« NHE training improves sprint performance »: a giant statement with feet of clay?

Written with Johan Lahti (MSc, PhD candidate, Université Côte d’Azur, Nice, France), including a point-counterpoint discussion with Lasse Ishøi (MSc, PhD candidate, Sports Orthopedic Research Center – Copenhagen (SORC-C), Copenhagen, Denmark).

IMPORTANT: Lasse is part of a point-counterpoint discussion, but has not otherwise been involved in the writing of this blogpost, nor does he share opinions with the authors.

At a recent international congress with major audience in the field of sports/football medicine and physiotherapy (see #FFMED on Twitter), a presentation and the comments made about some slides have drawn our attention. As seen in some statements made below, it basically appears that not only Nordic Hamstring Exercise interventions (NHE) protect against Hamstring Strain Injury (HSI), but it also improves sprint performance, by reportedly “3 to 5%”, which is assumed to allow players to “win the ball” and since football is such an easy game, eventually “maybe win the game”. All those who have played football know that this is a very far-fetched statement, but overall, we found this statement is actually a misinterpretation of the currently available knowledge on the topic.

Before all sprint and S&C coaches trying to improve players sprint speed resign, a couple of facts should be discussed…

Research shows that some isolated knee flexor strengthening exercises or programs have positive effects in terms of hamstring injury prevention (1, 2). Although we agree that isolated knee flexor strengthening exercises (using various modalities such as isokinetic, isoinertial or the NHE) are components of an effective preventive program, we think that the level of evidence supporting positive effects on sprint acceleration performance is comparably very low. At least not currently high enough to draw such conclusions.

In this post, we will discuss some partly published data and have a point-counterpoint discussion about the main randomized controlled trial (RCT) (3) testing the effect of an NHE training program on sprint performance in football players. We really appreciate the contribution of Lasse Ishøi in sharing their experimental data and discussing our points.

First things first, is hamstring strength, as assessed in isolated, single-joint tests (NHE or isokinetic) related to sprint performance?

Before discussing the potential training effects of NHE on sprint performance, it seems logical to study the cross-sectional relationship between these two variables: hamstring strength (as assessed during isolated, single-joint testing in NHE or isokinetic modes) and sprint acceleration performance (split times or running velocity over short distances, from 10 to 30m). Note that for the sake of clarity, we will use the term “hamstring strength” in this post for the magnitude of force output of the hamstrings in their knee flexors function as assessed during isolated, predominantly single-joint, “eccentric” tests such as NHE exercise or isokinetic testing.

To our knowledge, no study has shown such a direct, simple correlation between these two factors (hamstring strength and sprint acceleration performance). However, as we’ll see in the second part of this post, two studies quantified these variables in a pre-post training context but did not report correlations at pre- (3, 4).

One study by Mendiguchia et al. (5) included hamstring strength exercises (NHE, lunges, hip thrust, deadlifts, single-leg deadlifts etc), within a comprehensive and sprint-oriented training program for football players. In this study, hamstring strength was tested in an eccentric isokinetic mode (bilateral), as well as sprint performance (5- and 20-m peak velocity) but the pre-training correlation between these variables was not reported in the paper.

Here, we present that correlation, based on the raw data from this study:

Interestingly, the results from this study show that this multiple-exercise, hamstring-oriented strengthening programme including NHE but not only did not result in any change in the sprint performance of the players (pre-post % change reported for the group: +0.5% (90% Confidence Limits -2.8;3.8) on Pmax, and -0.6% (-1.6;0.5) for peak 20-m velocity. So the results of this comprehensive program contrast with the improvements reported by Ishøi et al. after a NHE-only program (see below).

Preliminary data from a novel study from our group (football and rugby players, presented by PhD candidate Johan Lahti at the World Congress of Biomechanics…LINK to the Abstract) shows a poor to moderate correlation between NHE torque capability (expressed relative to body mass, which is highly recommended since body mass influences NHE force measurements (6, 7) and 5-, 10- and 30-m time. Note that the NHE torque in our study was measured thanks to a recently developed device named “Hamtech” that can be seen in this slideshow (LINK to the Slideshow) and that will be presented in a coming post on this blog.

