who are close to returning to training (assuming that this would only be allowed once sufficient capacity deficits have been restored), I’d suggest monitoring the direction and magnitude of asymmetry. This seems less important to suggest during the early stages of rehabilitation, because the existing side-to-side differences are there for the one obvious reason. However, as capacity returns, it may be worth monitoring the direction of the imbalance to determine which of the following scenarios is occurring:

1. The direction of asymmetry still favours the uninjured limb (i.e., persistent capacity deficits still exist on the injured limb).
2. The direction of asymmetry is flip-flopping between sides. In this instance, and assuming raw scores are not drastically reduced compared to when the athlete was healthy, this may just be fluctuations in performance variability.
3. The direction of asymmetry actually now favours the injured limb (i.e., it is now performing better). Intuitively, this seems unlikely, although not impossible. But if this is the case, I would encourage practitioners to compare data to pre-injury scores, so that we can be confident that capacity in both limbs has not declined. Simply put, limb symmetry is probably of little use to our athletes if they are substantially weaker than before the injury occurred.

Finally, and not strictly part of the question per se, I also think it’s key to pick up on how we use these percent values in practice, because it may not always be clear what to do when you have that value.

A fantastic article by Tim Exell outlined that an asymmetry may only be considered “real” if the between-limb difference was greater than the intra-limb variability. In layman’s terms, this principle suggests that the percent difference probably needs to be greater than the coefficient of variation (CV) in order for it to be meaningful. Again, to provide two hypothetical examples here:

• Athlete 1: Right limb CV = 3.5%; Left limb CV = 2.9%; Asymmetry = 6.2%
• Athlete 2: Right limb CV = 6.7%; Left limb CV = 3.5%; Asymmetry = 5.5%

Athlete 1 shows an asymmetry value greater than both CV values (noting we can, of course, get this for both limbs, if we are recording separate limb data), but athlete 2 does not. This does not mean we have to run an intervention for athlete 1 right away. However, if athletes continue to exhibit real asymmetries during weekly test protocols and the same limb is persistently under-performing (let’s say for 4-6 weeks), practitioners may then wish to consider discussing why this athlete is showing those results. For example, is it simply a by-product of their sport over time, or is it a true capacity issue that perhaps warrants further investigation?

Given the context-specific nature of such discussions, practitioners should be able to answer this with their colleagues, when considering all associated factorsL lifestyle, training and injury history, test protocols, etc.

What are the biggest mistakes you see younger practitioners make and what advice would you give them to improve their practice/delivery, etc.?

I think the main one is to just do more coaching. For anything that isn’t an absolute science (e.g., coaching), it is likely to take a bit more time to get better and comfortable at that skill because the answers are not binary. To my surprise, we send out lots of opportunities to our MSc students for placements (~ 20 per year) and some never get filled, and it does make me wonder why more don’t take up these offers.