Hamstring injuries continue to be an issue in male football. Approximately 12% of all injuries concern the hamstring muscle. Taken into account that up to 35% of those will recur within the same season, it needs no further explanation that these injuries give rise to considerable time loss and will lead to decrements in athlete’s performance and overall fitness level, not to mention the repercussions on mental and psychosocial wellbeing.
Bearing these figures in mind, more and more time and effort is invested in injury prevention, mostly in the professional league, although this trend has definitely kicked in in the lower divisions at the amateur levels as well. Amongst others, the Nordic Hamstring Exercise is a very popular prevention item, due to the very intense eccentric loading and associated gains in eccentric load bearing capacity of the hamstring muscle-tendon unit.
The effectiveness of this exercise can be attributed to the fact that the hamstring unit mostly gets injured in football because it fails to withstand the high repetitive eccentric loads it needs to handle during repeated sprints. Indeed, lacking eccentric muscle strength, has been identified as a significant risk factor for football related hamstring injuries. Nevertheless, although this exercise has proven to be able to reduce the hamstring injury incidence, these eccentric strength training measures do not seem to be enough as recent epidemiological research has demonstrated that the incidence continues to increase as time passes.
This indicates that much more is at stake and more should be remediated rather than just isolated eccentric hamstring function. Because of the injury mechanism (which implies functional or structural hamstring muscle failure in the terminal front swing or initial stance phase of sprinting), me and my colleagues at the Department of Rehabilitation Sciences and Physiotherapy, Ghent University, were convinced that running technique is essential for promoting adequate running capacity and to allow the hamstrings to produce high levels of eccentric and concentric strength, in function of attaining maximal horizontal velocity without risking to get injured. As the hamstrings originate on the pelvis and have direct soft tissue connections with the sacrotuberous ligament, they should be highly dependent on the motion and postural control of the pelvis and lower spine, as movements in this proximal region will directly affect the tension on the hamstring muscle tendon continuum. In these terms, deficiencies in lumbopelvic control or ‘core stability’ might induce increased structural (more strain) and functional (need for additional stabilizing activity) loads on the hamstring unit. Such deficits would increase the risk of hamstring injury during high speed running in particular, because the biomechanical and metabolic responsibilities of the hamstrings are maximized under these circumstances.
To verify the possible existence of an association between features of lumbopelvic control (core stability) and hamstring injury susceptibility, we conducted a prospective cohort study throughout the 2013 – 2016 football seasons at the Ghent University (Department of Rehabilitation Sciences and Physiotherapy, Ghent, Belgium). Specifically, we intended to map the quality of core control (including trunk and pelvis) during both the stance and swing phases of the gait cycle in sprinting. To map these kinematical features, we used the Qualisys Track Manager system for three-dimensional (3D) motion analysis.
We recruited 60 healthy male soccer players playing in the upper division of the amateur soccer series in the region of Ghent (average age, body height and weight respectively: 24 y, 1.80 m and 75 kg). Throughout the month of July 2013, these subjects were invited to join our testing series which were conducted in Topsporthal Vlaanderen, Ghent, Belgium. This testing protocol consisted of repeated maximal over-ground sprints along a 40m running track.
During these sprinting trials, the participants were instructed to aim reaching maximal sprinting speed as fast as possible, as if they would be training for improving starting speed. Because this maximal acceleration phase is associated with substantial biomechanical loading and it necessitates highly adequate neuromuscular coordination, we wanted to objectify running kinematics during this specific phase of the sprinting trial.
To do so, we used the Qualisys Track Manager system, which consisted of 40 passive infrared reflective markers (12mm lightweight markers, Qualisys AB, Sweden) and 8 Oqus Infrared Cameras (Qualisys AB, Göteborg, Sweden) which were positioned around the running track to specifically map the athlete’s running kinematics in between meter 15 and 25 (Figures 1). We selected this part of the 40m running track, because this was the distance at which maximal acceleration for full speed sprint was reached (Figures 1, 2). The markers were attached in accordance with the LJMU Lower limb and Trunk Model for motion analysis (VanRenterghem J., Liverpool John Moores University), representing respective bony landmarks and segment clusters (Figures 2(a)).
Figure 1. Camera setup for 3D motion capture during maximal running acceleration for full speed overground sprint. The X’s indicate the positions of the 8 Oqus Camera’s surrounding the tracking volume.
Figure 2. (a) Attachment sites of anatomic and tracking markers in one of the participants (athlete is performing isolated knee flexion-extensions to dynamically detect the axis of rotation in the knee joint); (b) indicates the measuring volume size and demonstrates the position of the Optogait Bars which were used to differentiate stance and swing phases.