Finally, the same kind of correlation was obtained from the data of Ishøi et al.’s study (3) (pooled data of hamstring strength versus sprint performance for players from both groups. Note that we show here the strength data in absolute units (upper panels) but also normalised to body mass (lower panels):

From the experimental data shown above, the consistent absence of a clear positive relationship between hamstring strength as assessed during NHE (or isokinetic) testing and sprint performance makes total sense, given the significant differences between the testing modality and the way hamstrings actually operate during sprinting:

• Single joint testing versus multi-joint and multi-segmental system actions during running
• Overall geometrical conditions: knee, hip and trunk angles, associated muscle lengths
• Opposite ends on the force-velocity spectrum: low-to-very-low velocity versus high-to-very-high velocity
• Coordination with surrounding agonist and antagonist muscles (glutes, thigh, trunk, pelvis, lower leg muscles)
• Closed chain (NHE) versus combined Closed/Open-chain demands (sprint)
• Non-sprint-specific versus sprint-specific actions (easy one 😀)

Interestingly, in our 2015 paper published in Frontiers in Physiology (8), we discuss the “major role” played by hamstring in horizontal ground force production during sprint acceleration. At first sight (or if you only read papers titles or associated tweets), it seems to contradict the results presented above. However, in our 2015 treadmill study, we also found that isokinetic knee flexor torque was notrelated to sprint horizontal force output (and in turn acceleration performance, would the testing have occurred on a track). In fact, only when adding the variables of isokinetic torque and EMG activity measured specifically during the end-of-swing phase of sprinting did the regression model explain a significant part of the horizontal ground reaction force. To summarize these findings, isolated, low-velocity hamstring strength did not correlate with sprint horizontal force output, but the combination of this strength capability and the sprint-specific EMG activity did. Of course, numerous limitations apply to this study including the use of a treadmill (no other possibility to synchronise all measurements throughout the sprint), the use of surface bipolar EMG, and the use of an experimental research approach (sorry for that). Bottom line here is that isolated hamstring peak torque during isokinetic testing was not correlated with sprint force, power or performance outputs.

That being said, two studies from the same research group were published in 2017, investigating the training effects of NHE on sprint performance in football players: a first pilot study by Krommes et al. (4, access the paper HERE), and a randomised controlled trial by Ishøi et al. (3, access the paper HERE).

In the second part of this post, we will discuss some reasons why we think that statements such as “Nordics make you about 4% faster” are way too premature, in light of the available data from these two (very similar) studies. Since Lasse Ishøi was very cooperative in sharing the data from their randomized control trial and brought some points to answer our comments, we are happy to have him join the discussion in this second part. For the sake of clarity, his points will be written in blue and appear with a “Counterpoint” headline, wherever it made sense to him to intervene.

Do “Nordics make you about 4% faster”? A point-counterpoint discussion with Lasse Ishøi

The two studies share several features such as the objectives and the experimental protocol, but due to a low final number of players eventually compared in Krommes et al.’s study (9 in the NHE group versus 5 controls), we will essentially discuss the results of Ishøi et al. However, we should keep in mind that the results of Krommes et al. show contrasting changes in sprint performance after the NHE training: on average, the 9 players were 2.42±0.2 % slower on the 30-m but 9.4±0.15 % faster on the 5-m and 5.77±0.15 % faster on the 10-m…Statistics were not more detailed due to low sample size. However, the individual data show that the average 5.77 % faster 10-m was associated with 6 players out of 9 getting faster. This raises the point (see below) of group-average versus individual changes interpretations.

Counterpoint:
I thank JB and Johan for this opportunity.
Please note the control group in Krommes et al. was 3.88% slower on 30m (vs. NHE group: 2.42% slower).