This way, 3D kinematics of the trunk and the lower limb could be investigated thoroughly. All reflective markers were attached to the skin firmly by using double sided carpet tape, to prevent them from coming loose or falling off. The entire 3D data assembly of the acceleration phase in sprinting was carried out by the Qualisys Track Manager hard- and software systems. A 10m Optogait system (Microgate, Bolzano-Bozen, Italy) was used for step detection, as we wanted to perform kinematic analysis of one entire stride this would have been impossible using a force plate. First, one static trial and 4 functional joint trials (left and right knee and hip joints (figure 2(a)) were recorded to create a virtual model. Subsequently, 8 markers were removed (left and right acromion, greater Trochanters, left and right medial epicondyles of the knee joint and left and right medial malleolus at ankle level) as these were not necessary for motion capture (Figure 1-2(a)). Hereafter, the participant was instructed to perform a 5 minute warm up by running up and down the running track at a self-chosen pace, alternated with short sprint-intervals, as if they would perform starting speed drills. After warm up, each participant was instructed to sprint up the track over a distance of (at least) 30m, trying to reach maximal sprinting speed as soon as one could. 3D data collection and step-detection were conducted simultaneously throughout the 10m measuring volume (meter 15 – meter 25), as both software systems were synchronized within the QTM interface (Qualisys Track Manager). Every soccer player had to perform 12 maximal sprints, because a minimum of six left – and six right side strides was needed for post hoc data processing and analysis within the Visual 3D interface (Visual 3D v5 Professional, C-Motion Research Biomechanics, Germantown, USA). A sprinting trial could only be taken along for data analysis when it consisted of three full stance phases within the measuring volume.
After testing, the participants were submitted to a 1.5 season prospective monitoring period for (hamstring) injury registry.
Because we wanted to verify to what extent running kinematics and in particular, lumbopelvic motion and control, were associated with hamstring injuries we performed both retro- and prospective statistical analyses. First, we looked whether or not kinematical patterns differed based on the presence of an injury history (retrospective analysis), afterwards we checked for kinematical deviations in association with injury occurrence during the 1.5 season follow-up period (prospective analysis).
All motion trials (captured with QTM), were converted to C3D files after which kinematical data were analyzed in Visual 3D (V3D, C-Motion Inc., USA). For each phase (front-swing, stance and backswing), kinematic data were normalized to 101 data frames, enabling in-between-subject comparison. This time-normalization procedure was necessary because the duration of stance and swing phases differed substantially in between individuals and trials. (maximal over-ground sprinting, speed was not standardized)
For both retro- and prospective statistical analysis purposes, we used Statistical Parametric Mapping (SPM). This method retrieves results on statistical inference, based on the Random Field Theory. Whilst commonly used for the analysis of functional brain images it is becoming commonplace in biomechanical data analysis as well.2-3
Using the mean joint- or segment angle and its corresponding standard deviation at each of the 101 data frames within one phase, it allows statistical comparison of the entire kinematic profile instead of the commonly used peak – or mean amplitude comparison which only takes into account one or a couple of point estimates, irrespective of time. For the SPM analysis within Matlab, α was set at 0.05. All statistical analyses within the scope of this study were implemented using the open-source SPM1D code (http://spm1d.org).
30 of the 60 soccer players within our cohort, had a recent history of hamstring injury (injury occurred less than 2 seasons ago). Comparative analysis of lower limb and trunk kinematics in between those two groups, did not reveal any significant differences. On the contrary, running kinematics appeared to be quite similar in both groups.
For the prospective statistical analysis, exclusively the kinematic data of the control group were taken along for analysis, because we wanted to exclude the potentially biasing effect of injury history.
In this control group, 4 subjects sustained a first hamstring injury during follow up. SPM and logistic regression analyses revealed that kinematics of the trunk and pelvis differed significantly in the injured subjects compared to the control cohort. The injured subjects presented more pronounced sagittal plane and frontal plane kinematics in the pelvis girdle during the swing phase of sprinting, this deviating motion pattern was characterized by higher amounts of anterior pelvic tilting (Figure 3, Table 1), thoracic side bending and pelvic lifting (Figure 4, Table 1) during the airborne phase. Although also taken along for analysis, no association was found between any of the kinematical features of the lower limb (hip, knee and ankle) and hamstring injury.
Figure 3. Sagittal Plane core and pelvis kinematics in terminal swing phase of running in (a) a healthy control subject and (b) a player sustaining a hamstring injury during prospective monitoring.
Figure 4. Frontal Plane core and pelvis kinematics in terminal swing phase of running in (c) a healthy control subject and (d) a player sustaining a hamstring injury during prospective monitoring.
Our findings indicate that running coordination might actually be highly associated with the risk of sustaining a hamstring injury. Lacking control of the lumbo-pelvic unit (insufficient ‘core stability’), presented by excessive pelvis and trunk (range of) motion during the swing phases of running presented to be related to the primary hamstring injury risk. Therefore, assessing and addressing sprinting kinematics / running technique seem to be indispensable in primary hamstring injury prevention. This study confirms the essential role of the core function in sports performance and injury prevention.
Albeit preliminary, the results of this study are of magnificent value for practitioners occupied with performance, prevention and rehabilitation in sports. We gratefully acknowledge the Qualisys Team, because this study could not have been conducted without the use of their high quality hard- and software systems.
Department of Rehabilitation Sciences and Physiotherapy
Faculty of Medicine and Health Sciences