The important decrease between subjects recruited and eventually tested is attributable to drop-outs and technical issues. The authors mention that “as the overall compliance rate to NHE training was 60% with only few players above 70% compliance, we did not perform per-protocol analyses as planned”.

The players in the NHE group who could do all the training sessions “ingested” a weekly dose of about 90 reps on average during the weeks 5-10 of the program, at a cruising altitude of 3 sessions or 3 sets of 10 reps on average. In total, 10 weeks (27 sessions) of NHE training.

Good news is that this resulted in a 17-19 % improvement in NHE strength (depending on the metrics considered), on average for the group. It is important here to note that NHE group had clearly lower peak eccentric hamstring strength than control at pre-training (322 vs 382 N), and eventually reached at post-training (383 N), the same average value as the controls pre-training value (382 N). This led the authors to analyse data using an analysis of covariance to take this different basis level of strength into account.

Other good news is that “total sprint time”, i.e. the sum of 24 sprints of 10-m performed in 4 sets of 6 sprints (15s and 3min of rest) decreased by 1.84% on average (so 0.03 s per sprint), versus 0.33% for the control group. Note that the standard deviation around this group change was large (about as large as the average value) which shows a high variability in the responses, as seen in the individual data plotted in Figure 2 in Ishøi et al.’s paper. When considering individual changes, 8 players out of 11 improved their total sprint time above the smallest worthwhile change (SWC) that we considered here to be 0.2 times the pooled (NHE + control group) pre-training standard deviation for this variable (9) (so here -0.78%). The SWC line is drawn in red in the figures below. This amount of “positive responders” was 9/11 players for the fastest 10-m time and 8/11 players for the last 10-m time outcome. In the control group, 6/14 players improved their total sprint time, 2/14 their fastest 10m and 8/14 their last sprint time. Note that the results of this SWC analysis must be interpreted carefully, due to the fact that the SWC (threshold of practically meaningful individual change) was computed on the initially available players but then applied to the pre-post training changes of a different size sample, i.e. the players who completed the entire protocol without dropping out.

Counterpoint:
I agree the SWC is relevant when assessing the effect of an intervention. However, I disagree that the SWC is applied on a different sample size. The SWC is calculated based on the pre-intervention pooled SD of the 11 vs. 14 players who all completed the sprint testing at pre- and post-intervention. Thus, the SWC is applied on the exact same cohort of players. Based on this, SWC for total sprint time, fastest sprint, and last sprint is 0.78%, 0.85%, and 0.94%, respectively, with 72.7%, 81.8%, and 72.7% of players in the NHE group improving above the SWC. For a team of 25 players that would indicate a relevant improvement in 10m sprint for 18-20 players – I consider that a positive “adverse” effect for a protocol designed to reduce the risk of hamstring injuries.

It is interesting to note that a very recent study that will be presented at the European Congress of Sport Science this summer by Duhig et al. (abstract presented below is publicly available here) shows that sprint performance (sessions of 10x80m) “did not differ” within an eccentrically trained group (5-wk NHE training) despite increased knee flexor strength. Much more details will be provided at the ECSS Congress presentation but this study adds some points to the ongoing discussions.

 

Given (i) the contrast between the major improvement in NHE strength and the minor and variable improvement in sprint performance, (ii) the lack of correlation between NHE strength and sprint performance shown in the first part of this post, and (iii) the obvious question: is the improvement in NHE associated with an improvement in sprint time, on an individual basis, we performed additional analyses to discuss the correlations between training-induced individual changes in NHE strength and sprint performance.

Our point is that the study shows that NHE on the one hand, and sprint performance on the other hand overall improve (by a much larger and clearer magnitude for NHE strength, with a less systematic trend for sprint time as shown by the SWC data). But it does not answer the question “are player’s improvement in NHE correlated with an improvement in sprint performance”?

Counterpoint:

Yes, we observed large improvements in NHE strength and small-to-medium improvements in sprint performance. However, a one-to-one relationship is not expected, and therefore I see no contrast in these findings. Furthermore, it could likely be that NHE strength is not the primary variable driving adaptations in sprint performance – this is discussed in below.

Regarding the lack of correlation between baseline NHE strength and sprint performance, interestingly, the correlation analysis from Mendiguchia et al. (presented above) actually show a moderate positive correlation (r=0.31) between peak velocity at 10m and relative eccentric hamstring strength measured using gold standard isokinetic dynamometry. As NHE strength is “NHE strength” and not gold standard eccentric hamstring strength it is not surprising that no correlation exists between NHE strength and sprint performance.

Furthermore, there seems to be no correlation between eccentric hamstring strength and peak velocity at 30m in Mendiguchia et al.’s study. Collectively, this is interesting as Mendiguchia et al. reported a moderate between-group difference in short distance sprint performance (5m), but unclear difference for 20m following targeted eccentric hamstring training.

Therefore, these results are not in contrast to our RCT and pilot RCT (as stated by JB and Johan), showing improvements in short distance sprint (5m and 10m) but not at 30m sprint. Therefore, it is not surprising that Duhig et al. did not find improvements in long distance sprint (80m) following eccentric hamstring training, and this does not, in my opinion, question the effect of NHE intervention for short distance sprint.

The data below show the correlations between individual changes in NHE strength and sprint performance, using data from Ishøi et al.’s study (data for both pre- and post- measurements were available for only 10 players in the NHE training group).

The SWC line appears in red for the sprint performance (0.2 times the standard deviation of the pooled pre-training data, expressed in %):

This additional, more individual-responses centred analysis, brings two interesting points:

First, there was a high variability in the sprint performance responses, although a majority of players improved performance above the SWC. Contrastingly, all players improved their NHE strength. Of course, other physiological variables than NHE strength may be developed through a NHE training, and contribute to sprint performance, but only NHE strength has been measured in this study.

The second information is that the poor correlations between NHE strength and sprint performance improvements may be interpreted as the fact that “on an individual basis, improvements in NHE strength are not associated with improvements in sprint performance”. Some players improved NHE strength a little and sprint performance a lot, some others improved both a lot, and some others improved NHE a lot but not sprint performance at all.

Added to the low sample size, we think that (over)-simplified statements such as those listed at the beginning of this post are inaccurate and misleading, and not fully supported by the experimental data presented.

Counterpoint:

Although it is premature to conclude on secondary correlation analyses with major lack of statistical power, the lack of correlation between changes in NHE strength and sprint performance is indeed interesting. However, please note that these correlation analyses solely raise the question whether changes in NHE strength drive improved sprint performance. This does not seem to be the case, which is not surprising since (i) NHE strength is not the same as gold standard eccentric hamstring strength and (ii) a lot of muscular and neural adaptations occur following high-load strength training. Our group has recently found hamstring rate of force development to be much more associated with short sprint performance than peak hamstring strength.

Thus, the correlation analyses provided by JB and Johan, is not related to the effect of the NHE intervention on sprint performance – on a group or individual level. On a group level, two NHE RCTs and one hamstring RCT (Mendiguchia et al.) have indicated beneficial improvements on short distance sprint, whereas individual changes from our RCT indicate that >70% improve above the SWC.

It may be true that “on an individual basis, improvements in NHE strength are not associated with improvements in sprint performance” and “that the level scientific evidence on the relationship between NHE strength and sprint performance is low (quote, conclusion)”, but none of these statements answer the question “does the NHE intervention improve short distance sprint performance?”

Our level 1 RCT (including pilot RCT) answer that question (at least for amateur football players) – on a group and individual level. This blogpost highlights some interesting points, but these suggest that NHE strength may not be the primary driver for improved sprint performance, NOT if the NHE intervention improves sprint performance or not – these are two very different yet important aspects.

I look forward to future studies regarding the effect of hamstring training (and NHE) on sprint performance and the mechanisms driving this.

One last update is that we performed the same analyses on the data of Mendiguchia et al.’s study, with pre-post individual changes in knee flexor eccentric strength (average of both legs, isokinetic testing) compared to changes in sprint performance (maximal sprinting velocity at 10 and 30-m). The SWC threshold (0.85 and 0.87% computed showed 12/26 positive responders to the multifactorial, hamstring- and sprint-oriented training program, in terms of peak velocity at 10-m, and 10/26 positive responders for the 30-m maximal velocity variable. As for the previous cohort, the correlations between the individual changes in sprint performance and knee-flexor eccentric peak strength were very low:

Conclusions

Our team is not “pro-NHE” or an “anti-NHE”. Surely some of my colleagues in academics or training fields are, given their public comments and opinions on the topic (Twitter, workshops, conferences), but I trust scientific evidence, and so far my opinion is that the level scientific evidence on the relationship between NHE strength and sprint performance is low. Just as low, for instance, as the role of “trunk and pelvis stabilization” in hamstring injury risk discussed in recent works by J Schuermans et al. (10-12). So definitely, “more research is needed” should be a fair comment here…

It is well-known that “love makes you blind” but if we want our field of research (sport science and medicine) to keep-gain credit in the eyes of the sports performance field, we should adopt a balanced approach in our statements, and not seek for misleading, Twitter-buzz oriented statements. Considering the small sample size we usually deal with especially within sports science research, when concluding, vaguer language should be considered as more appropriate and even the norm. This This is even more crucial as a large part of our audience actually does not read the studies details and stick to the titles or our Twitter comments…

REFERENCES

1. van der Horst N et al. The Preventive Effect of the Nordic Hamstring Exercise on Hamstring Injuries in Amateur Soccer Players. Am. J. Sports Med. 43,1316–1323 (2015).
2. Petersen J et al. Preventive effect of eccentric training on acute hamstring injuries in men’s soccer: a cluster-randomized controlled trial. Am. J. Sports Med. 39,2296–303 (2011).
3. Ishøi L et al. Effects of the Nordic Hamstring exercise on sprint capacity in male football players: a randomized controlled trial. J. Sports Sci. 36,1663–1672 (2018).
4. Krommes K et al. Sprint and jump performance in elite male soccer players following a 10-week Nordic Hamstring exercise Protocol: a randomised pilot study. BMC Res. Notes 10,669 (2017).
5. MendiguchiaJ et al. Effects of hamstring-emphasized neuromuscular training on strength and sprinting mechanics in football players. Scand. J. Med. Sci. Sports 25,e621–e629 (2015).
6. Roe M et al. Eccentric knee flexor strength profiles of 341 elite male academy and senior Gaelic football players: Do body mass and previous hamstring injury impact performance? Phys. Ther. Sport 31,68–74 (2018).
7. Buchheit M et al. The Effect of Body Mass on Eccentric Knee-Flexor Strength Assessed With an Instrumented Nordic Hamstring Device (Nordbord) in Football Players. Int. J. Sports Physiol. Perform. 11,721–726 (2016).
8. Morin JB et al. Sprint Acceleration Mechanics: The Major Role of Hamstrings in Horizontal Force Production. Front. Physiol. 6,404 (2015).
9. Buchheit M. The Numbers Will Love You Back in Return-I Promise. Int. J. Sports Physiol. Perform. 11,551–4 (2016).
10. Schuermans J et al. Deviating running kinematics and hamstring injury susceptibility in male soccer players: Cause or consequence? Gait Posture 57,270–277 (2017).
11. Schuermans J et al. Proximal Neuromuscular Control Protects Against Hamstring Injuries in Male Soccer Players: A Prospective Study With Electromyography Time-Series Analysis During Maximal Sprinting. Am. J. Sports Med. 45,1315–1325 (2017).
12. Schuermans J et al. Prone Hip Extension Muscle Recruitment is Associated with Hamstring Injury Risk in Amateur Soccer. Int. J. Sports Med. 38,696–706 (2017).